US Import Refusals of Agricultural Products and Their Impact on the Participation of Indian Firms

Abstract

Developed countries are major importers of agricultural and food products from developing countries. While international trade in agricultural commodities has expanded, there has been growing consciousness by consumers about the quality of imported food products. The WTO members are legally authorized to impose sanitary and phytosanitary (SPS)-based maximum residual level (MRL) standards on imported food products to ensure that the imports are free from contaminants. Compliance with SPS-based standards and regulations is challenging for firms belonging to developing countries, for reasons including information asymmetry and lack of technological capabilities. This article uses data on Indian firms, and the SPS measures imposed by the USA on Indian products to analyse how SPS measures impact on the performance of firms, especially on the probability of the firms to participate in the export market and their export earnings. We find that the presence of standards discourages firms to participate in the export market. However, bigger firms are more productive and able to absorb the impact of SPS measures and continue to participate in export markets, even in the presence of these regulations.

Keywords

SPS agreement non-tariff measures standards and regulations firm heterogeneity Melitz model

Introduction

There has been a surge in use of regulations¹ in agricultural and processed food sector across the developed and developing countries in the recent past. Most of these standards relate to food safety and human health issues. The impact of such standards on international trade flows has been mixed—they can have both trade-enhancing and trade-reducing effects. These standards can improve market performance by avoiding trade friction, as the exporters in developing countries can use the regulations as yardsticks for their products. On the other hand, these standards can become costly and difficult to meet, if they are set stringent, that is, above the Codex (Kallummal & Gurung, 2016), or frequently changed to accommodate technological developments (Kallummal & Gurung, 2013). The agricultural exporters and producers from developing countries have to incur extra expenditures to meet such high standards. Given this background, the article aims to test the hypothesis of trade-enhancing versus trade-restricting effects of sanitary and phytosanitary (SPS) measures.² This is done by examining the performance of firms in the presence of SPS standards.

In what follows, a brief explanation is provided with regard to the different types of non-tariff measures (NTMs), before undertaking a review of the literature. The gap in the literature is then discussed, followed by a presentation of the research problem and questions. Thereafter, the theoretical framework, data and methodology are discussed, followed by the analysis and results. The article concludes with the main findings and elucidates the limitations and caveats of the study.

The Rise of Non-tariff Measures³

After several rounds of General Agreement on Tariff and Trade (GATT)/WTO negotiations, there has been a significant reduction in average tariff levels across all products. However, at the same time, there has been a rise in the use of NTMs as barriers to trade. This can be traced way back to 1963, when the first meeting of the Food and Agriculture Organization (FAO) was held in Rome. Figure 1 illustrates the falling average tariff levels for all WTO members and the USA. The figure indicates the most favoured nation (MFN)-applied tariff rates of the Indian exports to the USA for all product categories. Tariffs have declined continuously for the chemicals, manufacturing and mineral categories over the period between 1989 and 2014, while tariff rates for agricultural commodities are fairly constant since 1998. Renewing the commitment towards free trade, the WTO member countries have reduced tariff barriers consistently.

However, there has been a simultaneous growth in NTMs, which is true for WTO members and for the USA (Figure 2). This figure demonstrates how the counts of NTMs have grown in volume over time, from 1995 to 2014. The SPS notifications have increased from a mere 385 to 2198, from 1995 to 2014.

In the case of the USA, the notification increased to 54 in 1995, and then to a peak at 555 in 2007, only to drop thereafter. It is important to note that once a notification has been brought, it will continue to regulate the trade until the same is withdrawn or replaced with a new measure. The rising number of NTMs highlights how these measures are being viewed as a new trade policy tool and used as protection in international trade, against the goal of free trade.

Baldwin (1970, p. 5) defined NTMs as ‘any measure (public or private) that causes internationally traded goods and services, or resources devoted to the production of these goods and services, to be allocated in such a way as to reduce potential real world income’. In the words of Hillman (1996, p. 2), it refers to ‘any governmental device or practice other than a tariff which directly impedes the entry of imports into a country and which discriminates against imports, but does not apply with equal force on domestic production or distribution’.⁴

In other words, NTMs are any measures other than ordinary price-based tariff measures that can potentially have an economic effect on international trade in goods, changing quantities traded, or prices, or both (DITC, UNCTAD, 2010).

A single notification can cover multiple traded products. Figure 3 indicates the magnitude of the impact in the case of US notifications, the SPS notifications being solely applied to agricultural products. The US notifications cover 283 unique agricultural products,⁵ the yearly notifications having aggregated up to 22,327 products at four-digit level. Most of these were duplicate product coverage under different notifications, in over two decades.

Figure 1.

Source: Kallummal, CWS presentation on WTO member countries’ tariffs.

Figure 2.

Source: Adopted from Kallummal.

Figure 3.

Source: Centre for WTO Studies, web portal on SPS and TBT.

The WTO classified NTMs under three headings: (a) those applying to imports, such as import quotas, import prohibitions, import licensing and custom fee; (b) those applying to exports, such as export quotas, export subsidies, export prohibition and voluntary export restraints; and (c) legislation applied behind the border, on grounds of health, technical, environmental, product and labour standards. The focus of the analysis here is on the third type of NTMs, in particular, on the impact of environmental and food safety standards on trade.

