The role of global economic growth in Pakistani agri-food exports.

Haq, Zahoor Ul ; Gheblawi, Mohamed ; Muhammad, Safdar 等

This analysis uses least squares and Heckman maximum likelihood estimation procedures with fixed effects to explore the role of economic growth in 36 developed and developing economies--categorised as low-, lower-middle-, upper-middle-, and high-income--in explaining their agri-food import of 29 products from Pakistan during 1990 to 2000. We reject the hypothesis that the economic growth of these economies does not influence Pakistani agri-food product exports. However, the estimated income elasticities are statistically elastic only for lower-middle income countries, suggesting that their expenditure on Pakistani agri-food exports will increase disproportionately as their economies grow. Hence, lower-middle-income countries provide good export opportunities for Pakistan's agri-food products.

JEL Classifications: F14, Q17

Keywords: Economic Growth, Agri-food Trade, Income Elasticities, Developing Countries

1. INTRODUCTION

The agriculture sector is still the largest sector of Pakistan's economy despite structural shifts towards industrialisation. The sector accounted for 26 percent of gross domestic product (GDP) in 2000, but gradually shrank to 21 percent in 2007. It employed 44 percent of the total employed labour force in 2007, and is the mainstay of the rural economy around which socioeconomic privileges and deprivations revolve [Pakistan (2009)]. The agriculture sector consists of the crops, livestock, fishing, and forestry subsectors, with the crop subsector further divided into major crops consisting of wheat, cotton, rice, sugarcane, maize, and gram, and minor crops consisting of pulses, potatoes, onions, chillies, and garlic. Historically, the crops subsector accounted for the bulk of the agricultural portion of GDP but its share has been declining since 2000, accounting for 48 percent--a little more than the livestock subsector (47 percent). By 2007, the contribution of the crops subsector had declined to 45 percent while the livestock subsector had increased its share to 52 percent. Since 2000, trade (i.e., the sum of exports and imports) has accounted for about one third of the country's real gross national product (GNP), and agricultural trade for 80 percent of total trade. Hence, the performance of the agriculture sector affects the performance of the country's entire economy.

Pakistani exports are highly concentrated among a few countries and consist of a small number of commodities; consequently, they are vulnerable to external shocks. The major markets for Pakistani exports are the US, the UK, Germany, Hong Kong, and the United Arab Emirates (UAE). Exports to the US accounted for 20 percent, Hong Kong (24 percent), UK (13 percent), Japan (13 percent), and Germany (7 percent) in 2007. Such a high concentration of exports to a few destinations raises the question whether there is any opportunity for Pakistani agri-food exports to other developing and developed countries. This question becomes more important as developing countries outperform developed countries in economic growth and we need to know whether Pakistani exports benefit from this disproportionate global economic growth. It is also important to mention that, due to their rising income, developing countries' share of agri-food trade has increased. They import half of the agricultural products produced by developed countries and export 61 percent of their agricultural products to the latter. Similarly, developing countries as a group are the second-largest traders with the European Union, with exports of $162 billion and imports of $128 billion of agricultural products in 2000-2001 [Aksoy and Beghin (2005)].

This study investigates the role of income in agri-food exports from Pakistan by estimating the income elasticities of developed and developing countries for these exports. The study tests a number of specific hypotheses about the estimated income elasticities. We hypothesise that (i) the income of developed and developing countries does not determine the import of agri-food products from Pakistan, (ii) the income of developing countries does not determine their import of agri-food products from Pakistan, (iii) the demand for Pakistan's exports of agri-food products is statistically elastic in the importing countries, and (iv) the income elasticities of Pakistani agri-food products are the same for developed and developing countries. The results of these tests will also help to understand the heterogeneity of preferences for the country's exports to other developed and developing countries.

The article is organised into five sections. The next section presents theoretical and empirical models. The third section describes the data used in the analysis, followed by a discussion of results in Section 4 and conclusions in Section 5.

