Safeguard Measures and Fresh Produce Trade: The Case of U.S. Blueberry Imports

Safeguard measures are used to limit excessive import growth and protect domestic producers from unfair import competition. The global safeguard investigation for blueberries highlights these concerns and raises questions about the relationship between imports, prices, and the wellbeing of U.S. producers. Although the U.S. International Trade Commission (USITC) ruled that imports have not caused serious injury to U.S. blueberry producers, it was important to further examine this issue. In this study, we employ an inverse demand model and dynamic Vector Autoregressive (VAR) procedure linking source-specific fresh blueberry imports from countries like Mexico and Chile to U.S. blueberry prices. Our results mostly support the USITC ruling. Results indicate that declines in U.S. prices are small when compared to the level of import growth. Impulse response functions indicate that import price shocks are not long lasting. Abstract Safeguard measures are used to limit excessive import growth and protect domestic producers from unfair import competition. The global safeguard investigation for blueberries highlights these concerns and raises questions about the relationship between imports, prices, and the wellbeing of U.S. producers. Although the U.S. International Trade Commission (USITC) ruled that imports have not caused serious injury to U.S. blueberry producers, it was important to further examine this issue. In this study, we employ an inverse demand model and dynamic Vector Autoregressive (VAR) procedure linking source-specific fresh blueberry imports from countries like Mexico and Chile to U.S. blueberry prices. Our results mostly support the USITC ruling. Results indicate that declines in U.S. prices are small when compared to the level of import growth. Impulse response functions indicate that import price shocks are not long lasting.

industries from the consequences of excessive import growth and providing temporary relief to producers injured by imports (Becker and Hanrahan 2002). While safeguard policies apply to all goods, agricultural products are especially vulnerable to import competition given the perishability and seasonal nature inherent in agricultural production. In fact, about a quarter of all U.S. safeguard investigations from 1975 to 2018 were for agricultural products (Becker and Hanrahan 2002).
In September 2020, the U.S. Trade Representative issued a request to the U.S.
International Trade Commission (USITC) to initiate a global safeguard investigation to determine whether fresh, chilled, or frozen blueberry imports cause serious injury to U.S.
growers. The safeguard investigation was prompted by agricultural producers expressing strong concerns about the potential negative impacts of increased blueberry imports on prices and profitability. U.S. blueberry imports from Mexico and South America have persistently increased, while domestic blueberry prices have decreased in recent years, leading many to question whether imports were the cause of this decline (Office of the U.S. Trade Representative 2020a). This has been a heightened concern for U.S. blueberry farmers, especially producers in The global safeguard investigation for blueberries and USITC ruling raises questions about the relationship between U.S. blueberry imports and the viability of domestic producers.
The final ruling suggests that domestic price declines are not necessarily due to imports, or that price declines overall from import growth have been minimal. We consider this issue in this study, with a particular focus on fresh (including chilled) blueberry imports. In 2020, fresh blueberry imports exceeded one billion dollars and was the third leading U.S. fruit import behind avocados and bananas. Frozen blueberry imports were smaller by comparison ($234 million in 2020) (U.S. Department of Agriculture 2021). The uniqueness of the fresh produce market justifies a distinct focus for this analysis. Fresh fruit producers face unique challenges because of the short marketing window for sales. Import surges before or during this marketing window could depress prices and ill effect producer profitability (Office of the U.S. Trade Representative 2020b).
The primary objective of this study is to assess how fresh blueberry imports affect U.S.
prices. It must be noted that the impact of imports on U.S. prices necessarily depends on the impact of imports on import prices. For instance, if increases in fresh blueberry imports from Chile or Mexicowhich tend to occur when U.S. blueberries are not in seasondo not result in lower import prices, then it would be difficult to argue that imports put downward pressure on in-season domestic prices. Past studies of the effects of import competition on prices have appealed to the law of one price (Miljkovic 1999), and have primarily focused on import price transmission, i.e., the impact of import prices on domestic prices (Pincinato and Asche 2018;Warr 2008). However, given our focus on how import quantities affect domestic prices, we take a two-step approach in studying this issue (Brester, 1996). First, we examine the relationship between import quantities and import prices. We then examine the relationship between import prices and U.S. prices. Given these relationships, we can infer how fresh blueberry imports affect U.S. prices.
Important to this study is how source-specific imports (e.g. fresh Chilean blueberries) affect import prices and domestic prices. A price-dependent demand framework (inverse demand) is clearly more suited for this type of analysis. The predetermined nature of fresh fruit availability and the likelihood that short-run supply is highly inelastic suggests that an inverse demand framework is conceptually more applicable (Brown, Lee, and Seale 1995). Regarding inverse demand models, the literature offers several functional forms to choose from and studies have provided methods for optimizing on this choice (Barten and Bettendorf 1989;Brown, Lee, and Seale 1995;Holt 2002). As Asche and Zhang (2013) note, there is a lack of consensus on the most suitable functional form in spite of the many studies employing selection methods. In this study, we simply use the most common functional form developed by Eales and Unnevehr (1994) and Moschini and Vissa (1992), the Inverse Almost Ideal Demand System, which is the inverse version of the more popular Almost Ideal Demand System (AIDS model) developed by Deaton and Muellbauer (1980). To derive the relationship between import prices and U.S. prices, we employ a Vector Autoregressive (VAR) model to estimate the dynamic relationship between domestic and import prices. Using the inverse demand and VAR estimates, we assess the impact of import quantities on U.S. prices.
The remainder of this paper proceeds as follows. In the following section, we discuss U.S. global safeguard policies, select investigations for agricultural products, and the research on the implications of safeguard policy. We also provide background information on U.S. blueberry imports. We then present the empirical model, followed by a discussion of the data, estimation procedure, and results. Lastly, we discuss findings in the context of the global safeguard investigation and USITC ruling. We close the paper with final remarks regarding the investigation and analysis.