The NTMs are increasingly being imposed in international trade. It has been seen that in most cases, the high-income countries have been much more active than developing and least developed countries in the imposition of SPS notifications (Boza & Fernandez, 2016). The literature on imposition of SPS and technical barriers to trade (TBT) standards suggests that such standards can affect market performance. Standards set by the developed countries impose extra cost to producers and exporters in developing countries, which increases the cost of compliance and which also changes with every new standard introduced (Kallummal, 2012).

Exporters often lack the necessary technological capabilities and the information to adhere to these standards, which impedes their capacity to export. Moreover, if similar kinds of products are produced in different countries, standards could raise the elasticity of substitution, with the result being that only the more efficient producers survive in the export market (Bao & Qiu, 2010; Disdier, Fontagné & Mimouni, 2008; Fontagné, Orefice, Piermartini & Rocha, 2013; Jongwanich, 2009; Kallummal, 2012; Maskus, Otsuki & Wilson, 2005). Imposition of standards can also have positive effects on market performance, because by specifying the standards, exporters in developing countries can use them as a yardstick to gauge the expectations of importers regarding food safety. Hence, standards could positively affect market performance as they reduce trade friction and transaction costs by making information available. In addition, standards provide assurance to the consumers in importing countries on concerns related to quality, safety and health. In this way, standards can promote trade (Bao & Qiu, 2010; Moenius, 2004; Saini, 2009).

Most literature shows either one of the effects, or both of the effects. In this light, literature survey is provided and divided into two sections. The first section looks at the studies addressing the sectoral-level impact of the imposition of standards; the second section looks at the firm-level performance in the presence of standards on food safety and human health-related concerns. In the USA, these clearly account for close to 80 per cent of the total SPS measures.

An SPS measure is a sub-category of technical regulation. These measures are applied by WTO members under the SPS Agreement, which addresses the variety of measures used by governments to ensure that human and animal food is safe from contaminants, toxins, disease-causing organisms and additives, and measures to protect human health from pests or diseases carried by plants and animals.

Macro Impact of NTBs on Trade Flows

A study done by the Jongwanich (2009) elucidates the impact of food safety standards, in particular SPS standards, on processed food exports from low-, middle- and upper-middle-income developing countries, from 1980 to 2006. The proportion of processed food exports in total of world food exports registered a huge increase from 44 to 63 per cent in 1980–2006, with the bulk of the increase coming from developing countries. However, this increase in food exports was not systematically distributed across all developing countries. The middle- and upper-middle-income countries fared better in performance relative to low-income countries. To measure the incidence of SPS measures on processed food exports in developing countries, the authors have calculated cases of detention in the US market and the associated export value, from which they have arrived at the export value per detention. This means that higher SPS standards on average decrease the processed food exports from developing countries (Jongwanich, 2009).

The gravity model framework has been extensively employed to study the impact of standards on bilateral trade flows. They found that the higher the shared standards between two countries, the higher the volume in manufactured trade; that is, NTMs have a positive impact on trade flows. However, in the case of agricultural trade flows, the impact was negative. The author attributed this to the cost of adaptation of standards which outweighs the transaction costs involved in the gathering of information about standards. However, Moenius (2004) has not taken into account one important factor affecting the trade flows: the tariff rate. This means that one cannot differentiate the impact on trade flows which results from tariff and non-tariff barriers. Moenius (2004) and Disdier et al. (2008) improved on this aspect and considered both standards and tariffs in their specification in analysing the structure of SPS and TBT measures in agricultural trade. Their sample includes OECD, developing and least developed countries, and 690 products, for 2004. The authors have used a gravity model framework to study the impact of SPS and TBT measures as notified by the importing countries. In order to quantify this impact, the authors have used two measures: the frequency index and ad-valorem equivalents (AVEs) of NTMs. To control for tariff barriers on trade flows, the authors have introduced market access measure at bilateral level. They found that, for trade flows among OECD members, SPS and TBT measures do not have a significant impact. But the SPS and TBT measures significantly reduce the exports going from developing and least developed countries to the OECD countries. This suggested that the developing countries are restrained by a variety of factors, such as technological limitation, lack of financial resources and information asymmetry.

Bao and Qui (2010) examined the impact of TBT measures imposed by China on the country’s imports for 1998–2006. The authors used the frequency index and found that TBT requirements restricted trade, but when they used the coverage ratio, they found that the trade restriction effect of TBT measures is not statistically significant. Moreover, they found that TBTs have a trade-restricting effect on agricultural goods, but a trade-promoting effect on manufactured goods. The authors applied different tools to measure if the TBT had a trade-promoting or trade-restricting impact on agricultural and manufactured goods. They found contrasting impact of TBT measures in the case of trade of agricultural and manufacturing products. The use of different tools can yield different results. Hence, we need to formulate and test our hypothesis on the trade distorting effect of NTMs.