2. THEORETICAL AND EMPIRICAL MODELS

We use the theoretical and empirical frameworks developed by Hallak (2006) and modified by Haq and Meilke (2007, 2008, 2010). The framework assumes that demand in each country i is generated by a representative consumer with a two-tier utility function. The upper-tier utility function is weakly separable in sub-utility indices defined over differentiated goods [X.sub.f] where f = 1,..., F and for each homogenous product [X.sub.h] where h = F+1 , ... , H. The sub-utility index [u.sup.i.sub.f] is assumed to have a constant elasticity of substitution (CES) utility function. Maximising the CES approximation of preferences subject to the expenditure on imports generates demand functions for each variety of product f It is further assumed that importing country i consumes different varieties in sector f of the same quality and price. Hence, the value of the bilateral trade flow of country i's imports from country j in sector f in year y ([imp.sub.ijfy]) is given as

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

where [[sigma].sub.f] is the elasticity of substitution between any two products within a sector faced by a consumer in country i; [[tau].sub.jfy] is the trade associated cost between countries i and j for product f; [P.sub.jfy] represents the price of each variety fin country j in year y; [P.sub.jxy] [[tau].sub.jfy] represent the trade cost-adjusted price of the product f, [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] represents the price index of all the varieties and [[bar.I].sub.iy] is the average per capita income of country i, and represents the expenditures made on any sector fin country i since no expenditure data is available.

We assume that trade costs ([[tau].sub.jfy]) are determined by distance (dist), trade partners sharing a common border (DCB), landlocked countries (Landl), island countries (Island), a common language (DComlang), bilateral trade partners colonising each other (DColony), and trade protocol among developing countries (DPTN). (1) This relationship is given in Equation (2) and based on the insights from previous studies [Hallak (2006)].

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

Taking the logarithm of both sides of Equation (1) and substituting for transaction cost ([[tau].sub.jfy]) in Equation (1), we obtain the following equation for the value of imports:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

where [[epsilon].sup.j.sub.if] = (1 - [[sigma].sub.f])[V.sup.j.sub.if]. In Equation (3), [P.sub.jfy] is captured by exporter fixed effects; however, since only Pakistan's exports are being considered, these fixed effects are not required. The variable [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] represents importing country-specific effects, and importing country fixed effects ([[PSI].sub.i]) capture these effects. These importing country-specific fixed effects also allow us to control other unobserved factors such as product quality characteristics and technical and non-technical barriers. The analysis covers 29 agri-food products over 11 years, therefore product- ([[PSI].sub.f]) and year ([[PSI].sub.y])-specific fixed effects are also added to Equation (3) to account for the product and time dimensions. Let [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] so that Equation (3) can be rewritten, including the fixed effects as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

Since the study tests a number of hypotheses that require product-specific income elasticities for low-, lower-middle-, upper-middle-, and high-income countries, the per capita income variable in Equation 4 is split into [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] representing the per capita income of low-income, lower-middle-income, upper-middle-income, and high-income countries, respectively, thereby allowing for different income elasticities. Per capita income ([[bar.I].sub.iy]) is interacted with dummy variables representing the level of economic development to obtain income elasticities for low-, lower-middle-, uppermiddle-, and high-income countries as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

where [D.sup.i.sub.LI], [D.sup.i.sub.LMI], [D.sup.i.sub.UMI] and [D.sup.i.sub.HI] are dummies that represent the development level of importing countries: [D.sup.i.sub.LI] is 1 for low-income countries and 0 otherwise, [D.sup.i.sub.LMI] is 1 for lower-middle-income countries and 0 otherwise, [D.sup.i.sub.MI] is 1 for upper-middle-income countries and 0 otherwise, and [D.sup.i.sub.HI] is 1 for high-income countries and 0 otherwise. Equation (4) is augmented by the income shifters and reproduced below as Equation (6):

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)

Equation 6 is used to test our proposed hypotheses and estimated using ordinary least squares (OLS) and the Heckman maximum likelihood (ML) procedure. The choice of the Heckman selection procedure is motivated by zero-trade flows in the data. Omitting these zeros from the analysis could lead to selection bias [Heckman (1979)]. The Heckman selection procedure corrects the selection bias by including the inverse Mills ratio (IMR) in the regression model. Omission of the IMR from the regression model, when it is statistically significant, leads to an omitted variable bias [Heckman (1979)].