Safeguard Research
The research related to the implementation of safeguard measures in quite extensive. Studies have addressed the potential damage to industries from import competition, the effects of temporary protection on welfare and imports across countries and industries, the efficiency of trade remedies implemented through Section 201 actions, and the potential for individual country safeguards to be upheld within the WTO. Messerlin (2004) examines both the exposure and use of antidumping and safeguard measures by large developing economies focusing on China and its accession to the WTO. Bown and Tovar (2010) find that countries use antidumping and safeguard measures to reverse commitments of lower tariff bindings, using India as a case study. Nelson (2006) provides an extensive survey of the political economy use of protection including antidumping and safeguards, and subsequently uses a structural gravity modeling framework to show that import volumes and welfare effects of antidumping and safeguard mechanisms are negative, but relatively minor. Similarly, Bown (2009) found that use of safeguard measures increased substantially during and shortly after the 2008 financial crises. Vandenbussche and Zanardi (2010) find much larger negative effects on import volumes across countries from the use of antidumping and safeguard measures, with import volumes decreasing as much as 5.9% for new users of these mechanisms. Beshkar (2010) investigates the role of the WTO's ability to reduce trade skirmishes given its safeguard policies; however, Pickard and Kimble (2007) question whether use of temporary protection under Section 201 will be upheld given the current WTO framework.
There has also been research to determine whether outcomes are different given the use of nondiscriminatory safeguard measures versus discriminatory antidumping policy. Evidence shows that either policy has essentially the same result when considering U.S. steel imports (Bown 2013). Grossman (1985) questioned whether imports have been the most substantial cause of injury to the U.S. steel industry and finds that for the time-period 1976-1983, relief from imports was not warranted, but that when investigating a shorter time frame, 1979-1983, it is less clear whether imports cause substantial harm to the industry. Ryan (2012) investigated the effectiveness of the safeguards implemented for wheat gluten, lamb meat, and circular welded carbon-lined pipes and found that in all three cases, the industries were not restored to sustained competitiveness after the temporary trade barriers under Section 201 were lifted.