Micro Impact of NTMs on Firm Behaviour

Only few studies have been conducted on the impact of NTMs on trade. Maskus et al. (2005) examine the costs borne by the firms in 16 developing countries in meeting TBT standards. The authors have used survey data by World Bank’s TBT database, which was administered to 689 firms in 17 developing countries, for 2002. The authors have calculated the elasticity of variable cost with respect to the severity of foreign standards. They found that for a firm faced with more stringent TBT regulations, the incremental production costs are higher, and that firms have to employ additional labour and capital to meet these standards. Such an increase in the costs is crucial in determining the export success of the firms.

Fontagné et al. (2013) examined the impact of SPS measures on exports of French firms for 1995–2005. The authors have used the specific trade concerns (STC) database of the WTO, which records data on the concerns pertaining to the SPS measures notified by partner countries which are raised in SPS committees of the WTO. The fact that the particular SPS measure raises a concern means that it constitutes a barrier to trade. This study is an improvement over the other studies which have used Trade-Related Analysis and Information Systems (TRAINS) database relating to NTMs which only provides information on the presence and absence of NTMs that may or may not constitute a ‘barrier’ to trade. The regression analysis was conducted for the behaviour of exporters (in terms of export-market participation, market price and value of the goods shipped) on the SPS measures and variables reflecting firm-specific characteristics, controlling for the importance of export destination and the tariff rate. They found that the presence of an SPS measure decreases the probability of firms to export. In addition, they found that the firms meeting SPS standards tend to raise the price in certain export markets. Therefore, it clearly established the direct link of the SPS measures and exports at the firm level.

Objectives, Data, and Methodology

Based on the survey of literature, we find that the literature does not address the behavioural response of firms in the presence of SPS and TBT measures in the Indian context. In addressing this research gap, the present study looks at changes in the behaviour of firms in terms of participation in export markets and their export earnings when confronted with an SPS barrier. It looks at the effects of SPS standards on export decisions and export earnings of Indian firms. The study uses panel data of Indian firms over 1998–2015.

There are two main research questions that the study addresses. First, do SPS measures reduce the export-market participation of the firms? Specifically, do SPS measures act as an impediment to trade and reduce the participation of firms in export markets? A subset of the first question is whether the size of the firm plays a role in determining export-market participation in the presence of SPS barriers. In this case, the Melitz model hypothesis is tested, to determine whether larger firms are able to overcome these barriers vis-à-vis smaller firms.

The second question concerns the impact of SPS measures on the export earnings of the firm. In one strand of literature on NTMs, it is found that the producers have to incur extra costs to meet the requirements of compliance. The present study examines the impact of these restrictive measures on the export earnings of the firm. Although this does not directly address the actual increase in cost, it does suggest an overall impact on the resources of the firms in meeting the increased requirements by a stringent standard.

Construction and Meaning of Variables and Data Sources

There are three main data sources, the first being for NTMs with trade linkage (at four-digit Harmonised System (HS)), which is collected from the databases on SPS and TBT measures of the Centre for WTO Studies (CWS). The corresponding import refusal based on SPS measure information was collected from the ‘Oasis database’, based on Protecting and Promoting Your Health from the US Food and Drug Administration (USFDA). The second source for firm-wise data is the Centre for monitoring Indian Economy (CMIE) Prowess database. And, lastly, for actual trade flows (at HS six digits), these are collected from the World Integrated Trade System (WITS) Comtrade database. The CWS, Indian Institute of Foreign Trade (IIFT), maintains a unique dataset on SPS and TBT barriers on all WTO notifications submitted to the Secretariat by 162 members. In 85 per cent of the SPS and TBT notifications, countries do not specify the HS codes (or the trade link); the same is updated in the database maintained by CWS at four-digit HS levels. The CMIE Prowess database provides data for nearly 28,000 Indian-listed companies and is comprehensive on financial variables and trade at the company-level (from 1995 to the present). Trade data in value terms and applied MFN tariffs have been extracted from the online database of World Integrated Trade Solutions.

Details of Dataset Preparation

The general assumption for this study is that all the refusals of imports based on SPS and TBT measures by the USA were already notified to WTO.

The first dependent variable, ‘export-market participation’, is a dummy variable, which takes on the value ‘1’, if a particular firm participates in the export market in a given year, and ‘0’ otherwise.

Participation 1 if the firm participates in the export market in the year t 0 otherwise

The second dependent variable, ‘export earnings’, for the firms, was taken from the CMIE Prowess database for exporting as well as non-exporting firms which belonged to identified categories,⁶ for 1998–2015.⁷

The variable SPS_i,s,t was provided by the SPS notification database of the CWS. The study uses the SPS notifications issued by the USA on Indian products at four-digit level and maps it with industry-level National Industrial Classification (NIC) codes, and paper assigns value ‘1’ when SPS is applied in a certain industry in a particular year, and ‘0’ otherwise.

SPS_i,s,t 1 if SPS is imposed 0 otherwise

In order to control for the firm-specific characteristics, we have used the size and visibility of the firm as the other two variables. The ‘sales’ of the firm are taken as a proxy for the ‘size’ of the firm. Both these measures are lagged by a period, as performance of the firm in the year ‘t − 1’ would affect the future export in the year ‘t’.

