Long‐run impacts of trade shocks and export competitiveness: Evidence from the U.S. BSE event

This paper examines how comparative advantages of major beef exporters changed following the 2003 bovine spongiform encephalopathy (BSE) outbreak, which significantly disrupted the U.S. beef trade until approximately 2007. Using longitudinal data on beef export values and constructed revealed comparative advantage measures, we show that while some measures of the long‐run impacts of BSE on U.S. beef export competitiveness have returned to pre‐2003 levels, the U.S.’s comparative advantage has not. We also examine a hypothetical scenario of no BSE event in 2003 and predict that in the absence of the BSE outbreak, the U.S. beef sector would have been increasingly more competitive by 2017 than it actually was. Long‐term trade competitiveness may not simply return to normal even after a short‐term disruption.


Introduction
Shocks from trade disputes and phytosanitary emergencies occasionally impact agricultural export markets. Affected exporters always hope that such events are short-lived. A trade dispute beginning in 2018 between the United States and China is a pertinent example that has led to additional tariffs on U.S. agricultural products including corn, soybeans, cotton, and pork (Marchant and Wang 2018). Notably, the price spread between U.S. and Brazilian soybean exports widened to a record-high immediately after tariffs were imposed by China around the middle of 2018, with the U.S. soybean export price remaining low relative to the Brazilian price throughout the rest of the year (Good 2018). By assuming the tariffs on U.S. agricultural goods remain in effect for the next 10 years, the United States Department of Agriculture (USDA 2019) predicts that U.S. soybean exports would not return to pre-trade-war levels until 2028. However, even if the U.S. and China ended their disputes in 2019 or 2020 and the tariffs returned to pre-2018 levels, the competitive structure of markets may have adjusted in the interim. U.S. farmers are rightly concerned that adverse effects of even short-lived disruptions could permanently alter market relationships as other exporters erode the U.S.'s share in global markets (Balistreri et al. 2018;Elmer 2019;Hirtzer 2019). In particular, once China finds new trading partners, renegotiation costs can slow the U.S. in regaining market share just as it did for U.S. grain markets following the short-lived 1980 U.S. embargo of the former Soviet Union (Balistreri et al 2018).
In this paper, we shed light on how long it takes an export market to recover from a trade disruption. Although at this writing, the current U.S. and China trade disruptions are making headlines, it is difficult to forecast long-run outcomes for something that has limited data. We simply do not know the extent to which the disputes are permanently changing export 2 relationships. The outbreak of Bovine Spongiform Encephalopathy (BSE), also known as mad We create an empirical proxy to measure country-level beef industry comparative advantage over time: the indicator for competitiveness in our study. We show that while the U.S.
beef export values has mostly recovered back to their pre-BSE levels, the U.S.'s comparative advantage has yet to return to where it was prior to the BSE outbreak. We jointly estimate the effect of the BSE outbreak on the comparative advantages of other major exporters. We find that in the absence of the 2003 U.S. BSE event, the U.S. would have kept its comparative advantage in beef; moreover, its competitiveness would have grown over time. The results indicate significant lingering impacts of BSE on U.S. beef competitiveness that are less obvious when examining export values alone.
This study contributes to the literature in the following ways. First, we construct a modified revealed comparative advantage (RCA) index proposed by Yu et al. (2009) to present the trends of comparative advantage of the U.S. beef industry and other major competitors from year to year. The RCA index was first proposed by Balassa (1977) and reformulated in Balassa (1986), and is used frequently when looking for changes in a country's trade status (Gortan et al. 2000;Ferto and Hunnard 2003). However, the original RCA index features unsatisfactory 3 characteristics that are not appropriate to be used in statistical analyses (Vollrath 1991;Hoen and Oosterhaven 2006;Yu et al. 2009;Laursen 2015). The reformulated RCA index developed and used by Yu et al. (2009) resolves these limitations. To the best of our knowledge, we are not aware of any studies that have used an RCA index to study the impact of BSE on the U.S.'s beef sector competitiveness.
Different aspects of the adverse economic impacts of BSE outbreaks have been studied both in the U.S. and other countries. Research finds negative impacts on consumer demand for beef products in the aftermath of local BSE events (e.g., Burton and Young 1996;Mangen and Burrell 2001;Verbeke and Ward 2001;Peterson and Chen 2005). Other studies have examined the effects of food scares from BSE outbreaks on cattle futures prices and beef sales adjustments finding evidence of significant structural breaks of futures prices and adverse effects on beef sales following BSE events (Jin et al. 2008;Marsh et al. 2008;Schlenker and Villas-Boas 2009;Taha and Hahn 2014). We add to this literature by providing evidence of the impact of an outbreak of BSE on a country's international competitiveness. The aforementioned studies tend to focus on the short-run impacts of BSE, our results highlight the significance of long-run impacts on a country's trade performance. This paper proceeds as follows. In Section 2, we provide background information on the U.S. beef sector's competitiveness in the world market, as well as a brief history of the BSE outbreak in the United States. Section 3 describes the data construction process and presents summary statistics. Section 4 lays out the empirical analyses and discusses results. Finally, Section 5 concludes and discusses future research.