The Heckman selection procedure consists of selection and outcome equations. The selection equation is specified as probit and the outcome equation as the least squares regression equation. Both equations are simultaneously estimated using the ML procedure. The Heckman model can also be estimated in two steps, but we have chosen to use the ML procedure because it estimates homoscedastic standard errors [Greene (2003)]. This is important in the context of this study since we are using cross-sectional data. In the case of the Heckman selection model, the specification of the selection equation is motivated by the earlier studies of Linder and de Groot (2006), Bikker and De Vos (1992), and Hillberry (2002). Finally, the Heckman ML procedure does not directly estimate the IMR but estimates rho and sigma, calculating the arc hyperbolic tangent of rho and the natural logarithm of sigma, and then including these variables in the regression model to control for the selection bias.

3. DATA

The study uses trade data from the World Trade Analyzer (WTA) coveting trade flows from 1990 to 20002 [Statistics Canada (2004)]. The data is organised by the Standard International Trade Classification (SITC), revision 3, at the four-digit level. The agri-food products included in the study are given in Table 1. The countries included in the analysis are given in Table 2. These countries are categorised as lower-income (LI), lower-middle-income (LMI), upper-middle-income (UMI), and high-income (HI), using World's Bank per capita GNP thresholds. The data on GDP and per capita GDP is from the World Bank's World Development Indicators. Estimates of the distance between capitals and border sharing are obtained from the World Bank's website [World Bank (2007)]. The data required for the other gravity variables in the trade model has been compiled from Glick and Rose (2002).

4. RESULTS AND DISCUSSION

However, before discussing the estimated results, it is important to provide an overview of the per capita GDP, population, GDP, per capita GDP growth of the selected countries, and structure of trade between Pakistan and low-, lower-middle-, upper middle-, and high-income economies. The selected countries cover a wide range of importing countries, with real per capita incomes ranging from $92 for Ethiopia to $35,667 for Japan, with an average per capita income of $10,911 during 1990-2000 (Table 2). Similarly, the average population of the selected countries ranges from 3.6 million in Ireland to 1,203 million in China. The inclusion of countries with such diverse economic characteristics helps explain the structure of agri-food trade.

During 1990-2000, the nominal per capita GDP in the world grew at 1.3 percent (Figure 1): lower-middle-income economies accounted for the highest average nominal per capita GDP growth of 4.5 percent, followed by low-income (2.3 percent), high-income (1.9 percent), and upper-middle-income (0.8 percent) countries. However, Figure 1 also shows that growth in high-income economies was more stable than in others. It is also important and relevant that, although the growth in high-income economies was lower than in other economies, the absolute increase in the former's GDP was greater than that in middle-income economies, given that the high-income countries had larger economies.

[FIGURE 1 OMITTED]

Table 3 shows the total value of Pakistan's agri-food exports to low-, lower-middle-, upper-middle-, and high-income economies. On average, Pakistan's agri-food exports were valued at $154.3 million per year during 1990-2000. More than 66 percent of these exports were to high-income economies, followed by 18 percent to lower-middle-income and 14 percent to low-income economies. Upper-middle economies imported, on average, only 1 percent of agri-food exports per year from Pakistan during this period, but these exports showed higher growth (24.4 percent) than those of other economies. Overall, while the data shows a high degree of export concentration in high-income economies, there was higher export growth in the middle-income economies. Also, export growth was more stable in the developing economies (2 percent) than in high-income (7.9 percent) economies, as shown by the coefficient of variation. (3)

The estimated results are compiled in Table 4, while the hypotheses are tested in Table 5. Table 4 shows that importing country-specific effects and commodity fixed effects are statistically significant across all the procedures while time (year) fixed effects are statistically significant only for the Heckman ML procedure. Hence, omitting these fixed effects from the estimated equation would have produced biased estimates. The F-statistics yielded through OLS and the Wald test in the case of the Heckman ML procedure test the hypothesis that all the coefficients in the regression model (except the intercept) are zero. This hypothesis is consistently rejected at a 99 percent level of significance for all the procedures, indicating that the explanatory variables are collectively statistically significant in determining the per capita bilateral trade flows of Pakistani agri-food exports. The explanatory power of the model estimated using OLS shows that 49 percent of the variation in the dependent variable is explained by variations in the independent variables.

