NATIONALEKONOMISKA INSTITUTIONEN Uppsala Universitet
Examensarbete C
Handledare: Jan Pettersson Författar: Louise Skärvall HT 2011
Does Swedish aid help or hinder bilateral trade
-An empirical study on the effect of Official Development Assistance and Aid for Trade
Louise Skärvall 2012-01-20
2 List of acronyms
AfT Aid for Trade
DAC Development Assistance Committee DCs Developing Countries
LDCs Least Developed Countries OA Official Aid
OECD The Organization for Economic Cooperation and Development ODA Official Development Assistance
OOF Other Officials Flows
GATT General Agreement on Tariffs and Trade GNI Gross National Income
MDG Millennium Development Goal NGO Non-Governmental Organization
SIDA Swedish International Development Cooperation Agency TRA Trade Related Assistance
UN United Nations WB World Bank
WTO World Trade Organization
3 Abstract
A growing consensus regarding the positive benefits of trade has during the last decades attracted actors from the developing world to enter the world market. Official Development Assistance (ODA) and more especially Aid for Trade (AfT) have thus come to play an important role in development strategies. This paper seeks to empirically analyze the link between bilateral trade and ODA and AfT, given by Sweden. The paper uses a gravity model of trade and includes 126 countries during the period 1996 to 2009.
The estimates indicate that both Sweden and its trading partners engage in more bilateral trade when ODA-disbursements are made. It’s further suggested that AfT-receiving- countries export more to Sweden than non-recipients do. The paper also control for differences between geographical areas and show that the ODA effect strongly varies in degree and across regions.
Keywords
Aid and Trade; Gravity Model; Foreign Aid; Official Development Assistance; Aid for Trade; Bilateral Trade; Developing Countries.
1. Introduction
Trade Related Assistance (TRA), also known as Aid for Trade (AfT) is a relatively new focus of the development agenda. Trade has long been considered an important mean for achieving economic growth but it is only in the last decade that TRA has evolved into a key tool for the realization of economic development. The Aid for Trade initiative emerged within the Doha Round of trade negotiation in 2001, aiming to find solutions to trade related problems. The concept entailed strategies that would help developing countries expand their trade trough market access. By doing so, all parties would benefit - developed as well as less developed countries.
The objective of this paper is to empirically analyze the link between Official Development Assistance (ODA), given by Sweden, and bilateral trade. The concept
“bilateral trade” will in this paper refer to trade between Sweden and the developing countries subject to Swedish aid. Since AfT has been an expanding focus area of the developing dimension ever since its launch, I will also investigate its impact on bilateral trade behaviours. A number of investigations have already shown that bilateral aid has an empirical effect on exports from donors to recipient countries. Yet, the effect has shown to vary by donor and over time. Despite the overall goal to find benefits for all parties, only a limited range of literature dealing with the inverse relationship – exports to donor countries. One of the reasons why this relation is important to consider is that many countries spend large sums of money on ODA every year. Sweden spent $US 4.732 billion on ODA in 2008, an amount equivalent to 0.98% of Gross National Income (GNI). This makes Sweden the tenth largest donor in terms of volume (OECD/DAC, 2012), which has also resulted in increased demands of performance of the aid given.
Considering that much ODA consists of TRA flows, it’s reasonable to expect increased trade opportunities, not least between the aid recipients and the donor countries. Despite that donors’ might act according to multiple objectives when dealing with development aid, one would assume that TRA should have a positive impact on recipients’ exports - also towards donor countries. I expect an increased volume of trade for several explanations:
4 (i) I presume that the Swedish ODA and more especially AfT, will create customer relations and distribution channels that will reduce the effective cost of distance between the two trading countries. Aid has shown to have the possibility of speeding up the learning-by-doing process when practicing trade as well as generate reputation and trade habits. Pettersson and Johansson (2011) clarify this connection by showing that aid can create links between the donor and the recipient that will “open doors” and enhance trade with the donor country. By adapting physical and technological inputs used by the donor, the recipient country would also gain an advantage over other countries seeking to export equivalent products and it’s likely that Sweden would choose to import from the developing partner rather than other developing countries.
(ii) I expect exports to industrialized countries to be more desirable for developing countries since it can enhance knowledge “spillovers” and thereby be beneficial to the recipient country in the long run (Grossman and Helpman, 1991). Even if better trade opportunities primary would increase trade with the neighbouring countries due to shorter and cheaper means of distribution, exports to industrialized countries could generate higher revenues over time.
(iii) ODA can be used to secure import flows of strategically interesting products and materials and thus increase exports from the aid recipient country to the donor country.
The method I have chosen for my investigation is the gravity model of trade. The underlying theory of this model is that trade between two countries is explained by the distance between the economic centres of the exporter and importer as well as the size (estimated by GDP and population) of the economies. The model will include 126 countries and cover the period 1996 to 2009. Even though the paper seeks to examine and explain a wide range of the effects that ODA can have on bilateral exports, the analysis and its implications are limited. It is for example probable that ODA has varying effects in different sectors and regions over time and in range. This paper doesn’t disaggregate the ODA by sector, a method that has been proved to explain significant parts of trade behaviours (Johansson and Pettersson, 2011). The paper also won’t examine the causality between ODA and trade, i.e. whether increased trade opportunities attract Swedish ODA flows or if the relation is reverse.
The paper proceeds as follows. Section 2 presents a conceptual framework including a description of ODA and AfT. Section 3 develops the volume and structure of Swedish ODA while section 4 focuses on the theoretical background and the problems associated with aid. Section 5 reviews recent literature on trade and aid. Section 6 presents the model specification and data sources. Section 7 contains the main results and finally section 8 presents the conclusions.