United States Beef Export Competitiveness
The U.S. beef industry operates in a highly competitive global marketplace. Major competitors include Canada, Australia, New Zealand, Brazil, and Mexico (USDA-FAS 2019). Historically, the United States has held a comparative advantage in beef production due to a well-developed infrastructure and a reputation for both meat quality and food safety. However, the United States can be at a disadvantage relative to cost of production. For example, a pound of grass-fed beef can typically be produced at lower cost, where the majority of U.S. beef is grain-fed.
Competitive advantages can also be built around the sophisticated use of information. Globally, animal identification and traceability are important components of managing animal and human health and food safety (Schroeder and Tonsor 2012). Traceability systems also enhance communication and coordination by delivering information up and down the supply chain to benefit producers, processors, and consumers. Smith et al. (2005) reported that the United States is "lagging behind many countries in developing traceability systems for food in general and especially for livestock, and their products" (p. 174). Of the world's eight largest exporters, six have in place mandatory cattle animal identification and traceability systems. Only the United States and India have not adopted mandatory national identification and traceability systems (Schroeder and Tonsor 2012). In short, the United States' beef industry today faces a highly competitive and developing global market place (Murphy et al. 2009;Schroeder and Tonsor 2012;Pendell et al. 2013). Trade relationships, exchange rates, and economic growth rates in other countries all affect the export demand profile.

U.S. BSE Outbreak
Bovine spongiform encephalopathy (BSE) is a neurological disorder of cattle that cannot yet be treated or vaccinated against. Cattle affected by BSE experience degeneration of the nervous 5 system. BSE can be categorized into two types -classical (C-type) and atypical (H-type or Ltype). Only the classical BSE is zoonotic, where humans can become infected through consumption of diseased beef products, but symptoms do not appear for some time, making diagnosis, and hence food recalls, more difficult. 1 The disease became officially recognized in the 1980s, and the first diagnosis of classical BSE was reported in the United Kingdom in 1986, spreading throughout the country and lasting for almost a decade. Thousands of classical BSE cases were reported during this period, raising public health concerns across the world. Basic summary statistics and the sources of all the data used in the analysis are placed in the appendix. We collect data on cattle stock and cattle slaughtered, as they contribute to beef production and its export market share. Cattle stock and cattle slaughtered are measured in heads.
To allow for comparison of cattle production across countries, we further construct a cattle 7 stock-to-slaughter ratio. Exchange rates affect the relative prices of beef exports. Given that the exports are not bilateral, we choose the national currency per SDR (special drawing rights) as the preferred indicator. 4 Ideally, we would also like to control for other sources of cattle production costs.
Because not all exporters in our analysis have available producer prices, we instead use the inflation rate of local consumer prices, as the rate of change between consumer and producer prices tend to be similar. 5 Corn futures prices are used as a proxy for feed costs. For the United States, we also include meat slaughtering labor cost to better proxy for the cost of beef production. Such data are not available for the other countries. To capture any underlying technological progress that could also contribute to the change in a country's comparative advantage in beef, we include a linear trend variable. We control for nonlinearities in the following manner. The main interest of the study is to estimate the long-run effect of the BSE outbreak on comparative advantage. Hence, to examine any possible nonlinear impact over time, we generate nonlinear trend variables using the restricted cubic spline function, instead of generating a simple dummy variable indicating the BSE event. 6 We do so by interacting nonlinear trend variables, generated using STATA's mkspline command, with the BSE dummy variable. The five knots are generated at years 1983, 1991, 2000, 2009, and 2016. Given only post-BSE periods are fitted with nonlinear trend variables, only two knots (2009 and 2016) are used. 4 The value of SDR is determined by a basket of currencies, including the British pound sterling, the Chinese renminbi, the euro, the Japanese yen, and the U.S. dollar. 5 Consumer prices for all 12 countries are collected from the World Bank. Data for Argentina is missing after 2013. We fill its missing values from the Bank for International Settlements (BIS). 6 See https://www.stata.com/manuals13/rmkspline.pdf for the discussion of the underlying generating process. 8

