Aid and Regional Trade Integration with a CGE application for Low-Income Countries
Bruno Versailles University of Oxford May 2007
A thesis submitted in partial fulfillment of the requirements for the degree of M.Phil. in Economics
Acknowledgements I wish to thank my supervisor Dr. Christopher Adam for all the hours spent in guiding me and my thesis towards a successful conclusion. Discussing development macroeconomics with one of the best in the field has proven to be both enjoyable and very stimulating. It was exactly what I hoped for when returning to academics. Thank you Chris.
1
Contents
Introduction
7
1 Aid, Dutch Disease, the Transfer Paradox and Regional Integration: Literature Review
10
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
1.1
Aid and Dutch Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
1.1.1
Theoretical Framework
. . . . . . . . . . . . . . . . . . . . . . . .
11
1.1.2
How important are Dutch Disease effects likely to be? . . . . . . .
13
1.1.3
Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
1.1.4
Short-run and Long(er)-run effects . . . . . . . . . . . . . . . . . .
16
1.2
The Transfer Paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18
1.3
Regional Trade Agreements . . . . . . . . . . . . . . . . . . . . . . . . . .
20
2
1.3.1
Some basic theory . . . . . . . . . . . . . . . . . . . . . . . . . . .
21
1.3.2
The ‘New’ Regionalism . . . . . . . . . . . . . . . . . . . . . . . .
21
1.3.3
African RTAs: The record so far . . . . . . . . . . . . . . . . . . .
23
1.3.4
North-South RTAs and Economic Partnership Agreements . . . . .
25
1.3.5
Aid and Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
2 Theoretical Models
29
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
2.1
Aid, Tariffs and the Real Exchange Rate . . . . . . . . . . . . . . . . . . .
30
2.1.1
Model set up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
2.1.2
Comparative statics . . . . . . . . . . . . . . . . . . . . . . . . . .
32
Adding the Regional Dimension . . . . . . . . . . . . . . . . . . . . . . . .
35
2.2.1
Adding a regional good to the model . . . . . . . . . . . . . . . . .
35
2.2.2
Equilibrium in ‘Home’ and ‘Foreign’ . . . . . . . . . . . . . . . . .
36
2.2.3
Regional links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38
2.2.4
Interpreting the model: regional integration and aid . . . . . . . .
39
2.2
3
2.2.5
Towards a CGE modeling strategy . . . . . . . . . . . . . . . . . .
41
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
3 Computable General Equilibrium model: Structural Equations
44
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
3.1
Set up of the CGE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
3.2
Price Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.3
Quantity Equations (including the regional market) . . . . . . . . . . . . .
49
3.4
Income Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
3.5
Expenditure Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
3.6
Market Clearing and Closure . . . . . . . . . . . . . . . . . . . . . . . . .
53
3.7
Real Exchange Rates in the CGE . . . . . . . . . . . . . . . . . . . . . . .
54
3.8
Distributional aspects of the CGE . . . . . . . . . . . . . . . . . . . . . .
55
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
56
4 Calibrating the CGE to Rwanda
57
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
57
4.1
57
Macroeconomic and trade data . . . . . . . . . . . . . . . . . . . . . . . .
4
4.2
Factors of production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
4.3
Elasticities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
62
4.4
Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63
5 Aid and Regional Integration: Simulating Exogenous Shocks
64
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64
5.1
Simulation I: Aid increase fully consumed . . . . . . . . . . . . . . . . . .
65
5.1.1
Simulation II: Regional free trade . . . . . . . . . . . . . . . . . . .
69
5.1.2
Simulation III: Aid increase fully consumed, with regional free trade 71
5.1.3
Simulation IV: Aid to increase sectoral productivity, with regional free trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.4
71
Simulation V: Non-tradable productivity, subsistence consumption and the transfer paradox . . . . . . . . . . . . . . . . . . . . . . . .
73
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
74
Conclusion
75
A Deriving An Expression for φ
77
B List of Variables and Parameters
81
5
B.1 Endogenous Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
81
B.2 Exogenous Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83
B.3 Parameters calculated through Calibration . . . . . . . . . . . . . . . . . .
84
B.4 Other Parameters used in the context of CGE simulations . . . . . . . . .
85
C List of Model Equations
86
C.1 Price Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
86
C.2 Quantity Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
87
C.3 Income Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88
C.4 Expenditure Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
88
C.5 Market Clearing and Closure . . . . . . . . . . . . . . . . . . . . . . . . .