5 2. Conceptual framework
2.1. Official development assistance (ODA)
The Organization for Economic Cooperation and Development (OECD) classifies Official Development Assistance (ODA) as official flows, administered to promote economic development and welfare of developing countries. The ODA includes governmental grants, total net loans, capital subscriptions and other capital flows. They must also meet three criteria:
(i) Be undertaken by the official sector
(ii) Promote economic development and welfare as its main objective (iii) Have a grant element of at least 25%
It excludes Official Aid (OA) 1and Other Officials Flows (OOF)2 and considers neither military nor private aid given by Non-Governmental Organizations (NGOs).The ODA is further classified as bilateral ODA (flows from one donor country) and multilateral ODA (flows from international institutions such as the UN, the EU or the WB). (OECD, 2011)
2.2. Aid for Trade (AfT)
Aid for Trade (AfT) is part of the ODA and includes flows that aim to help developing countries (DCs), in particular the least developed countries (LDCs), to build the trade capacity and infrastructure they need to benefit from a global economy. Aft is a broad concept and hard to define, especially if you take into account that aid itself is a complex concept where one flow often affects several sectors. Following the WTO definition, AfT is considered as:
(i)Technical trade-related assistance: for example, helping countries to develop trade strategies, negotiate trade agreements, and implement their outcomes
(ii)Trade-related infrastructure: for example, building roads, ports, and telecommunications networks that connect domestic markets to the global economy
(iii) Productive capacity building (including trade development): for example, providing support to allow industries and sectors to build on their comparative advantages and diversify their exports
(iv)Trade-related adjustment assistance: helping developing countries with the costs associated with trade liberalisation such as tariff reductions, preference erosion, or declining terms of trade
(v) Other trade-related needs: if identified as trade-related development priorities in partner countries’ national development strategies
(OECD/WTO, 2010)
1 Flows which meet conditions of eligibility for inclusion in ODA, other than the fact that the recipients are on Part II of the DAC List of Aid Recipients.
2 Transactions by the official sector with countries on the DAC List of Aid Recipients which do not meet the conditions for eligibility as ODA or OA, either because they have a grant element of less than 25% or that they are not primarily aimed at development.
6 3. Background
3.1. Trade and the emerge of ODA and AfT
The world has long been optimistic about the prospects of helping poorer countries through external assistance. How to design and where to distribute the assistance has however changed over the decades and it was first during the 1950s and 1960s that trade was recognized as a central element for development strategies. Controlled imports were initially the most important concern in trade policies in developing countries. This strategy arose since many countries stood no chance to compete against industrial economies. But this conception slowly changed after seeing the first successful generation of Asian countries. Their export-oriented industrialization lead the way for a new conclusion; exports could lead to rapid growth and necessary structural changes. The new export-led growth quickly gained acceptance and new trade related liberalizations where introduced to improve the efficiency of the economy by exposing the domestic production to the international market. (Page, 2007) During the Uruguay Round of trade negotiations (1986-1994), trade-related assistance (TRA) was established as one of the key components of ODA strategies to help developing countries negotiate and implement trade agreements (OECD/WTO, 2010).
Despite this new growing conviction, the aim of the development agenda quickly changed during the 1990s and emphasis was suddenly put on poverty reduction rather than exports and economic growth. Although trade was still considered important for the overall trade, its negative impact on the increased gap between rich and poor people could no longer be ignored. As a response to this, the United Nations (UN) and its 192 members made a promise in the year of 2000 to reduce the extreme poverty and fight for development all around the word. This pledge became the eight Millennium Development Goals (MDG) to be achieved by 2015 and has had a major impact on the world agenda for sustainable global development ever since. The Millennium Development Goal 8 (MDG8) Develop a global partnership for development addresses more especially the problems associated with economically and politically complex markets and includes commitments to improve production and export capabilities in developing countries by means of ODA (UN, 2010; 2011). But with inadequate infrastructure, weak product chains and unstable public and private institutions etc., it had and would be hard find benefits for all countries. Subsequently, the already existing TRA scope, developed during the Uruguay Round, expanded and the Aid for Trade initiative emerged within the Doha Round of trade negotiation in 2001. These two initiatives – MDG8 and AfT was together expected to strengthen the link between poverty reduction, trade and economic growth and help developing countries benefit from globalization and liberalization. The Aid for Trade Initiative was further developed during the sixth Ministerial Conference in Hong Kong in 2005 where an integrated framework on Aid- for-Trade was created. The AfT commitments have increased steadily since its launch both in real terms and across sectors and income groups (WTO, 2012) and disbursements reached $US 29 billion in 2009. Sub Sahara Africa surpassed Asia in 2009 and is currently the region receiving the largest share of total AfT with disbursements of approximately $US 13 billion in 2009. (OECD/WTO, 2011)
7 3.2. The volume and Structure of Swedish ODA
The overall target for Swedish ODA is to reduce the world poverty and create fair and sustainable development. To do so, the Swedish International Development Cooperation Agency (SIDA) 3 have defined five areas in focus; (i)Democracy, equality and human rights, (ii) Economic development, (iii) Knowledge, health and social development, (iv) Sustainable development and (v) Human security (SIDA, 2012). The AfT initiative concerns all five areas but isn’t known to be the most prioritised aid strategy when fighting extreme poverty. Sweden is on the other hand known to be one of the most generous donors, considered in relative figures. The standard measurement for developing funding is the ODA as a percentage of Gross National Income (ODA/GNI).
As mentioned in the introduction, Sweden spent 0.98 % of its GNI on ODA in 2008. The country is thus one of only five that have reached the UN target4 of 0.7% (OECD/DAC, 2009). Figure A1 in the appendix shows the Swedish ODA– to – GNI ratio from 1990 to 2009. It demonstrates that the ODA allocations have exceeded the UN target 0f 0.7%
during the whole period. Concerning the geographical distribution, assistance to Sub- Saharan Africa and Eastern Europe has increased over the years whilst assistance to Latin America and Asia has decreased (OECD/DAC, 2009). Figure A2 in appendix illustrates that Sub Sahara Africa is the largest ODA receiving region followed by Asia. Figure A3 helps clarifying this by showing that 7 of the top 10 country recipients are located in Africa. Swedish ODA has during the last couple of years been distributed to a reduced number of countries and sectors. If we turn to Figure 4 we can see that the largest sectors for Swedish ODA in actual numbers have been: (i) Democratic governance and human rights; (ii) Natural resources and environment; (iii) Humanitarian assistance and (iv) Health. These sectors are all in line with the overall target.