Revealed Comparative Advantage in the Beef Sector
To examine whether the U.S. has held a comparative advantage in beef exports, we construct revealed comparative advantage (RCA) indices for the U.S. as well as for eleven other major beef exporting countries. The RCA index was first proposed by Balassa (1977) and reformulated in Balassa (1986) as an empirical proxy for Ricardian comparative advantage. While useful, Balassa's original RCA index features some unsatisfactory characteristics. In particular, the index only indicates whether the country itself has a comparative advantage in a specific product/sector, but it does not hold either cardinal or ordinal properties. Therefore, one cannot compare Balassa's indices across countries or over time. Following recent literature, we adopt the normalized revealed comparative advantage (NRCA) index proposed by Yu et al. (2009) 7 . The NRCA index allows for symmetry and comparability, facilitating its use in an examination of changes to international competitiveness. For a country exporting good (beef in our analysis), the NRCA index is defined as where is country 's export of good , is country 's total export of all commodities, is world's export of good , and is world's total export of all commodities. Under this formulation, a country has a comparative advantage in beef (i.e., advantage if otherwise (i.e., < 0). 8 Figure 2 shows each country's NRCA in beef exports over the sample period.
[Insert Figure 2: Normalized Revealed Comparative Advantage] Australia has a lower beef production cost and indeed, we see that Australia has been leading other competitors in comparative advantage. The U.S. had been enjoying an increasingly strong share in the beef export market relative to its total exports since the 1980s, and started to have a comparative advantage after the 1990s until the discovery of BSE in December 2003.
Like the impact on beef export values, trade bans on U.S. beef resulted in an adverse shock to its competitiveness, where the U.S. NRCA in beef fell to levels last seen in the late 1980s. What figure 2 also shows is that no single country completely snatched the lost U.S. market advantages. Instead, U.S. competitiveness appears to have been re-distributed to a handful of other exporters. This graphical evidence suggests that the market moved toward higher competitiveness all-around. 9

Seemingly Unrelated Regressions (SURs) Estimation
We next move to estimating the impacts of the BSE outbreak on the U.S. and other countries' comparative advantage in beef. We limit the sample period to cover 1981 to 2017,ending just prior to the U.S. engagement on trade renegotiations. We correlate the NRCA index with variables we consider would contribute to its variation. Because the NRCA index is constructed using exports to all importing countries instead of bilateral exports, we can view these exporters as serving the world market together. As a result, we estimate the twelve equations simultaneously as a seemingly unrelated regression system, with potentially correlated, crossequations errors. The system of equations is specified in equation (2): (2) Consistent with equation (1) One concern with equation (2) is that while _ contributes to the variation of NRCA, it is also likely a channel through which the BSE outbreak impacts a country's NRCA in beef. For instance, the U.S. may decrease its cattle slaughter rate in response to import bans from other countries after the BSE outbreak. This suggests that the effect of the BSE outbreak on NRCA can be biased towards statistical insignificance once _ is held constant in the regression. In the extreme case, if one believes that _ is the only channel through which the BSE outbreak impacts NRCA, then one would erroneously conclude that there is no correlation between the BSE outbreak and beef export competitiveness from the regression result after controlling for _ . To address this issue, we implement a first-stage regression of _ on the post-BSE trend variables and obtain the predicted error term for each country: (3) By construction, the predicted error term ϵ �, which we interpret as the residual variation of _ , is not correlated with the post-BSE trend variables. Therefore, we replace _ with ϵ � in equation (2) to estimate the following system of equations: 11 (4) where all variables are the same as in equation (2), except for ϵ �, the residual variation of _ .
To aid in discussion, we standardize the NRCA indices in equation (4), thus the interpretation of the estimated coefficients will be how many standard deviations of change in NRCA given a unit change in a given right-hand-side variable. The results of interest are the predicted values of NRCA from the regression model. Once equation (4) is estimated, we obtain the predicted NRCA indices for all countries. We also predict the NRCA indices under the counterfactual scenario of no BSE outbreak in 2003 by replacing the BSE dummy variable with zero values in the post-BSE periods to study the impacts of the BSE outbreak.

12
For the sake of brevity, although the system of equations is for the top 12 beef exporting nations, the discussion of findings will focus on the U.S. as well as the top five exporters: Australia, Brazil, Canada, Mexico, and New Zealand.

Conclusions
As trade disruptions made headlines in 2018 and 2019, one concern has been the long-run impacts to export competitiveness. Such impacts are difficult to ascertain until more data become available. Phytosanitary emergencies can provide insight into these potential impacts because they cause disruptions that are often expected to be short-lived, similar to trade disputes. What we see however is that even a short-term market closure can lead to long-term consequences to market structure that lingers beyond the phytosanitary event's conclusion. Another criticism of our method might be in misinterpreting why the U.S. (and Canada) suffered for such a long time from the BSE outbreak. It could be that consumers changed their preferences for U.S. beef, something for which our model does not account. This is possible, however, Marsh et al. (2008) studied this issue precisely and conclude that the impacts of BSE on demand come from the trade bans, not from changes in consumer preferences.
Research studying the potential impacts of the China-U.S. trade disputes (2018-19) is important but are limited to descriptive analyses or simulation studies for which changes to 15 market structure (e.g. equilibrium displacement) can only be guessed in the short-run (Marchant and Wang 2018;Balistreri et al. 2018). As more data become available, these studies take on greater information. Even with its "comparing apples to oranges" limitations, the lesson from our BSE case study has an important implication. Markets disrupted do not easily bounce back after the disruption. Using longitudinal data on beef exports that spans before and after the 2003 BSE event, we directly observe longer impacts of a significant, albeit arguably short-lived, trade interruption and show that a country's export competitiveness can take a long time to recover, if at all.