The ‘size’ of the firm has been defined as the ‘sales’ of the firm (in ₹ million): the higher the sales of a firm, the bigger the firm. The ‘visibility’ of the firm is computed on a yearly basis as the ratio of export earnings of the firm to total export value in that sector. The variable ‘visibility’ depicts shares of the firm’s export value in the total sector’s export value; it can be interpreted as signifying the importance of a firm in total sectoral exports.

Two interaction terms has been used in the article: the first (SPS and size) and the second (SPS and visibility) have been included in order to examine whether the presence of an SPS measure impacts heterogeneous exporters differently. This, in fact, tests the Heterogeneous Trade model, which hypothesizes that bigger firms are more productive, and, hence, are easily able to overcome the ‘adaptation cost’, or the extra expenditure incurred by the firm in meeting the SPS measure.

The variable ‘tariff_s,t’ has been taken from the WITS TRAINS database. This is the tariff imposed by the USA on Indian product lines (HS codes) from 1998 to 2014. Similar treatment of the procedure is followed with tariff lines, as with the other trade variables in HS codes for concordance with firm-level data.⁸

The Melitz Model Framework

The Melitz model establishes a theoretical framework and enables us to analyse the relationship between the exports and productivity of firms. In a simple depiction of the Melitz model, a large number of prospective firms enter the market. The entry is subject to bearing a fixed cost of investment. Firms then draw a productivity parameter from a random distribution. Firms decide to stay in, or exit from the market, conditional upon the productivity drawn.⁹ The Melitz model predicts the participation decisions of a firm based on the fixed cost of entry and the productivity drawn by the firm (Greenaway & Kneller, 2007). In the context of our study, the fixed cost of entry into the market is the cost incurred by the producer in meeting the standards set by the developed countries.

We have adapted the model in the case of our study as follows.

Concordance of Trade Codes (HS) with the Industry Codes (NIC)

The need for this step arose as the data on SPS are in HS codes, while the firms in Prowess are identifies using the NIC codes. Here, a two-fold methodology has been employed: first, the industries facing SPS measures were identified using concordance of HS to International Standard Industrial Classification (ISIC) (NIC). In order to facilitate comparison, the NIC codes (2008) confirm one-to-one with the ISIC revision 4, up to four-digit level of classification, that is, up to ‘class’. The five-digit level of classification, that is, the ‘sub-classes’, is then divided as per the national requirements. In order to concord HS to NIC, a three-step procedure was followed:

the correspondence between HS (combined) to ISIC revision 3 was obtained from WITS;

UNSTATS provided correspondence between ISIC revision 3 and ISIC revision 3.1, and between ISIC revision 3.1 to ISIC revision 4;

the ISIC revision 4 corresponded one-to-one with NIC 2008.

Hence, starting with HS (combined) and moving progressively to ISIC revision 3, ISIC revision 3.1 and then subsequently to ISIC revision 4, the concordance of HS codes to the NIC codes was obtained.

Identification of Agricultural Product Refusals

In order to identify the product categories which have been subjected to SPS measures, we have used a study done by Kallummal and Gurung (2013). In their study, the authors have identified the HS codes of the imported commodities from Brazil, Russia, India, Indonesia and China (BRIIC) economies which have been rejected by the US FDA over 2002–2012.

Regarding the concordance of industry codes with the HS codes, it was observed that one ISIC code corresponded to several HS codes. This is natural, since an industry is a bigger classification and embodies several products. Hence, the key assumption made here is that if an SPS measure is imposed on a product line, then that SPS measure is imposed on the industry as a whole.

For the above-mentioned NIC chapters, the data at the firm-level have been taken from the CMIE Prowess database. Both the HS code and the NIC codes have been taken at the four-digit level of disaggregation, that is, up to the ‘heading’ level.

Table 1 provides the industries which faced SPS measures in India.

Rationale for Including Time- and Sector-fixed Effects

Fixed effects are useful to analyse the effect of variables that vary over time. The assumption behind using a fixed-effects model is that there exists a correlation between the independent variable and the error term. Each entity (in this case, the firm) possesses its own characteristics that may influence the explanatory variable itself and may lead to biased results. By using fixed effects, we remove these time-invariant characteristics. Hence, we can estimate the net effect of the explanatory variables on the response variable.

Table 1.

Industries Facing SPS Measures/Barriers

NIC Codes Faced with SPS Measures	Description
111	Growing of cereals (except rice), leguminous crops and oil seeds
113	Growing of vegetables and melons, roots and tubers
121	Growing of perennial crops
143	Raising and breeding of camels and camelids
1010	Processing and preserving of meat
1030	Processing and preserving of fruit and vegetables
1040	Manufacture of vegetable and animal oils and fats
1050	Manufacture of dairy products
1061	Manufacture of grain mill products
1062	Manufacture of starches and starch products
1071	Manufacture of bakery products
1072	Manufacture of sugar
1073	Manufacture of cocoa, chocolate and sugar confectionery
1074	Manufacture of macaroni, noodles, couscous and similar farinaceous products
1079	Manufacture of other food products n.e.c.
1080	Manufacture of prepared animal feeds
1101	Manufacture of beverages
1102	Manufacture of wines
1104	Manufacture of soft drinks; production of mineral waters and other bottled waters
1200	Manufacture of tobacco products

Source: Compiled by authors from ASI Reports 2014, CSO, MOSPI.