4. Theoretical Background
4.1. International trade theories and aid
International trades theories have long argued that increased openness of an economy will boost economic growth and raise a country’s welfare. Behind this theory lies the assumption that trade allows countries to exploit their productive potential by permitting it to specialize according to the theory of comparative advantages. By doing so, the returns to those factors of productions which are less scarce in the country will improve and all trading parties will benefit. Another general argument is that developing countries may need net capital inflows to be able to match the opportunities caused by trade. A traditional macroeconomic approach is that bilateral transfers can be used to alleviate the lack of investment in the developing country and thus be one way of securing productivity and income (Page, 2007). ODA has thereby come a piece of the puzzle of the reallocation of resources and is now representing one of the most important external financial resources for developing countries (UN, 2011).
3 SIDA is a government organization under the Swedish Foreign Ministry and administer approximately half of Sweden's budget for development aid.
4The target refers to a repeated commitment of the developed countries to spend 0.7 % of their GDP on ODA every year. The target was first pledged in a 1970 General Assembly Resolution and later reaffirmed in e.g. the 1992 United Nations Conference on Environment and Development, Rio de Janeiro and at the 2002 UN International Conference on Financing for Development, Monterrey.
8 4.2. Problems associated with aid
In recent decades, new liberalizations have taken place under the WTO and the General Agreement on Tariffs and Trade (GATT), forcing developing countries to change to new markets and products. Several studies have shown that poor people have gained from these trade liberalizations but the WTO also acknowledges short turn losses for some developing countries due to disadvantages on the international market (e.g.
Drabek and Laird, 2001; Page, 2007 and UN, 2011). One way of meeting these additional costs from WTO settlements have been to strengthen substantial assistance to developing countries as explained in the background section.
But foreign aid is also associated with transfer paradoxes, meaning that the country receiving aid might attain lower levels of utility than the donor country and sometimes even suffer losses. One explanation is that large inflows of foreign currency might appreciate the exchange rate in the developing country and cause a non-trade-goods effect. This is comparable with a price boom in countries suffering from the Dutch disease5. The appreciated exchange rate changes the prospects of trade, and export and import-substitution products become less preferable as exports become more expensive.
The existing capital will flow into the market of non-tradable goods and cause a pro import bias with no new revenues from abroad. The increased flows of imports will eventually depreciate the exchange rate and raised exports will sooner or later be the only way to stimulate growth as the domestic sources of demand will decrease (Page, 2007)..
Hence, new aid flows are most likely to be required to make up for income losses.
(Suvwa-Eisenmann and Verdier, 2007). This threatens to permanently reduce the productivity and thereby worsen the prospects of economic growth. Despite this recognition, the general perception is that both donors and recipients can benefit from aid if it’s properly “spent” and “absorbed” (McKinley, 2005).
One other theory is that aid crowds out domestic savings. Even though one would expect aid flows to have a positive effect on savings and investments within the developing country, there is a risk that the new capital will replace the domestic commitments. To avoid this problem donor countries have a history of tying their aid.
This means that the donor obligates the recipient to spend the foreign aid on imports from the providing country. As late as in the early 1990s, approximately 50% of all foreign aid was tied by the donors (Wagner, 2003). There are many reasons to believe that this type of aid generates an immediate impact on the imports to the recipient countries from the donor and tied aid has therefore been compared with export subsidies. The subsidies indicate in this situation that the donors seek to increase their own production (e.g. Tajoli, 1999 and Osei et al, 2004). Few empirical investigations have been done on the relative effectiveness of untied and tied aid. Yet, the OECD-DAC adopted a recommendation in 2001 to untie much ODA to LDCs after finding evidence of raised costs of goods, services and works by 15-25%, (OECD, 2009). As a respond to this, Sweden was the first country to present an integrated policy for global development in 2003, the so-called Government Bill 2002/03:122 Shared responsibility - Sweden’s policy for global development. The Bill presented a new objective for the development work in interaction with the common developing goals where untied ODA flows to the LDCs would be one way of securing the efficiency of cooperation activities. Sweden reported the tying status of to be 0.1 % of total ODA disbursements in 2009 (OECD, 2010).
5 The Dutch disease is primarily associated with a natural resource discovery and refers to the Dutch economic crisis that arose in the 1960s when natural gas was found in the North Sea. It can thus result from any large increase in foreign currency, including foreign direct investment, foreign aid or a substantial increase in natural resource prices.
9 5. Recent literature on trade and aid
It has long been difficult to prove that ODA creates bilateral trade opportunities and reduce poverty. How to measure and investigate the effects of aid in different countries and over time have proved to be especially hard. One of the main problems has been the establishment of a solid model which is comprehensive for all regions and types of aid flows. To be able to investigate the outcome of country specific donor-funded AfT programs, case-studies have been conducted as a mean to evaluate the results of these strategies. Many of the case studies have been initiated by the donor countries themselves and the most recurrent conclusion have been that the volume of total exports is likely to increase if receiving aid. But that this relationship is of course complex and hard to substantiate. Another common positive conclusion is that AfT programs have opened up for national trade dialogues and increased the awareness of trade relates policy issues.
(OECD/WTO, 2011 and WTO, 2012)
Despite problems when analyzing the effects of aid programs, a number of studies have been dedicated to investigate the link between aid and trade. Several studies have focused on bilateral trade, which in most cases also demonstrated positive correlation between aid and donors’ exports (WTO, 2012). The first one (to my knowledge) is Wagner’s study from 2003 where a gravity model of trade is used to statistically test the link between aid and export expansion. The paper investigates the correlation between exports and aid (tied and untied) and the results suggest that exports increase with 133%
of the aid. Wagner also makes comparisons between donors and finds that Japan, a country which previously has been accused of using aid to achieve commercial advantages, doesn’t increase its exports more than an average donor. This is consistent with the assumption that aid creates bilateral links and supports the underlying expectation that both developed and less developed countries can benefit from aid.