The estimated models included variables such as distance, trade partners sharing a common border, landlocked countries, island countries, common language, trade partners that have colonized each other, trade partners colonised by the same coloniser, and protocol on trade among developing countries. It is expected that an increase in distance between trading partners leads to a fall in trade while countries adjacent to each other, i.e., with a common border, trade more. Similarly, landlocked and island countries are expected to trade less while countries colonised by a common coloniser, with a common language, border, and colonial history are expected to trade more. Table 4 shows that the effect of distance on Pakistani agri-food exports is negative and statistically significant. The effect of common borders on Pakistani exports is statistically insignificant, which could be because, with the exception of China, Pakistan does not export intensively to its neighbours India, Afghanistan, Bangladesh, and Iran. The effects of other variables on exports are as expected when statistically significant. The direction of the effects of variables across the estimation procedures is consistent but the magnitudes of the estimated parameters are not directly comparable since the Heckman selection procedure does not directly yield marginal effects. Marginal effects can be generated for the Heckman selection model, but this is beyond the scope of this paper.

4.1. Does Global Economic Growth Affect Pakistan's Agri-Food Trade?

The role of income in explaining the trade of differentiated agri-food products is explored by estimating the income elasticities of low-, lower-middle-, upper-middle-, and higher-income countries, and then testing specific hypotheses concerning the role of these income elasticities. Our analysis considers all commodities collectively and does not draw separate conclusions for different product sectors. The results imply that we can accept the hypothesis that income elasticities are different from 1 for low-income, lower-middle-income, and high-income countries when using either the OLS or Heckman procedures, but not for upper-middle-income countries. Interpreting the results of these hypotheses and income elasticities given in Table 4 suggests that, in the case of lower- middle-income economies, the proportionate increase in their per capita income leads to a more-than-proportionate increase in their exports from Pakistan. The premise that developing countries' incomes do not determine trade is rejected when using both procedures (Table 5).

The individual significance of income elasticities (Table 4) for Pakistani exports shows that low- and high-income countries' incomes do not significantly determine Pakistani exports, when using either the OLS or Heckman procedures. The income elasticity of upper-middle-income countries is statistically insignificant when estimated by OLS but statistically significant when using the Heckman procedure. Hence, the choice of estimation procedure can change the results of the hypothesis testing. However, in the case of upper-middle-income economies, income elasticity estimated using the Heckman procedure is negative, indicating that the growth in per capita income of upper-middle-income countries leads to a decrease in their demand for Pakistani exports. Lower-middle-income countries' estimated income elasticities are statistically elastic, implying that, as their income increases, their expenditure on agri-food imports from Pakistan increases disproportionately. Hence, lower-middle-income countries are viable growth markets for Pakistani exports.

5. CONCLUSION

As the predominant sector of the country's economy, agriculture--including agri-food and cotton products--accounts for 80 percent of the country's exports. However, these exports are concentrated in very few markets, most of them, developed countries. The slow economic growth of developed countries, coupled with the recent financial crises, could negatively affect their demand for Pakistani exports. Using agri-food export data on 29 products exported to 36 developed and developing countries, this study has estimated a series of import demand functions and investigated the role of economic growth in the importing countries in their demand for Pakistani agri-food exports. The analysis shows that lower-middle-income countries are the best growth market for Pakistani agri-food exports since only economic growth in these economies can potentially enhance the demand for agri-food imports from Pakistan.