U.S. Blueberry Sector
In the United States, cultivated blueberry production for the fresh market has increased by about 50% over the last decade (currently around 179 thousand metric tons). However, demand growth has outstripped supply, increasing the need for imports to satisfy demand (Kramer 2020;Kramer, Simnitt, and Calvin 2020). Table 1

Inverse Demand Model
The Inverse Almost Ideal Demand System (IAIDS) is used to assess how total and sourcespecific fresh blueberry imports affect import prices in the United States (Eales and Unnevehr 1994;Moschini and Vissa 1992). For the analysis, fresh blueberries from the ith or jth source (exporting country) are considered a distinct product (e.g., fresh Chilean blueberries) that is part of the product group imported fresh blueberries. Denote the price and quantity of imported fresh blueberries from the ith source as and , respectively, and total expenditure on all fresh blueberry imports as ∑ , where is the value of fresh blueberries imported from the ith source; n is the total number of sources where both i = 1 2,…, n and j = 1 2,…, n.
Also, denote the conditional import share for the ith source as ⁄ . Given these terms, the IAIDS is specified as follows (Moschini and Vissa 1992).
Equation (1) states that the share of imports from the ith source ( ) is a function of aggregate imports as measured by a quantity index ( ) and source-specific imports ( ). is a random error term.
Following Brown, Lee, and Seale (1995), we use the differenced IAIDS for estimation.

∑
( 2) is the differenced conditional import share and is the jth import quantity in log differences. is the Divisia volume index: ∑ ̅ , which is a measure of change in aggregate imports; ̅ ( ) is the average conditional import share between periods t and t-4. is a random error term. Note that the data used for this analysis are quarterly and the 4 th difference is used to correct for seasonality.
There are two important benefits of using equation (2)  , and are fixed parameters to be estimated. According to demand theory, the following parameter restrictions should hold true: ∑ ∑ (adding-up); ∑ (homogeneity); and (symmetry). The adding-up condition is automatically satisfied since the differenced import shares sum to zero (∑ ). The homogeneity and symmetry restrictions must be imposed on the parameters (Brown, Lee, and Seale 1995).
Using the parameters in equation (2), we can derive price flexibilities with respect to the total imports (scale) and source-specific imports (Eales and Unnevehr 1994). (3) The scale flexibility, equation (3), measures how the price of fresh blueberries from the ith source responds to changes in total fresh blueberry imports. The uncompensated own and crossprice flexibilities, equations (4) and (5), measure how the price of fresh blueberries from the ith source responds to changes in source-specific imports. Whereas equation (4)  We estimate the inverse demand model represented by equation (2) using the generalized Gauss-Newton method in TSP (version 5.0), which is a maximum likelihood procedure for equation systems (Hall and Cummins 2009). We test and correct for autoregressive disturbances using a procedure for singular equation systems developed by Beach and MacKinnon (1979).
Homogeneity and symmetry are imposed on the model.
The vector includes import prices and a representative domestic price (all are in logs), are square coefficient matrices, and is a vector of random disturbances. The advantage of using levels, as opposed to differences, is that the estimates remain consistent regardless of prices being integrated or not. Furthermore, standard inference on impulse responses in levels will remain asymptotically valid, and the inference is asymptotically the same even in the presence of cointegrated prices (Lütkepohl and Reimers 1992;Sims, Stock, and Watson 1990). Using this procedure, we derive impulse response functions to assess how domestic prices respond to import price shocks.
The inverse demand model estimates are reported in Table 2. Note that these estimates are best understood as price flexibilities, which are reported in Table 3. The scale flexibilities are all significant, negative, and indicate that for every percentage increase in total imports, the price of fresh blueberries from Chile, Mexico and Rest of World decrease by 1.27% 1.06%, and 1.12%, respectively, whereas imports from other South America decrease by only 0.46%.
According to Eales and Unnevehr (1994), scale flexibilities are less than -1.0 for necessities and greater than -1.0 for luxuries. This suggests that imports from other South American countries like Peru are relatively more valued by consumers, as reflected by total imports having a smaller effect on prices.
The own flexibilities are highly significant and show a similar pattern, with Chilean prices being the most sensitive to "own" imports (-1.03). That is, for every one-percent increase in fresh blueberry imports from Chile, Chilean fresh blueberry prices decrease by 1.03%. The other fresh fruit CPI as a proxy for U.S. prices, the estimates are not as large (Table 5). The only significant causal relationship is between the U.S. CPI and the import price for Chile (0.09), albeit the estimate for Mexico (0.07) is not highly insignificant (P-value = 0.51). This estimate suggests that the initial impact on U.S. prices is only 0.09% for every one-percent increase in Chilean blueberry prices (0.07% for Mexico), which is considerably smaller by comparison.
Impulse response functions (IRFs) based on the estimates in Table 5 for Chile and Mexico are shown in Figure 3. We do not report the IRFs for the other sources, which are overwhelmingly insignificant. The IRFs show the impact of an import price shock on either the U.S. average blueberry price or other fresh fruit CPI. After the initial price response to an import price shock for Chile or Mexico, the confidence bans for the responsiveness of U.S. prices include the zero axis after the 2 nd quarter, which is indication that the U.S. price response to import prices are not long-lasting. Note that this is the case regardless of how U.S. prices are measured (U.S. average price versus the CPI).