Another example is Martinez-Zarzosos et al. study from 2008 where a static and dynamic gravity model of trade is used to investigate the effect on aid to 138 recipient countries during the period 1962 to 2005. By distinguishing between bilateral ODA (given directly by Germany) and multilateral ODA (given by the European Community), the authors manage to show that German bilateral aid has a positive correlation with exports whereas the effect doesn’t apply to multilateral aid. This may be due to several reasons, including the fact that the EU gives more humanitarian aid that doesn’t seek intensified trade. This investigation also suggest that the impact of aid on trade depends on the countries involved in trade, the type of aid given and that the impact can change over time.
The range of literature dealing with the opposite relationship – exports from recipients to donors has on the other hand been very scarce. Pettersson’s and Johansson’s article from 2011 is one of few example and they use a gravity model to investigate the relation between bilateral foreign development assistance and bilateral exports in 184 countries between 1990 and 2005. In their article they demonstrate that bilateral aid is associated with donor exports as well as recipient exports to donors. By using disaggregated forms of ODA (technical assistance, general budget support, AfT, different export categories etc), they manage to interpret the effect of different types of flows on trade. They manage to show that the positive correlation between aid and exports is linked with both donors and receivers. This result suggests that aid creates customer relations and distribution channels that can benefit both parties. The results showed also (unfortunately) that AfT had little impact on trade but that the aid-trade link was additionally strong between strategically materials and recipients exports.
10 6. Specification and estimation of the gravity model
6.1. Model Specification
The method I have chosen for my investigation is the gravity model of trade, named after its analogy with Newton´s law of universal gravitation. The underlying theory of this model is that trade between two countries is explained by the distance between the economic centres of the exporter and importer (-) as well as the size [estimated by GDP (+) and population (+)] of the economies. The model was first employed by Tinbergen in 1962, who adapted the normal gravity model to economic purposes, and is today considered one of the most successful empiric models in economy (Anderson and van Wincoop, 2003). The model is used to predict bilateral trade and is normally expanded to include other factors that can facilitate or prevent trade between two countries. Examples of such variables include common official language, trade agreements, common currency, colonial ties and level of democracy, corruption and stability. Given that this paper only studies the relationship between Sweden and its trading partners, I find several of these factors irrelevant and exclude them (e.g. Sweden doesn’t have a modern history of colonization, no other country except Finland shares Swedish as official language and the currency is only viable within the national borders). Yet, level of democracy, corruption and stability would have been very interesting to include for several reasons. As mentioned before, donor countries act according to multiple objectives and aid programs can thus be influenced by political and economical goals. It is therefore likely that these factors have a countable affect on aid distribution. Unfortunately, because of the limited scope of this paper as well as difficulties in estimation, these variables won’t be included in the model.
A general estimation of the traditional gravity model of trade can be specified as:
where
Ln denotes variables in natural logs
represent trade flows between the trading countries.
GDP indicates the GDP of each country Pop is the population in each country
Distance is the geographical distance between the economic centres the trading countries is the error term
As indicated above, the gravity model of trade has evolved over the years to explain a wider range of the trade patterns. Time has for example shown to have a significant effect on trade behaviours because of factors such as inflation, the business cycle and economic policies. To account for this, year-dummies are added to the traditional gravity model.
Anderson and van Wincoop (2003) have argued that the so called multilateral resistance – the resistance to trade with one country in relation to trade with other countries, can clarify a substantial amount of the variance not explained by the traditional gravity equation. By including country-specific-dummies I hope to be able to account for a number of these trade friction related factors. Consequently, variables that are time invariant (in this case distance) are dropped from the equation6.
6 These variables are already included in the country-dummy and cannot be estimated
11 The gravity model is specified as:
(1) where
c represents the trading countries s represents Sweden
t is the period of time
Ln denotes variables in natural logs
represent the exports to Sweden from country c / Swedish exports to country c, in period t
GDP indicates the GDP of respective country deflated into 2009 $US constant prices Pop is the population in country c in millions
represents year dummies
denotes country-specific dummies
is the error term
6.2. Model specification including ODA variables
To study the ODA impact on bilateral exports, I will add ODA to the typical gravity model of trade. It is likely to assume that ODA flows have a delayed effect on trade behaviours. This can be problematic when measuring the effects of the ODA flows, especially since it’s distributed in several different ways, with varying results in degree and over time. To correct for this, one-period lagged ODA flows (t-1) will be used. I use dummies for NODA (No ODA) that will take the value of one in those cases where there are no ODA flows. (i.e. ). The coefficient measures the ODA elasticity of exports giving that ODA is received (given). Exports from (to) ODA receivers (giver) exceeds exports from (to) no ODA receivers when
.
The extended gravity model is specified as:
(2) where
is the log of the ODA that country c receives from Sweden year t-1 is a no ODA-dummy taking the value of one when
6.3. Model specification using disaggregated aid
To investigate the difference between AfT and other types of ODA, I then add two new variables to the extended gravity model; (i) total Swedish ODA excluding AfT (ODA-net-AfT) and (ii) Swedish AfT. Both variables are expected to have a positive effect on the exports since ODA is expected to create customer relation between donors and recipients. In order to investigate this connection, I will have to turn to disaggregated forms of aid. This means that you divide the aid into subsectors and subtract them from the total ODA. Following Pettersson and Johansson (2011), I disaggregate AfT from total ODA by summarizing three subsectors:
12 (i) Trade policy
(ii) Trade-related infrastructure
(iii) Capacity-building to improve production and export capacities
These sectors are reported by The OECD – DAC database - The Creditor Reporting System online database (CRS) and are specified in appendix. I also include dummies in the same way as in equation (2) where AfT (No AfT) take the value of one where . The coefficient measures the AfT elasticity of exports giving that AfT disbursements are made. Exports from (to) AfT receivers (giver) exceeds exports from (to) no AfT receivers when . The coefficient
measures the total ODA-net-AfT elasticity of exports giving that country c (donor) is receiving (giving) ODA. Exports from (to) ODA receivers (giver) exceeds exports from (to) no ODA receivers when Since few studies have been able to prove that multilateral aid promote bilateral trade (e.g.