The overall policy implication of the analysis is that Pakistan should, accordingly, focus more heavily on middle-income economies and take advantage of their rising economic growth. Demand for Pakistani products in developed countries has declined and, given their economic growth and income elasticities, may decline further still. Further, Mustafa (2003) indicates that, compared to developing economies, developed economies have higher sanitary and phytosanitary (SPS) requirements, which Pakistan's weaker infrastructure is not necessarily equipped to deal with. Hence, the country must diversify its exports and take advantage of the higher economic growth in developing economies. However, further analysis is needed to identify those specific countries within the lower-middle-income bracket that drive these results. Such analysis could also determine which individual product sectors to focus on and investigate the rationale for bilateral and multilateral trade agreements to take advantage of the growth occurring in middle-income economies.

REFERENCES

Aksoy, M. and J. C. Beghin (2005) Global Agricultural Trade and Developing Countries. Washington, DC: The World Bank

Bikker, J. A. and A. F. de Vos (1992) An International Trade Flow Model with Zero Observations: An Extension of the Tobit Model. Brussels Economic Review 135: 379-404.

Glick, R. and A. K. Rose (2002) Does a Currency Union Affect Trade? Time Series Evidence. European Economic Review 46, 1125-1151.

Greene, W. H. (2003) Econometric Analysis. (5th ed.) Upper Saddle River, N J: Prentice Hall.

Hallak, J. C. (2006) Product Quality and the Direction of Trade. Journal of International Economics 68:1,238-265.

Haq, Zahoor and K. Meilke (2007) The Role of Income and Non-homothetic Preferences in Trading Differentiated Food and Beverages: The Case of Canada, the United States, and Selected EU Countries. Canadian Agricultural Trade Policy Research Network. (Working Paper 2007-5).

Haq, Zahoor and K. Meilke (2008) The Role of Income Growth in Emerging Markets and the BRICs in Agri-food Trade. Canadian Agricultural Trade Policy Research Network. (Working Paper 2008-10).

Haq, Zahoor and K. Meilke (2010) Do the BRICs and Emerging Markets Differ in their Agri-food Trade? Journal of Agricultural Economics 61:1, 1-14.

Heckman, J. (1979) Sample Selection Bias as a Specification Error. Economterica 47:1, 153--61.

Hillberry, R. H. (2002) Aggregation Bias, Compositional Change and the Border Effect. The Canadian Journal of Economics 35:3, 517-30.

Linders, G. M. and H. F. de Groot (2006) Estimation of the Gravity Equation in the Presence of Zero Trade Flow. (Tinbergen Institute Discussion Paper No. TI 2006072/3).

Mustafa, Khalid (2003) Barriers against Agricultural Exports from Pakistan: The Role of WTO Sanitary and Phytosanitary Agreement. The Pakistan Development Review 42:4, 487-510.

Pakistan, Government of (2009) Economic Survey of Pakistan. Ministry of Finance, Islamabad.

Statistics Canada (2002) World Trade Analyser. Ottawa, Ontario, Canada. World Bank (2007) Worm Bank Development Indicators. The World Bank, Washington, DC. Available at: www.worldbank.org.

Zahoor ul Haq <zahoor.haql@gmail.com> is Dean of the Faculty of Arts at Abdul Wall Khan University Mardan, Khyber Pakhtunkhwa. Mohamed Gheblawi is Assistant Professor at the Department of Agribusiness and Consumer Sciences, United Arab Emirates University, UAE. Safdar Muhammad is Associate Professor at the Department of Agribusiness and Consumer Sciences, United Arab Emirates University, UAE.

(1) Factors affecting the tariff structure between trade partners, such as preferential trade agreements, are not included because Pakistan does not have such arrangements with the countries in the sample for the years 1990-2000.

(2) Although, this study uses data from 1990 to 2000, more recent data shows that the structure of trade has not changed much since 2000. In 2007, Pakistan exported about 64 percent of its agricultural products to high-income countries, 19 percent to low-income countries, 12 percent to lower-middle-income countries, and 5 percent to upper-middle-income countries.

(3) The coefficient of variation (CV) is a normalised measure of the dispersion of a probability distribution and is defined as the ratio of the standard deviation to the mean.