Global Safeguard Implications
The recent global safeguard investigation for blueberries was based on the notion that blueberry imports increased significantly within recent years causing "substantial and serious injury" to the IRFs suggest that this impact would not be long lasting. That is, U.S. prices would likely recover in the following quarter.
The U.S. Bureau of Labor Statistics tracks U.S. blueberry prices at the producer level (blueberry producer price index), but these data are only reported for the second and third quarter due to the periodic nature of U.S. blueberry production. Consequently, we could not use this price series in our econometric analysis. However, we do compare this price series to out-ofseason import changes. In Figure 4, we report deseasonalized quarterly import growth for Chile, other South America, and Mexico from 2012 (first quarter) to 2020 (fourth quarter) and deseasonalized quarterly percentage changes in blueberry prices paid to U.S. producers. Note that deseasonalized growth or percentage changes are based on the following formula. Given what occurred in 2018 and 2019, it is understandable why U.S. producers would assert that increased imports of fresh blueberries affected their prices and profitability.

Conclusion
The global safeguard investigation for blueberries raises questions about the relationship between U.S. blueberry imports, prices, and the wellbeing of U.S. producers. Although the USITC ruled that imports have not caused serious injury to domestic producers, it was important to further examine this issue to understand the factors driving the USITC decision. Fresh blueberry imports have, in fact, more than doubled since 2011, with significant increases in more recent years (U.S. Department of Agriculture 2021). As Kramer, Simnitt, and Calvin (2020) note, the recent growth in fresh blueberry imports from countries like Mexico is "…likely to continue to put downward pressure on [in-season] blueberry prices." However, they also highlight expanding domestic production as also contributing to lower prices. The IRF results in this study suggest that the impact of import price shocks on U.S. prices would not be long lasting and that U.S. prices would likely recover in the following quarter. This suggests that negative shocks to import prices should not be a concern for U.S. producers.
The results of this study should be taken with some caution, primarily due to the limited number of observations. While this limitation does not negate the validity of the inverse demand estimates since the number of observations (36 observations) far exceed the number of products in the system (4 products), it does raise concerns about the reliability of the VAR estimates and IRF results. Unfortunately, we are limited by the data in this regard.
We do not address quantity endogeneity in this study. While issues of demand endogeneity have mainly focused on ordinary demand and endogenous prices, there have been some studies of inverse demand and endogenous quantities. However, these studies have focused on products with much shorter production horizons such as live seafood capture (Huang 2015).
Given the perishable and non-storable nature of fresh produce and the relatively long production season, quantity endogeneity should not be an issue. Note that quantity endogeneity is based on the notion that higher price expectations could result in greater output or quantity supply resulting in small, insignificant, or even positive own-price flexibilities. This is clearly not the case in this study.
The results of this study do not necessarily refute the claim that imports depressed U.S. blueberry prices. Note that the IRFs are based on one-time price shocks, and it could be argued that repeated shocks from persistent import growth could lead to a sustained decrease in U.S.
prices. It could also be the case that the most recent price declines