The random effect model, on the other hand, assumes that the error term is uncorrelated with the explanatory variables. Therefore, in a random effect model, the time-invariant characteristics also influence the dependent variable.

For each of the models, the study uses random-effects and fixed-effects estimations. In the first model, with participation as the dependent variable, we have used the probit model, with random- and time-fixed effect specifications. In the second model, we have used random effect, sectoral (NIC)-fixed effect, year-fixed effect, and year- and sector-fixed effect specifications.

Econometric Model and Variable Description

The rationale of using such a model is to find the impact of imposition of SPS measure on the export-market participation and export earnings of firms, controlling for firm-specific characteristics (size and visibility) and applied tariff. The subscript ‘i’ refers to the ‘firm’, ‘s’ represents the sector at the level of the NIC code, and ‘t’ represents time.

\begin{matrix} y_{i, s, t} = α + β_{1} S P S_{i, s, t} + β_{2} \ln {(s i z e)}_{i, t - 1} + β_{3} (S P S_{i, s, t} * \ln {(s i z e)}_{i, t - 1}) \\ + β_{4} \ln {(v i s i b i l i t y)}_{i, s, t - 1} + β_{5} (S P S_{i, s, t} * \ln {(v i s i b i l i t y)}_{i, s, t - 1}) \\ + β_{6} t a r i f f_{s, t} + ϕ_{t} + ϕ_{s, t} + ε_{i, s, t} \end{matrix}

Hence, the dependent variable y_i,s,t is defined in two ways:

the export-market participation of the firm (Model 1); and

the export earnings of the firm (Model 2).

We have constructed two different modes, taking the above variables as dependent variables. The variables ϕ_t and ϕ_s,t represent the time-fixed effects and sector-time-fixed effects. They control for any shocks that occurred through time, or any shocks that affected the sectors.

Limits of Empirical Estimation

We have used a dataset comprising of 1831 firms (in NIC chapters 01, 10, 11, 12) over 1998–2015. This makes it a panel dataset of n cases over t time periods.

For the first model, the dependent variable is export-market participation, which is a binary variable and takes value ‘1’ for the exporting firms and ‘0’ for non-exporting firms. The objective is to calculate the response probability (Wooldridge, 2010):

P (​ y = 1 | x) = P (y = 1 | x_{1}, x_{2}, \dots \dots \dots . x_{n})

where, ‘ x ’ denotes the set of explanatory variables affecting the binary dependent variable ‘y’ (ibid.).

P (​ y = 1 | x) = G (β_{0} + β_{1} x_{1} + \dots \dots \dots \dots . + β_{k} x_{k}) = G (β_{0} + x β)

where ‘G’ is a function whose values strictly lie between ‘0’ and ‘1’, that is, 0 < G(z) < 1 for all real numbers ‘z’.

The cumulative distribution function G(z) for the probit model is expressed as follows (ibid.):

G (z) = ϕ (z) = \int_{- \infty}^{z} ϕ (v) d v

where ϕ (z) is the standard normal density, $ϕ (z) = {(2 π)}^{- \frac{1}{2}} \exp (- z^{2} / 2)$ (ibid.)

The standard normal cumulative distribution function G(z) ensures that the response probability lies between 0 and 1.

For the second model, export earnings are the dependent variable, which is a continuous variable, and for which panel data regression has been used.

Analysis and Results

Summary and Descriptive Statistics

The summary statistics for our data are shown in Table 2.

The descriptive statistics of the data are shown in Table 3.

Table 2.

Summary Statistics

Variable	Observations	Mean	Standard Deviation
Total number of firms	1831
Export earnings (`millions)	4897	841.25	3307.06
Size or Sales (`millions)	16,279	2245	12,517.88
Average tariff	20,474	3.875	3.072
Visibility (`millions)	3390	8278	292647

Source: Authors’ calculations based on various databases.

Among the 1831 firms, 15 per cent participated in the export market over 1998–2015, while 84.56 per cent did not participate. The SPS measures have been imposed on 50.55 per cent of the firms in the data, while on 49 per cent of the firms, no SPS was imposed.

In Figure 6, the orange line depicts the rise in SPS measures as ratios, while the black line depicts the participation ratio of firms. We can observe that over 1998–2015, as the SPS measure ratio has increased (from 0.2 to 0.5 in 1998–2015), the participation of the firms in the export market has fallen (from 0.14 to 0.08 in 1998–2015).

Table 3.

Descriptive Statistics

Variable	Dummy	Frequency	Per cent
Participation	0	29,014	84.6%
	1	5297	15.4%
SPS	0	17,028	49.5%
	1	17,047	50.6%

Source: Authors’ own calculations based on the dataset.

Econometric Results: First Model

The first model for the regression of export-market participation on SPS and other controls was estimated using the data as follows:

\begin{matrix} P a r t i c i p a t i o n_{i, s, t} = α + β_{1} S P S_{i, s, t} + β_{2} \ln {(s i z e)}_{i, t - 1} \\ + β_{3} (S P S_{i, s, t} * \ln {(s i z e)}_{i, t - 1}) \\ + β_{4} \ln {(v i s i b i l i t y)}_{i, s, t - 1} \\ + β_{5} (S P S_{i, s, t} * \ln {(v i s i b i l i t y)}_{i, s, t - 1}) \\ + β_{6} t r a i f f_{s, t} + ϕ_{s, t} \end{matrix}

We have used a probit model for the estimation of this regression. The results of the first specification of the model are shown in Table 4.