Martinez-Zarzoso et al., 2009), this variable won’t be taken into account. Because the model only includes bilateral trade flows, conclusions also won’t be made on the effects of ODA and AfT on the overall trade.
The extended gravity model including AfT flows is specified as:
(3) where
denotes total ODA net of AfT [i.e. ln(ODA-Aft)] given by Sweden to country c, year t-1
denotes AfT given by Sweden to country c, year t-1
is a no AfT-dummy taking the value of one when
6.4. Data sources and variables
The investigation covers the years 1996-2009 and includes 126 developing countries and territories from the DAC List of Aid Recipients7 for which there is available data.
Data over bilateral ODA flows are collected from the OECD Development Assistance committee’s online database International Developments Statistics. This database is divided into The Development Assistance Committee online database (DAC) that report aid disbursements and The Creditor Reporting System online database (CRS) that report aid commitments. Since I’m only interested in the actual disbursements and not the reported commitments during this particular period of time, I use ODA statistics from the DAC database. Unfortunately it doesn’t report data of disbursement on disaggregated ODA and I have to turn to the CRS database to locate statistics from the different sectors.
The respective sectors are, as mentioned above; (i) Trade Policy and Regulations, (ii) Building Productive Capacity and (iii) Investments in Trade-Related Infrastructure.
Following Pettersson and Johansson (2011) I assume that the share of commitments is the same as the share actually disbursed in each sector, and the data from the CRS database will be used as a measure to disaggregate total disbursements. There are several problems
7 The DAC List of Aid Recipients shows all countries and territories except G8 members, EU members and countries with a firm date for entry into the EU, eligible to receive ODA.
13 associated with this method of estimating aid. It is fore example likely to over- or underestimates aid flows which can result in incorrect conclusions regarding the impact of aid on trade. A commitment is nothing more than a commitment and it is impossible to know if, when or where the aid will be distributed. The purpose of disaggregating aid is, despite a substantial risk, the advantage to easily compare different aid coefficients. In this case it helps to examine the effect of AfT on bilateral trade - the purpose of this paper.
Data over exports (reported as Swedish imports and exports) is collected from the OECD online database International Trade and Balance of Payments Statistics. Income (GDP) and population variables are obtained from The World Development Indicators Database. As exports (Swedish imports and exports) and GDP is reported in current $US, the variables are deflated to constant $US using a GDP price index of 2009 calculated by the OECD Economics Directorate.
7. Results
7.1. Results using the gravity model of trade (excluding ODA)
The results of the regression using the gravity model of trade (excluding ODA variables) are presented in Table 1. Column (1) and (3) estimate the traditional gravity model (excluding distance) and column (2) and (4) estimate specification (1). Due to the large size of the dataset, country and year dummies are not shown in the table.
Table 1 Bilateral trade
(1) (2) (3) (4)
DCs’ exports DCs’ exports (including year and
country dummies)
Swedish exports Swedish exports (including year and
country dummies)
lnGDP 1.351*** 1.184*** 1.341*** 0.934***
(GDP country c) (0.0393) (0.179) (0.0588) (0.101)
lnPop 0.0454 -0.248 0.381 0.0422
(population country c) (0.0403) (0.567) (0.295) (0.317)
Constant -12.28*** -9.936*** -10.53*** -6.553***
(0.310) (1.672) (0.388) (0.909)
N 1684 1684 1762 1762
R2 0.612 0.840 0.902 0.905
adj. R2 0.612 0.826 0.895 0.897
Notes: The dependent variables are log exports of the recipients (column 1 and 2) and log exports of Sweden (column 3 and 4). Year and country dummies are not shown. Standard errors in parentheses
* p < 0.05, ** p < 0.01, *** p < 0.001
Turning to Table 1, we find that GDP has a large and highly significant effect on exports. This result follows the assumption stated in the traditional gravity model - that larger economies have a tendency to engage in more trade. In this case it means that Sweden trade (export and import) more with DCs with higher GDP. This relationship is expressed in absolute figures and makes however no suggestions about the relative relationship. The estimated parameters using the traditional gravity model show that the effect on exports, more or less is the same for all trading parties (1.351≈1.341). Yet, the effect seem to be smaller for Swedish exports than DCs exportswhen country and year
14 dummies are included (0.934<1.184). If we study the population variables, we find that all coefficients are insignificant and show different signs and levels. Three out of four coefficients indicate that the size of the population in the DC is positively correlated with the volume of exports. This is very probable as more people indicate more potential buyers. Several authors (e.g. Martinez-Zarzosos et al., 2008) have still argued that countries with a larger population usually engage in national trade as the domestic source of demand is high. This could help explain the negative coefficient presented in column (2). If we compare the estimations from the two models, it becomes clear that it is important to control for specific unobservable heterogeneity along the recipients as well as time fixed effects. The following estimations will therefore include both country and year dummies.
7.2. Results using specification (3) and (4)
Table 2 includes ODA flows and show estimations using specification (2) and (3).
The depended variable in column (1) and (2) is the log of ODA recipients’ exports whilst column (3) and (4) uses the log of Swedish exports.
We then turn the table, we see that the GDP and population parameter estimates from Table 1 are essentially unaffected by the inclusions. Country and year dummies (not shown) also confirm similar estimations as before. When we analyze the parameters specified in equation (2), ODA is estimated to have a zero effect on recipients’ exports and a positive and significant effect on Swedish exports. The later result matches the outcome of earlier studies and shows that ODA is positively correlated with donor exports. Turning to the no-ODA-dummy, a similar effect on recipient exports is also observed. The estimated parameter imply a great but negative effect (-0.466) on exports for those countries that do not receive ODA. This suggests that that exports of no-ODA- receiving-countries only accounts for about 63%8 of the exports by ODA receivers. One interpretation of this is that ODA indeed affect DC’s trade behaviours. To better understand the relation between ODA and bilateral exports (despite insignificant results), I will apply the ODA elasticities from column (1) and (3) on Tanzania. I will use statistics over bilateral trade flows and ODA disbursement, obtained from the data sample.
Country and year dummies will not be included in the calculation.