Table 1

List of Selected Agri-Food Products at Four-Digit SITC Level

 Number
No. SITC Description of Cases Percent

1 Apples, fresh 55 1.48
2 Beans, peas, lentils and other 209 5.64
 leguminous
3 Cereal grains, worked/prepared 55 1.48
4 Chocolate and other food preparations 220 5.93
5 Crustaceans and molluscs, fresh, chilled 319 8.61
6 Crustaceans and molluscs, prepared or 55 1.48
 preserved
7 Edible nuts (excluding nuts used for 176 4.75
 extraction)
8 Edible products and preparations 308 8.31
9 Fish fillets, fresh or chilled 55 1.48
10 Fish, dried, salted or in brine; smoked 121 3.26
11 Fish, fresh (live/dead) or chilled 187 5.04
12 Fish, prepared or preserved 110 2.97
13 Fruit otherwise prepared or preserved 22 0.59
14 Fruit, fresh or dried 352 9.5
15 Fruit, temporarily preserved 33 0.89
16 Grapes, fresh or dried 110 2.97
17 Jams, fruit jellies, marmalades 77 2.08
18 Juices; fruit and vegetable 176 4.75
19 Malt extract; preparation of flour 33 0.89
20 Meat of bovine animals, fresh, chilled 44 1.19
21 Oranges, mandarins, clementines, 209 5.64
 and other
22 Other citrus fruit, fresh or dried 66 1.78
23 Other fresh or chilled vegetables 198 5.34
24 Other prepared or preserved meat 22 0.59
25 Potatoes, fresh or chilled 66 1.78
26 Tea 33 0.89
27 Vegetables, dried dehydrated or 154 4.15
 evaporated
28 Vegetables, frozen or in temporary 22 0.59
 preserved
29 Vegetables, prepared or preserved 143 3.86
 Total 3,707 100

 (2) Although, this study uses data from 1990 to 2000, more recent
data shows that the structure of trade has not changed much since
2000. In 2007, Pakistan exported about 64 percent of its
agricultural products to high-income countries, 19 percent to
low-income countries, 12 percent to lower-middle-income countries,
and 5 percent to upper-middle-income countries.

Table 2

Average Real GDP, Population, and Real per Capita GDP of Selected
Countries for 1990-2000

 Real GDP
No. Country Income Level (Million $)

1 Bangladesh Low income 35,957
2 Brazil Lower-middle income 528,485
3 Canada High income: OECD 593,278
4 China Lower-middle income 798,284
5 Colombia Lower-middle income 76,981
6 Denmark High income: OECD 139,319
7 Egypt Lower-middle income 80,270
8 Ethiopia Low income 5,285
9 Finland High income: OECD 100,593
10 France High income: OECD 1,168,904
11 Germany High income: OECD 1,720,911
12 India Low income 350,419
13 Indonesia Lower-middle income 148,019
14 Ireland High income: OECD 64,168
15 Italy High income: OECD 980,106
16 Japan High income: OECD 4,470,770
17 Jordan Lower-middle income 6,952
18 Madagascar Low income 3,341
19 Mexico Upper-middle income 480,735
20 Netherlands High income: OECD 315,712
21 Norway High income: OECD 141,007
22 Peru Lower-middle income 44,866
23 Philippines Lower-middle income 63,697
24 Poland Upper-middle income 133,350
25 Portugal High income: OECD 91,093
26 Romania Lower-middle income 38,072
27 South Africa Upper-middle income 117,730
28 Spain High income: OECD 494,511
29 Sri Lanka Lower-middle income 12,823
30 Sweden High income: OECD 207,236
31 Switzerland High income: OECD 226,814
32 Tanzania Low income 7,662
33 Thailand Lower-middle income 108,525
34 Turkey Upper-middle income 168,673
35 United Kingdom High income: OECD 1,243,523
36 United States High income: OECD 8,155,109
 All Countries 647,866.1

 Population Real per Capita
No. Country (Million) GDP * ($)