Figure 4.

Source: Adapted from Kallummal (2012).

Figure 5.

Source: Adapted from Greenaway and Kneller (2007).

Figure 6.

Source: Author’s calculations based on CWS IIFT database and CMIE database.

The results of the second specification, in which we have taken year-fixed effects, are shown in Table 5.

Table 4.

Results of Random Effects Probit Model

VariablesDep var: Participation	Random EffectsCoefficient(Standard error)	Marginal Effect at Mean
SPS	-1.119***(0.424)	-0.1831587**(0.0866232)
Logsizelag	-0.0599(0.0521)	-0.0106894(0.0109116)
SPSlogsizelag*	0.310***(0.0607)	0.0576063***(0.0122025)
Logvisibility lagged	0.0657*(0.0350)	0.0155512**(0.00724)
SPSlogvisibility lagged*	-0.0799**(0.0383)	-0.0205669***(0.0077587)
Average Tariff	-0.0338(0.0236)	-0.0084114*(0.0050518)
Constant	0.00396(0.381)	–
Observations	2321	2321
Number of unique ID	488	488

Source: Authors’ own calculations based on the dataset.

Notes: Legend ***p < 0.01, **p < 0.05, *p < 0.1.

Table 5.

Results from Fixed Effects Probit Regression

VARIABLESDep var: Participation	Time Fixed Effects Coefficient(standard error)	Marginal Effect at Mean
SPS	-0.926*(0.498)	-.2033338**(0.1035934)
Logsizelag	0.200***(0.0696)	0.0434819***(0.0142225)
SPSlogsizelag*	0.198***(0.0739)	0.0404602***(0.0151453)
Logvisibility lagged	0.0663(0.0420)	0.0126942(0.0088234)
SPSlogvisibility lagged*	-0.0777*(0.0450)	-0.0154338(0.0093815)
Average tariff	-0.0300(0.0260)	-0.0077755(0.0055838)
Constant	-4.686***	–
Observations	2,321	2321
Number of unique ID	488	488

Source: Authors’ own calculations based on the dataset.

Notes: Legend ***p < 0.01, **p < 0.05, *p < 0.1.

In both the models, the coefficient on SPS is negative and significant, implying that the presence of an SPS measure decreases the probability of the firm to participate in the export market. The probability of the reduction in export-market participation is given by marginal effect, that is, on average, the presence of an SPS measure reduces the probability of participating in the export market by firm by 0.2 per cent. This clearly confirmed the result and is in line with the findings in the literature. According to a study by Chen, Otsuki and Wilson (2006), the producers in developing countries are plagued by the problem of information asymmetry. Lack of information poses a severe hindrance to exporters. Disdier et al. (2008) also find the same results. They reason that developing and least developed countries are restrained by their technological and financial capabilities to adhere to the SPS and TBT measures prescribed by the developed countries. One other reason for the high level of information asymmetry has been indicated by Kallummal (2012), who concludes that most of the notifications (nearly 85%) submitted to the WTO secretariat do not contain the information on the HS code. Thus, the interpretation of the measures is left to the discretion of custom authorities ‘behind the borders’. This makes these NTMs opaque by their nature. All these factors result in lower export-market participation by the firms.

The size or sales of the firm in previous periods have had a positive and significant impact on export-market participation in the current period, as indicated by the positive coefficient on Logsizelag in the second specification. On average, a unit increase in the sales of a firm in the previous period leads to an increase in probability of export-market participation in the current period by 0.04 per cent. This means that the bigger firms have a higher probability of participating in the export market. This is due to the operation of scale effect. Larger firms have lower average and marginal costs, leading to greater likelihood of participation in export markets.

Table 6.

Panel Regression of Export Earnings on SPS

VariablesDep var: ln(export earnings)	Random Effect Model	Sectoral Fixed Effect	Time Fixed Effect	Time and Sector Fixed Effects
SPS	-0.492(0.418)	-0.0815(0.441)	-0.281(0.426)	-0.214(0.488)
Logsizelag	0.689***(0.0763)	0.735***(0.0787)	0.687***(0.0714)	0.716***(0.107)
SPSlogsizelag*	0.0511(0.0631)	0.0420(0.0646)	0.0239(0.0622)	0.0648(0.0702)
Logvisibility lagged	-0.0220(0.0378)	0.0435(0.0378)	-0.0283(0.0358)	-0.000554(0.0378)
SPSlogvisibilitylagged*	0.0685*(0.0375)	0.0208(0.0376)	0.0747**(0.0358)	0.000580(0.0370)
Average Tariff	0.0269(0.0278)	0.0690**(0.0316)	0.0106(0.0276)	0.0525(0.0330)
Observations	4829	1646	1646	1646
Number of uniqueID	700	300	300	300
R square	0.3020	0.487	0.313	0.253

Source: Authors’ own calculations based on the dataset.