Example: Tanzania’s exports amounted to 6.3 million dollar in 2009 whereas Sweden’s exports were 69.8 million dollar. The bilateral assistance reached 111.33 million dollar (using one year lagged data). If we only look at the elasticity of Tanzania’s exports where =0.00757, a 5 % increase of ODA (5.5665 million dollar) would increase exports by approximately $US 42.000. This is about 0.7 % of the total exports. If applying the same method to proxy the effect for Sweden where =0.0440, exports would raise with approximately $US 245.000. This effect is much larger and accounts for about 3.9 % of total exports. A greater volume of ODA implies thus increased exports for both Sweden and Tanzania.
8 (Exp (-0.466)-1)*100= -37.2493 ≈ 37%. Exp (-0.466) = 0.62750 ≈ 63%.
15 Table 2 Bilateral trade and ODA
(1) (2) (3) (4)
ODA effect on recipients’
exports
ODA/AfT effect on recipients’
exports
ODA effect on Swedish exports
ODA/AfT effect on Swedish exports
lnGDP 1.186*** 1.158*** 0.917*** 0.911***
(GDP country c) (0.178) (0.180) (0.101) (0.102)
lnPop -0.291 -0.253 0.0355 0.0619
(Population country c) (0.565) (0.566) (0.318) (0.319)
lnODA 0.00757 0.0440*
(lagged ODA) (0.0327) (0.0183)
NODA -0.466** -0.234 -0.0440 0.0153
(no ODA dummy) (0.169) (0.146) (0.0941) (0.0824)
lnODA_AfT -0.0164 0.0278
(lagged ODA-net-AfT) (0.0311) (0.0175)
lnAfT 0.0148 0.0179
(lagged AfT) (0.0273) (0.0156)
NAfT -0.241* -0.0273
(no AfT dummy) (0.107) (0.0618)
Constant -9.788*** -9.517*** -6.412*** -6.364***
(1.668) (1.684) (0.911) (0.921)
N 1681 1681 1758 1758
R2 0.842 0.842 0.905 0.905
adj. R2 0.828 0.828 0.897 0.896
Notes: The dependent variables are log recipient’s exports/ log Swedish exports. The regression includes year and country dummies (not shown). Standard errors in parentheses * p < 0.05, ** p < 0.01,
*** p < 0.001
Turning to column (2) and (4), the total ODA variable is dropped and disaggregated forms of aid are added according to specification (3). The ODA parameter, when excluding AfT flows, is still positively correlated with Swedish exports whilst a negative correlation is observed with recipient exports. The estimations also suggest that DCs receiving ODA will export larger volumes than no receivers whilst Sweden will export more to countries not receiving ODA. These results are all contradictory as well as insignificant. Even though the estimations fail to show an effect on the bilateral trade, it’s important to consider possible explanations for such a result. One explanation is that disaggregated forms of aid, based on commitment shares, are associated with under and/or overestimations. This is described in section 6.4 and indicates that there is a possibly that the disaggregated ODA-variables are incorrect. This in turn can lead to erroneous estimates of the actual correlation. Another likely explanation is that the models don’t account for the complex causality chain between ODA and trade. I.e. the model doesn’t control whether increased trade opportunities attract Swedish ODA flows or if the relation is reverse. It is often argued that DCs that don’t engage in much trade with a donor country, neither receive aid. This is often related to economical goals established by the policy makers in the donor country. Yet, a reverse causality suggests that DCs that do not receive aid, neither engage in bilateral trade. This relation is
16 supported by the assumption stated above - that aid creates customer relations and other bilateral links between the trading countries. If we also assume that much ODA is given to countries suffering from acute problems like natural disasters or civil wars, it is plausible that aid exhibits a negative correlation with exports. This correlation often applies to humanitarian aid that doesn’t seek intensified trade opportunities. Given the heterogeneity of aid motives, it becomes clear that a limited subdivision of types of ODA can be very problematic. Another important aspect to consider is however that none of the models presented in this paper, aims to investigate the link between Swedish aid and total exports. Improved trade capacities increase, with high probability, the possibilities to engage in trade. This relation is especially strong with neighboring countries due to lower distribution costs. Aid may therefore show a positive effect on total trade, while a negative correlation is found for bilateral trade.
If we return to Table 2, we find that AfT is positively correlated with both donor and recipient exports. The no-AfT-dummies clarify this connection by showing negative correlation with exports. The parameter in column (2) imply a great but negative effect (-0.241) on exports for no-AfT-receiving-countries. The parameter suggests that these exports only accounts for approximately 75.5%9 of the exports when AfT in fact is received. This is an interesting observation since it indicates that AfT indeed has an effect on recipient exports to the donor country. This may be due to several reasons, including enhanced trade capacities and the creation of costumer relations. The later assumption is strengthened by a similar negative correlation with Swedish exports, even though it’s smaller and insignificant.
As discussed above, only few of the parameters are significant meaning that just a limited amount of effects on bilateral trade are found. Hence, no conclusions regarding the assumptions and
will be made.
7.3. Regional differences
To check the robustness of my results, I will redo the regression according to my preferred specification (3). The three regions are: (i) Africa, (ii) Asia and (iii) The rest of the world (i.e. all 126 countries excluding countries located in Africa and Asia). The reason for this division is the regional distribution of Swedish ODA where Africa and Asia are the two largest receives in absolute numbers (see figure A2 in appendix).
Studying Table 3 we can once again conclude that GDP is positively correlated with exports. This relation holds for all regions, although it varies in degree. The correlation between population and bilateral trade has on the other hand changed. If we compare the parameters, it becomes clear that effect strongly varies in degree and across regions.
What is interesting to note is that the size of population in Asia is great and negatively correlated with exports. Africa shows however, a great but positive correlation. This differs from the results in Table 1 and Table 2 but consists with the previous presented reasoning. From this we can suspect that Asian countries choose to consume domestic goods and/or trade more with other countries when the population increases. African countries choose, in contrast, to import more goods from Sweden as domestic demand increases.
9 (Exp (-0.241)-1)*100= -21.415837 ≈ 21.5%. Exp (-0.241) = 0.7858416 ≈ 78.5%.