1 Bangladesh 116.4 307.1
2 Brazil 161.5 3,265.5
3 Canada 29.4 20,165.8
4 China 1202.9 657.6
5 Colombia 38.5 1,994.3
6 Denmark 5.2 26,587.4
7 Egypt 61.3 1,301.9
8 Ethiopia 57.3 91.9
9 Finland 5.1 19,722.5
10 France 57.8 20,202.6
11 Germany 81.3 21,148.1
12 India 932.5 373.1
13 Indonesia 192.6 765.8
14 Ireland 3.6 17,588.6
15 Italy 57.2 17,121.4
16 Japan 125.3 35,667.3
17 Jordan 4.1 1,670.2
18 Madagascar 14.0 239.0
19 Mexico 90.8 5,279.8
20 Netherlands 15.4 20,415.5
21 Norway 4.4 32,277.0
22 Peru 23.8 1,872.3
23 Philippines 68.4 928.8
24 Poland 38.5 3,461.3
25 Portugal 10.0 9,063.9
26 Romania 22.7 1,675.4
27 South Africa 39.3 2,995.5
28 Spain 39.4 12,526.3
29 Sri Lanka 18.1 703.8
30 Sweden 8.8 23,624.3
31 Switzerland 7.0 32,415.9
32 Tanzania 30.7 249.4
33 Thailand 58.2 1,858.1
34 Turkey 61.8 2,721.7
35 United Kingdom 58.3 21,307.0
36 United States 266.1 30,558.2
 All Countries 111.3 10,911.2

* In 2000 $.

Table 3
Total Value of Pakistani Agri-Food Exports to Low-, Lower-Middle-,
Upper-Middle-, and High-Income Economies during 1990 to 2000
(Million $)

 Low- Lower-Middle Upper-Middle
Year Income Income Income

1990 19.1 (14.4) 14.6 (11.0) 1.1 (0.8)
1991 10.5 (8.6) 21.5 (17.7) 1.1 (0.9)
1992 18.2 (14.5) 17.8 (14.2) 0.5 (0.4)
1993 20.3 (14.0) 17.0 (11.7) 0.8 (0.6)
1994 18.2 (12.9) 16.2 (11.5) 1.7(l.2)
1995 18.5 (13.0) 15.2 (10.7) 1.9(l.3)
1996 28.6 (17.3) 26.3 (15.9) 1.4 (0.8)
1997 26.1 (13.7) 29.2 (15.3) 2.1 (1.1)
1998 25.9 (15.4) 45.1 (26.9) 4.1 (2.4)
1999 24.2 (13.8) 57.1 (32.7) 4.7 (2.7)
2000 34.7 (18.2) 45.3 (23.8) 4.1 (2.2)
Average 22.2 (14.4) 27.8 (18.0) 2.1 (1.4)
 Growth/decay
1990-91 -45.2 47.4 6.4
1991-92 73.4 -16.9 -52.5
1992-93 11.7 -4.8 55.6
1993-94 -10.4 -4.4 100.2
1994-95 1.6 -6.4 11.0
1995-96 55.2 73.3 -26.0
1996-97 -9.0 10.9 55.9
1997-98 -0.6 54.4 90.6
1998-99 -6.7 26.7 14.3
1999-2000 43.6 -20.8 -11.3
1990-2000 81.7 210.9 289.8
Average 17.8 15.9 24.4
CV 2.01 2.06 2.04

Year High-Income Total

1990 97.9 (73.8) 132.6
1991 88.1 (72.7) 121.1
1992 88.7 (70.8) 125.2
1993 107.2 (73.8) 145.3
1994 105.2 (74.5) 141.2
1995 106.3 (75.0) 141.9
1996 109.4 (66.0) 165.8
1997 133.5 (69.9) 191.0
1998 92.8 (55.3) 167.8
1999 88.8 (50.8) 174.8
2000 106.2 (55.8) 190.3
Average 102.2 (66.2) 154.3

1990-91 -10.0 -8.7
1991-92 0.7 3.4
1992-93 20.9 16.1
1993-94 -1.9 -2.8
1994-95 1.1 0.5
1995-96 2.9 16.8
1996-97 22.0 15.2
1997-98 -30.5 -12.1
1998-99 -4.3 4.2
1999-2000 19.6 8.9
1990-2000 8.5 43.5
Average 2.0 4.1
CV 7.86 2.47

Source: Authors' calculations from data.
Figures in parentheses show percentage of total value of trade
within a given year.