Notes: Legend ***p < 0.01, **p < 0.05, *p < 0.1.

The interaction term between ‘SPS and size’ is positive and significant in the second specification. Suggesting that bigger firms are more likely to absorb the impact of an SPS measure and continue exporting in the later periods, in spite of the presence of SPS measures, a unit increase in the size of the firm increases export-market participation by 0.04 per cent.

The visibility term is positive, indicating that firms which are more ‘visible’, or which have a higher share relative to their sector, are more likely to export. However, this effect is not found to be significant.

The second interaction term between ‘SPS and visibility’ has a negative sign. This means that in the presence of SPS measure, the firms having higher export shares (i.e., more visible firms) are less likely to participate in the export market. The magnitude of the reduction in the probability to participate in the export market is very low and found to be insignificant. The coefficient on tariffs is negative, suggesting that the presence of tariffs reduces the likelihood of a firm to participate in the export market. However, this is not significant.

The overall picture that emerges is that SPS reduces the probability of export-market participation by firms. However, bigger firms still continue to export even in the presence of an SPS measure. The results conform to the Melitz model: the bigger firms are more productive and able to overcome the fixed cost in meeting the SPS standards, and continue to export. The size of the firm emerges as a major determinant of export success for a firm. Hence, we can sufficiently infer that the SPS measures prove to be an impediment to export faced by firms. Moreover, these measures hurt the small-size firms more in comparison to the bigger firms.

Econometric Results: Second Model

The model for regression of export earnings on SPS was estimated using the data as follows:

\begin{matrix} L n {(E x p o r t e a r n i n g s)}_{i, s, t} = α + β_{1} S P S_{i, s, t} + β_{2} \ln {(s i z e)}_{i, t - 1} \\ + β_{3} (S P S_{i, s, t} * \ln {(s i z e)}_{i, t - 1}) \\ + β_{4} \ln {(v i s i b i l i t y)}_{i, s, t - 1} \\ + β_{5} (S P S_{i, s, t} * \ln {(v i s i b i l i t y)}_{i, s, t - 1}) \\ β_{6} t a r r i f f_{s, t} + ϕ_{s, t} \end{matrix}

Export earnings have been regressed on SPS and other controls. For our main variable, SPS, we find that its presence reduces the export earnings, ceteris paribus. However, this result is not found to be significant. This could be due to the larger weight of large firms and, as established in the first model of the SPS measures, it does not impact the participation of firms.

The effect of size on export earnings is positive and highly significant. On average, an increase in the size of firm by 1 per cent increases the export earnings by 0.68 per cent, ceteris paribus. The size or scale effect of bigger firms results in lower average and marginal cost, which, in turn, leads to higher export earnings. The interaction term between SPS and size bears a positive sign, which implies that bigger firms are able to absorb the impact of an SPS measure, and this does not impact their export earnings. However, this effect is not found to be significant.

The interaction term between SPS and visibility has a positive and significant sign. This suggests that the export earnings of more visible firms are not impacted in the presence of an SPS measure. In the presence of an SPS measure, a 1 per cent increase in the visibility of the firm will, on average, lead to 0.06 per cent increase in export earnings, ceteris paribus. More visible firms would have higher value of export relative to the sector and are easily able to overcome the cost of compliance associated with SPS measures, thereby resulting in positive export earnings. Fontagné et al. (2013) argue that the more visible firms can easily shift their resources from the region where SPS has been imposed, to the non-affected regions. In this way, they avoid the compliance cost and continue to earn positive profits.

The tariff does not seem significantly to affect export earnings. This suggests that tariffs do not play a role in affecting export earnings, ceteris paribus. This is in contrast to the economic theory that suggests that an increase in the tariff rates of the importing country results in lower market access for the exporting country producers. Here, tariffs do not prove to be a significant barrier for the export earnings of the firm.

Conclusion and Policy Recommendations

This study has examined the effects of SPS measures on firm performance using a panel of Indian firms over 1998–2015. The study has controlled for firm-specific factors in terms of size and visibility and the tariff barriers and examined the impact of SPS measures on the performance of the firm. It has been found that the presence of SPS measures prohibit the firms from participating in the export market and that the firms that are larger in size are less impacted and are able to overcome the higher cost of compliance meet by the US SPS-based standards. This suggests that SPS measures impact small firms more when compared to large firms. With regard to export earnings, the SPS measures do not play a significant role in determining the export earnings of firms. In both models, it was observed that the size of the firm was highly significant. This implies that firm size is a major determinant of market access, as well as export earnings. Size is a reflection of past success, as well as the scale effect. Larger firms may be more successful in the past, which determines future export success. In addition, bigger firms may be associated with lower marginal cost or average cost of exporting, which, in turn, increases the probability of exporting for these firms.

From a policy perspective, there are four important issues to be highlighted. First, based on the dataset used, it is found that the SPS measures prohibit the firm from exporting to the US market. This means that Indian exports do not meet the standards as set by the US authorities. There should be concrete efforts to improve the quality of agriculture and food exports from India to make them more competitive in the international market. India should be checking for standards compatibility under Article 5 of the SPS Agreement.