17 Turning to the aid variables, we can even in this case see that neither ODA, AfT nor ODA-net-AfT show significant results. We can on the other hand observe negative and significant correlations between no-ODA-disbursements and exports. This correlation is found for African exports as well as Swedish exports to “The rest of the world”. The parameters are identical and indicate that exports will decrease with 12%10. I.e. these exports will account for 88% of the exports measured when ODA disbursements are made. The no-AfT-dummy show few significant results but demonstrate a negative and significant correlation for exports of countries located in other regions than Africa and Asia. The parameter indicates that export are 32%11 lower for non receiving countries.
Even though these results aren’t consistent for all trading partners and across regions, it initiates that ODA can have a positive effect on trade for both receivers and donors. This is in line with the previous results. Yet, when comparing Table 2 and 3 it becomes clear that the geographical division provides a better understanding of conditions in the different regions.
Table 3 Bilateral trade by region
(1) (2) (3) (4) (5) (6)
African exports to
Sweden
Asian exports to
Sweden
Rest of the world exports
to Sweden
Swedish export to Africa
Swedish exports to
Asia
Swedish exports to the rest of the
world
lnGDP 0.975** 1.586*** 0.521 0.472** 1.318*** 0.975**
(GDP DC) (0.354) (0.274) (0.309) (0.172) (0.184) (0.354)
lnPop 2.822 -1.156* 0.698 3.684*** -1.220** 2.822
(Population DC) (1.846) (0.585) (1.203) (0.893) (0.399) (1.846)
NODA -0.131* 0.0623 0.0708 0.0287 -0.0230 -0.131*
(laggad no-ODA- dummy)
(0.0596) (0.0501) (0.0498) (0.0281) (0.0342) (0.0596)
lnODA_AfT -0.219 -0.193 -0.179 0.0125 -0.0749 -0.219
(lagged ODA- net-AfT)
(0.302) (0.219) (0.222) (0.146) (0.148) (0.302)
lnAfT 0.0189 0.0147 0.0348 0.0490 -0.00828 0.0189
(lagged AfT) (0.0507) (0.0363) (0.0498) (0.0250) (0.0248) (0.0507)
NAfT -0.308 0.0634 -0.400* 0.0591 -0.177 -0.308
(laggad no-AfT- dummy)
(0.210) (0.141) (0.188) (0.104) (0.0965) (0.210)
Constant -14.29*** -11.59*** -4.881 -9.811*** -6.870*** -14.29***
(3.829) (2.678) (3.096) (1.748) (1.803) (3.829)
N 604 444 633 637 463 604
R2 0.747 0.923 0.837 0.882 0.930 0.747
adj. R2 0.717 0.913 0.818 0.869 0.921 0.717
Notes: The dependent variables are log recipient’s exports by region / log Swedish exports by region.
The regression includes year dummies (not shown) Standard errors in parentheses * p < 0.05, ** p < 0.01, ***
p < 0.001
10 (Exp (-0.131)-1)*100= -12.2782 ≈ 12%. Exp (-0.131) = 0.877218 ≈ 88%.
11 (Exp (-0.400)-1)*100= -32.968 ≈ 32 %. Exp (-0.400) = 0.67032 ≈ 67%
18 8. Conclusions
Even though much ODA isn’t designed to boost trade capacities, this article shows that ODA is positively associated with recipient-donor exports as well as donor-recipient exports. The result supports the assumption that aid creates customer relations and distribution channels that can benefit both parties. The effect is however not consistent when AfT flows are excluded from the ODA-variable. From this founding it’s on the other hand plausible to assume that it’s the AfT flows that account for a significant part of the positive correlation with exports. This conclusion is further strengthened by a negative and significant correlation with recipient exports, found when analysing the no- AfT-parameter. The paper also divided the sample into different geographical areas to be able to check the robustness of the results presented above. Few of the new parameters shows significant results but the correlation between ODA and trade seem to be inconsistent both in degree an across regions.
If we summarize this, four important conclusions can be made:
(i) Swedish ODA has a positive effect on Swedish exports to recipient countries (ii) Swedish ODA has a positive effect on recipient exports to Sweden
(iii) No-AfT- disbursements entail a negative effect on recipient exports (iv) The ODA effect differs in degree and across regions
To some degree it is hard to analyze the effects of Swedish ODA on trade when we neither consider ODA flows given by other donors nor exports to other countries. If we also assume that much of the Swedish ODA addresses poverty reduction, it becomes clear that further subdivision are needed to estimate the actual effect of ODA designed to increase trade opportunities. My model entails even further difficulties since it doesn’t include sector specific exports. These factors would be interesting subjects for further research.
19 9. References
Anderson, J. E. and van Wincoop E. (2003), Gravity with Gravitas: A solution to the Border Puzzle. American economic review 93, no. 1: 171–192
Drabek, Z. and Laird, S. (2001), Can Trade Policy Help Mobilize Financial Resources for Economic Development? WTO and UNCTAD, Geneva
Grossman, G. and Helpman, E. (1991), Innovation and Growth in the World Economy. Cambridge: MIT Press.
Martínez-Zarzoso, I., Nowak-Lehmann D., F., Klasen, S. and Larch, M. (2008), Does German Development Aid Promote German exports? Ibero-Amerika Institut für Wirtschaftsforschung Instituto Ibero-Americano de Investigaciones Económicas Ibero- America Institute for Economic Research (IAI) Discussion Papers no 170
McKinley, T. (2005), Why is ‘the Dutch Disease’ Always a Disease? The Macroeconomic Consequences of Scaling up ODA. UNDP Working Paper no 10
Munemo, J., Bandyopaghyay, S. and Basistha, A. (2007), Foreign Aid and Export Performance: A Panel Data Analysis of Developing Countries. Federal Reserve Bank of St. Louis Working Paper 2007-023A.
Osei, R., Morrissey O. and Lloyd T. (2004), The Nature of Aid and Trade Relationships. The European Journal of Development Research 16, no 2: 354–74.