Table 4

Heteroscedasticity-Corrected Regression Results for Agri-Food
Exports (Real 2000 Dollars) Using OLS and Heckman ML Procedures

 OLS

Variable Estimate SE (A) p-value

Log of Distance -6.080 3.583 0.090
Common Border 0.044 1.669 0.979
Expenditure Elasticity of:
 Lower-Income Countries -0.331 1.219 0.786
 Lower-Middle-Income Countries 1.995 0.712 0.005
 Upper-Middle-Income Countries -1.089 0.747 0.145
 High-Income Countries 0.186 0.642 0.773
 Landlocked -2.224 3.048 0.466
 Island -0.629 0.478 0.188
 Common Colonizer 10.517 6.209 0.091
 Colony 5.093 3.325 0.126
 Common Language -4.720 3.692 0.201
 Protocol on Trade among Developed
 Countries -4.530 5.472 0.408
 Arc Hyperbolic Tangent of rho - - -
 Log (sigma) - - -
Fixed Effects
 Importing Country 18.0 0.000
 Year 0.7 0.728
 Commodity 26.1 0.000
Summary Statistics
 Uncensored Observations 1531
 Total Number of Observations -
 F-Statistics 18.8 0.000
 R-squared 0.49

 Heckman ML Procedure

Variable Estimate SE p-value

Log of Distance -11.939 4.279 0.005
Common Border 4.247 2.161 0.149
Expenditure Elasticity of:
 Lower-Income Countries -0.136 1.235 0.912
 Lower-Middle-Income Countries 4.146 0.896 0.000
 Upper-Middle-Income Countries -0.069 0.032 0.031
 High-Income Countries -0.556 0.764 0.467
 Landlocked -5.582 3.581 0.119
 Island 0.334 0.598 0.576
 Common Colonizer 29.254 8.027 0.000
 Colony 15.922 4.337 0.000
 Common Language -13.916 4.585 0.002
 Protocol on Trade among Developed
 Countries -14.391 6.557 0.128
 Arc Hyperbolic Tangent of rho 1.594 0.524 0.002
 Log (sigma) 0.838 0.117 0.000
Fixed Effects
 Importing Country 45.8 0.000
 Year 41.1 0.004
 Commodity 82.1 0.000
Summary Statistics
 Uncensored Observations 1531
 Total Number of Observations 3707
 F-Statistics 1345.2 (B) 0.000
 R-squared -

(A) All standard errors are robust.
(B) Represent Chi test statistics.

Table 5

Test of Hypotheses Using OLS and Heckman ML Procedures

 OLS

No. Hypothesis F-Statistics p-value

1 Agri-food imports of low-income
 countries from Pakistan are 1.3 0.264
 statistically different from 1

2 Agri-food imports of lower-middle-
 income countries from Pakistan are
 statistically different from 1 2.1 0.152

3 Agri-food imports of upper-middle-
 income countries from Pakistan are
 statistically different from 1 8.2 0.004

4 Agri-food imports of high-income
 countries from Pakistan are 1.7 0.194
 statistically different from 1

5 The effect of developed and
 developing countries' income 2.8 0.026
 elasticities on trade is 0

6 The effect of developing countries'
 income elasticities on trade is 0 3.7 0.012

 Heckman ML

No. Hypothesis Chi-Test p-value

1 Agri-food imports of low-income
 countries from Pakistan are 1.2 0.275
 statistically different from 1

2 Agri-food imports of lower-middle-
 income countries from Pakistan are
 statistically different from 1 2.0 0.162

3 Agri-food imports of upper-middle-
 income countries from Pakistan are
 statistically different from 1 7.8 0.005

4 Agri-food imports of high-income
 countries from Pakistan are 1.6 0.205
 statistically different from 1

5 The effect of developed and
 developing countries' income 2.0 0.162
 elasticities on trade is 0

6 The effect of developing countries'
 income elasticities on trade is 0 7.8 0.005