Second, size is a key determinant of response to export measures, even in the presence of barriers to trade. Therefore, policies should be designed to enlarge the size of firms, such that they increase the size or capacity of the exporters and make them more productive. Third, there needs to be harmonization of standards across all the WTO member countries. Harmonization will ensure that developed countries of the WTO do not discriminate against the developing countries by imposing different standards. Finally, there needs to be an efficient mechanism of information dissemination related to these standards, so that the producers are well aware of these standards in advance and can plan their production process accordingly.

There is a clear and present danger of selective market access for developing countries’ exports, and therefore, governments should be pushing for a transparency-related provision under the SPS and TBT agreements and seeking the WTO to mandate mechanisms thereof (Kallummal & Gurung, 2016). Another serious developmental impact of the imposition of SPS measures by the USA could be on employment in Indian firms which have vanished from exporting activity, as the small and medium firms have a higher share for this factor of production.

Footnotes

Notes

References

Baldwin

R. E.

(1970). Nontariff distortions of international trade. Washington, DC: Brooking Institution Press.

Bao

, & Qiu

L. D.

(2010). Do technical barriers to trade promote or restrict trade? Evidence from China. Asia-Pacific Journal of Accounting & Economics, 17(3), 253–278.

Boza

, & Fernández

(2016). World Trade Organization members’ participation in mechanisms under the sanitary and phytosanitary agreement. International Journal of Trade and Global Markets, 9(3), 212–227.

Chen

M. X.

, Otsuki

, & Wilson

S. J.

(2006). Do standards matter for export success? (World Bank Policy Research Working Paper, 3809). Washington, DC: World Bank.

Disdier

A.-C.

, Fontagné

, & Mimouni

(2008). The impact of regulations on agricultural trade: Evidence from the SPS and TBT agreements. American Journal of Agricultural Economics, 90(2), 336–350.

DITC

UNCTAD

(2010). Non-tariff measures: Evidence from selected developing countries and future research Agenda. United Nations, New York & Geneva

Fontagné

, Orefice

, Piermartini

, & Rocha

(2013). Product standards and margins of trade: Firm level evidence (Working Paper 2013–06). France: Centre d’Etudes Prospectives et d’Informations Internationales.

Greenaway

, & Kneller

(2007). Firm heterogeneity, exporting and foreign direct investment. The Economic Journal, 117(517), 134–161.

Hillman

J. S.

(1996). Non-tariff agricultural trade barriers revisited (Working Paper No. 96-2). International Agricultural Trade Consortium. Retrieved 25 December 2015, from http://ageconsearch.umn.edu

10.

Jongwanich

(2009). Impact of food safety standards on processed food exports from developing countries (Working Paper Series, No. 154). Philippines: Asian Development Bank.

11.

Kallummal

(2012). SPS measures and possible market access implications for agricultural trade in the Doha Round: An analysis of systemic issues (ARTNeT Working Paper Series No. 116). Bangkok: ESCAP.

12.

Kallummal

, & Gurung

H. M.

(2013). The United States SPS measures on agricultural products imports: Impact on BRICS countries. Agriculture and trade—Emerging issues and challenges in India. Retrieved 18 September 2015, from https://www.academia.edu/7430130/Agriculture_Products_Imported_by_United_Statts_and_SPS_Measures_based_Refusals_Impact_of_Refusals_by_the_FDA_of_United_States_on_BRIICS_Countries

13.

Kallummal

, & Gurung

H. M.

(2016). Global agricultural trade under science-based MRL standards regime: An impact assessment on India and other developing countries. Retrieved 5 August 2015, from https://www.academia.edu/25879342/Global_Agricultural_Trade_under_Sciencebased_MRL_Standards_Regime_An_Impact_Assessment_on_India_and_other_Developing_countries

14.

Maskus

K. E.

, Otsuki

, & Wilson

J. S.

(2005). The cost of compliance with product standards for firms in developing countries: An econometric study (World Bank Policy Research, Working Paper 3590).

15.

Moenius

Johannes

. (2004, February). Information versus product adaptation: The role of standards in trade. Retrieved from SSRN: https://ssrn.com/abstract=608022

16.

Saini

K. G.

(2009). Non-tariff measures affecting India’s textiles and clothing exports: Findings from the survey of exporters. Retrieved from http://www.igidr.ac.in/pdf/publication/WP-2009-008.pdf

US Import Refusals of Agricultural Products and Their Impact on the Participation of Indian Firms

Abstract

Keywords

Introduction

The Rise of Non-tariff Measures 3

Macro Impact of NTBs on Trade Flows

Micro Impact of NTMs on Firm Behaviour

Objectives, Data, and Methodology

Construction and Meaning of Variables and Data Sources

Details of Dataset Preparation

The Melitz Model Framework

Concordance of Trade Codes (HS) with the Industry Codes (NIC)

Identification of Agricultural Product Refusals

Rationale for Including Time- and Sector-fixed Effects

Econometric Model and Variable Description

Limits of Empirical Estimation

Analysis and Results

Summary and Descriptive Statistics

Econometric Results: First Model

Econometric Results: Second Model

Conclusion and Policy Recommendations

Footnotes

Notes

References

The Rise of Non-tariff Measures³