Page, S. (2007), The Potential Impact of the Aid for Trade Initiative. UNCTAD G-24 Discussion Paper Series no 45
Pettersson, J. and Johansson, L. (2011), Aid, Aid for Trade, and Bilateral Trade: An Empirical Study. The Journal of International Trade & Economic Development, oi:10.1080/09638199.2011
Suwa-Eisenmann, A. and Verdier, T. (2007), Aid and Trade. Oxford Review of Economic Policy 23:3, 481-507 15
Tajoli, L. (1999), The Impact of Tied Aid on Trade Flows Between Donor and Recipient Countries. The Journal of International Trade & Economic Development, 8:4, 373-388
Tinberg, J. (1962), Shaping the World Economy. The International Executive, Volume 5, Issue 1.
Wagner, D. (2003), Aid and Trade - an Empirical Study. Journal of the Japanese and International Economies, Volume 17, Issue 2, 153-173
References accessed online
OECD (2008), The Developmental Effectiveness of Untied Aid, (2008-12) http://www.oecd.org/dataoecd/5/22/41537529.pdf, (accessed 2011-10-28)
OECD/DAC (2008), DAC Recommendations on Untying ODA to the Least Developed Countries and Heavily Indebted Poor Countries. (2008-06) http://www.oecd.org/dataoecd/61/43/41707972.pdf, (accessed 2011-10-28)
OECD/DAC (2009): Sweden (2009) DAC Peer Review - Main Findings and Recommendations (2009) http://www.oecd.org/dataoecd/27/37/43278517.pdf, (accessed 2011-11-01)
OECD/WTO (2010), Aid for Trade: Is it working? (2010) http://www.oecd.org/dataoecd/30/36/45581702.pdf, (accessed 2012-01-02)
OECD/WTO (2011), Aid for Trade at a Glance 2011: Showing Results (2011) http://dx.doi.org/10.1787/9789264117471-en, (accessed 2012-01-02)
20 Sida (2012), Our fields of work (2012), http://www.sida.se/English/About-us/our- fields-of-work/, (accessed 2012-01-10)
UN (2010), GOAL 8 Develop a Global Partnership for Development, United Nations Summit (2010-09) http://www.un.org/millenniumgoals/pdf/MDG_FS_8_EN.pdf, (accessed 2011-11-01)
UN (2011), Millennium Development Report 2011, (2011), http://www.un.org/millenniumgoals/11_MDG%20Report_EN.pdf, (accessed 2011-11- 01)
WB (2012), The World Bank group: Millennium development indicators, (2012) http://data.worldbank.org/data-catalog/millennium-development-indicators, (accessed 2012-01-01)
WTO (2011), Doha Round Agenda (2011)
http://www.wto.org/english/tratop_e/dda_e/texts_intro_e.htm, (accessed 2011-09-30) WTO (2012), Aid for Trade and WTO Work Programme (2012)
http://www.wto.org/english/tratop_e/dda_e/background_e.htm (accessed 2012-01-05)
21 Appendix
1. Data description
All values are presented in $US million, constant prices. Exports and GDP are deflated into 2009 $US using a GDP price index of 2009 calculated by the OECD Economics Directorate. Ln signalises that the variables are in their logarithmic values.
When the ODA flows (ODA, AfT and ODA_AfT) are reported to be 0 (e.g. ODA= 0) or when figures are missing, the numbers will be changed into 1 so that its logarithmic value equals to 0. The same procedure will be made to correct for missing export values.
Exports – the figures of Swedish exports as well as the recipient’s exports are downloaded from the UN Comtrade data base on 27 December 2011. These values are reported in current $US million as Swedish total annual exports and imports.
GDP – is the GDP figures downloaded from the World Development Indicator the 11 November 2011 in current $US.
Pop – represents the population in millions in country c. The figures are downloaded from the World Development Indicator the 11 November 2011.
ODA – data over ODA is taken from the OECD Development Assistance committee’s online database International Developments Statistics 26 December 2011.This database in divided into The Development Assistance Committee online database (DAC) that reports aid disbursements and The Creditor Reporting System online database (CRS) that report commitments. The ODA disbursements are located from DAC and the ODA commitments used to proxy AfT flows are located from CRS.
AfT – AfT data is collected from the CRS database. The summarized sectors are:
(i)Building Productive Capacity; Banking and Financial Services (240), Business and Other Services (250), Agriculture (311), Forestry (312), Fishing (313), Industry (321), Mining (322), Construction (223) and Tourism (332)
(ii)Trade-Related Infrastructure; Transport and Storage (210), Communications (220) and Energy (230)
(iii) Building Productive Capacity; Banking and Financial Services (240), Business and Other Services (250), Agriculture (311), Forestry (312), Fishing (313), Industry (321), Mining (322), Construction (223) and Tourism (332).
ODA_AfT – is the ODA-net-AfT [i.e. (total ODA-AfT)]
NODA – takes the value of 1 when ODA is reported as 0 or when values are missing.
NAfT – takes the value of 1 when AfT is reported to be 0 or when values are missing.
22 2. Figures and tables
Figure A1 Swedish ODA-to-GNI ratio (1990-2009)
(Source: WB, 2012)
Figure A2 Swedish Gross Bilateral ODA by region (2008/2009 average)
(Source OECD/DAC, 2009) 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
29%
7%
4% 6%
6% 6%
42%
Sub-Saharan Africa South and Central Asia Other Asia and Oceania Middle East and North Africa Latin America and Caribbean Europé
Unspecified
23
0 5 10 15 20 25
Health Education Research Democratic governance and human rights Conflict, peace and security Humanitarian assistance Infrastructure Trade, industry and financial systems Natural resourses and environment Budget support for poverty reduction Other
Figure A3 Top 10 Recipients of Swedish ODA ($US million, 2008)
(Source: OECD/DAC, 2009)
Figure A4 Sector spending in by SIDA (percentage 2007)
0 20 40 60 80 100 120
Bangladesh Ethiopia Uganda Sudan Congo. Dem. Rep.
Kenya Palestinian Adm. Areas Afghanistan Mozambique Tanzania