88
6
Introduction
This MPhil thesis aims to link the literature on aid-induced Dutch Disease with regional trade integration in low-income countries. Dutch Disease related economic theory tells us that foreign capital inflows will, ceteris paribus, appreciate a country’s real exchange rate, which might negatively affect a country’s competitiveness. On the distributional front, a country’s sectoral terms of trade are likely to shift as part of the adjustment process, with the possibility of ‘transfer-paradox’ type issues. The literature on regional trade integration typically tells a Vinerian trade-diversion, trade-creation story. These two effects pull in different directions, with no clear-cut welfare results. Viner’s analysis has recently been enriched with a move away from standard neo-classical trade theory to allow for productivity issues such as technology and knowledge transfers and diffusion or dynamic comparative advantage and learning-by-doing issues. Relatedly, there has been an increased focus on behind-the-border issues and the complementarity of political integration to economic integration. The added value of the thesis lies in bringing together these different aspects of a low-income country’s external environment. Ultimately, we are interested in the macroeconomic effects of aid flows in a changing regional trading environment. Does regional integration and the changing trade patterns it causes, mitigate or worsen aid-induced real exchange rate appreciation? What are the effects on incentives for the production and
7
consumption for regional and international markets? Regional integration will most likely have winners and losers, as inter- and intra-country terms-of-trade respond to the newly integrated environment. This raises the possibility of aid being used to compensate losers, with possible trade-offs between more aid and real exchange rate related competitiveness issues. This raises further questions on how extra aid should be used. If it is invested and raises productivity, it might reduce the pressure on the real exchange rate, with a relatively smaller squeeze on exportables. Also, there might be a case for aid to fund regional public goods such as multi-country infrastructure projects, as externalities make it less likely that individual countries are able to fund these. Hence, under the current move away from project to budgetary support amongst donor agencies, it might make sense to give aid to a regional body rather than to individual countries1 . Again, one can then look at the effects of such regional projects on productivity and competitiveness (real exchange rates) in the regional partners. This is a potentially huge research topic. Hence the need to define clearly what this thesis covers and, perhaps as importantly, what it doesn’t cover. I do not touch upon issues such as political economy, dynamic behavior or uncertainty, whilst productivity and welfare are modeled in a very crude way. Also, in the CGE developed in later chapters, the set-up is only for 1 country. This implies that external relations governing regional trade are approached from 1 country’s perspective. The idea is to make a start with the analysis of the research questions listed above, whilst trying to gain some insight into these issues from both a theoretical and empirical point of view. This can then evolve into further work and richer results in a next (post-MPhil) stage. In the conclusion, I go into more detail regarding areas for possible future research. The thesis consists of two big parts. The first part (chapters 1 and 2) looks at theoreti1
One can think of a whole range of political economy issues that would affect such choices. That is not the focus of the thesis however.
8
cal aspects of Dutch Disease and regional integration. The first chapter briefly summarizes the existing literature, whilst chapter 2 sets up a neo-classical model of international trade that has the potential to analyze both Dutch Disease and regional integration issues. However, solving such a model for the variables we are interested in becomes very unwieldy very quickly. Hence, only the model’s set-up is discussed without actually solving it ‘by hand’. To show some results and more intuition for this class of models however, chapter 2 also includes an exposition of the familiar result of a real exchange rate appreciation as a result of an aid increase in the presence of a tariff. The first two chapters do not immediately answer the questions posed earlier on. Hence, the second part of the thesis builds a computable general equilibrium (CGE) model, loosely based on Rwanda. Chapter 3 sets up the foundations of the CGE, explaining its core equations. Chapter 4 calibrates the model’s baseline run, whilst in, chapter 5 we simulate the exogenous shocks we are interested in. These include increasing the aid inflow, decreasing regional trade barriers, looking at the effects of increased productivity through e.g. infrastructure projects funded by aid and subsistence consumption. Distributional issues are rigorously investigated under these different set ups. Increased regional political integration is looked at through increased ease of doing business in the region through changing some crucial elasticities. A short final chapter summarizes the main conclusions and suggests ways to extend the analysis.
9
Chapter 1
Aid, Dutch Disease, the Transfer Paradox and Regional Integration: Literature Review
Introduction
This first chapter introduces Dutch Disease and regional integration issues through a short tour of the relevant literature. The first section deals with Aid and Dutch Disease, a short second section with the transfer paradox and the third section with regional integration. I do not aim to give a comprehensive overview of the empirical literature, but where I do mention empirical results, they focus on sub-Saharan Africa and the member countries of the recently revived East African Community (EAC) in particular1 . For more substantive literature reviews, the reader is referred to Adam [2005] for the first two sections and to 1
Broken up in the 70s, the EAC was revived in 2003-2004 and comprises Kenya, Uganda and Tanzania. In 2005, Rwanda and Burundi were formally accepted as new members.
10
World Bank [2004] and Yang and Gupta [2005] for the third section.
1.1
Aid and Dutch Disease
In this section, I develop the standard aid-induced Dutch Disease argument and discuss its empirical relevance. We start by laying out the theoretical framework, after which we discuss how important Dutch Disease effects are likely to be. Thirdly, we look at the empirical literature, before closing off with a look at effects of aid in dynamic settings. Throughout the thesis, the real exchange rate is defined in the Salter-Swan sense as the ratio of the price of non-tradables to the price of tradables ( PPNT )2 . An increase in this ratio indicates the production of non-tradables becoming more profitable, and consumption more expensive, and is equivalent to an RER appreciation.
1.1.1
Theoretical Framework
Dutch Disease theory for a small, open economy (SOE) are based on the Salter-Swan model in a neo-classical framework of perfect competition and full markets3 . This set-up is used to analyze the effects of an (exogenous) increase in foreign exchange availability like aid on an economy’s structure. To equilibrate the balance of payments (BoP) when foreign exchange availability increases, and with PT fixed in world markets, PN and/or imports have to increase4 . The increase in PN will cause a real exchange rate appreciation, with negative incentive effects for the production of tradables (exports and import substitutes). 2
Alternatively, one can define the Purchasing Power Parity (PPP) real exchange rate as the nominal exchange rate multiplied by the ratio of an aggregate index of world prices to an index of domestic prices. For a lucid exposition of how these two RER concepts relate to each other, see Devarajan et al. [1993]. 3 The main theoretical contributions are Corden and Neary [1982] and Corden [1984] or van Wijnbergen [1986] in the context of the capital inflow taking the form of aid to Africa. 4 Note that for the moment, we are not considering bringing idle production factors into the production process as we are assuming no unemployment of production factors, i.e. the economy operates on its Production Possibility Frontier.
11
Production factors will be reallocated away from the tradables sector and towards the non-tradables sector. On the consumption side, the rise in PN incites economic agents to shift consumption from non-tradables to imports. This is a standard description of the forces leading an economy back to equilibrium after an exogenous shock, and as such no (government or other) intervention is called for. However, there are reasons why countries try to avoid an appreciating real exchange rate and a contracting exportables sector. Firstly, the relative development of tradables’ vs. non-tradables’ sectors is often considered not to be neutral to a country’s overall growth prospects. This non-neutrality typically takes the form of a positive externality, e.g. learning-by-doing5 , in the nontraditional exports sector6 . Exports in general are considered a likely source of productivity growth as, faced with competition in world markets, they face relatively more pressure to improve their performance7 . An appreciating real exchange rate will in general hurt the prospects of this sector and, through the above channels, also the economy’s overall growth outlook8 . As Van Wijnbergen puts it: ‘After all, the post World War II success stories in economic development involve without exception countries that promoted their traded goods sector aggressively 9 ’. Secondly, a country will ultimately have to be weaned off international assistance. Assuming some habit formation in consumption of imports and/or the unavailability of 5
Torvik [2001] however builds a model in which there is learning-by-doing in both the tradables and the non-tradables’ sectors, with the possibility of productivity spill-overs between sectors. He shows that the real exchange rate does not need to appreciate in such a framework in response to a foreign exchange inflow. 6 Non-traditional export sectors, particularly manufacturing are considered to have larger productivity pay-offs, with more traditional commodity exports thought of as problematic to fuel a country’s long-run growth in the face of their steadily declining terms-of-trade. 7 An alternative to the learning-by-doing channel is explored in the model by Melitz [2003]. Here an industry’s exposure to trade leads to inter-firm reallocations towards more productive firms. The model then goes on to show how the aggregate industry productivity growth generated by the reallocations contributes to a welfare gain. 8 Hence the ‘Disease’-tag. The term ‘Dutch Disease’ was first used to describe the adverse impact on the Dutch manufacturing sector of the discovery of natural gas in the North Sea in the 1960s. This happened through the windfall natural gas revenue appreciating the Dutch Real Exchange Rate. 9 van Wijnbergen [1984].
12
domestic substitutes for inputs in the production process, sudden reversals in aid flows can be costly. Hence the need for an alternative source of foreign exchange to pay the import bill and the importance of gradually increasing export receipts in the face of possibly declining future aid flows. Thirdly, the above analysis assumes aid is an integral long-run part of a country’s long-run equilibrium path. Once the possibility is raised that aid is only going to be a temporary phenomenon, it might not be optimal to let the economy adjust as suggested in the above analysis. We will say more about intertemporal issues in section (1.1.4) below. As an aside, and making the link to section (1.3) on regional integration, a case could be made that for an average sub-Saharan African low-income country, exports to the rest of the world might well be more important than regional exports for the described competitiveness/productivity effects to work. One reason is that regional trade is often relatively more concentrated in basic goods, which don’t need a high skill level in production and which often have low income elasticities (e.g. like foodstuffs). Providing goods like cash crops or manufacturing for the international market might thus well give more scope for the dynamic externalities mentioned previously to materialize.
1.1.2
How important are Dutch Disease effects likely to be?
Continuing our theoretical overview, what are the parameters that influence how much the real exchange rate will appreciate in the face of increased aid inflows? Firstly, it is important to know what the extra aid will be spent on. On imports or on non-tradables? In general, aid is channeled through government, which is often thought of as having a higher propensity to spend on non-tradables than the private sector. Secondly, a quick supply-side response would ease the pressure on the real exchange rate. Such a response might have a higher probability in low-income countries, e.g. in post-conflict areas. If 13
‘..countries can call on idle productive capacity to incite a quick supply-side response with aid being used productively to ease supply bottlenecks’ 10 , the appreciation will be muted. A third issue is that the real exchange rate appreciation will be less severe the higher domestic agents’ substitution possibilities between importables and non-tradables. Fourthly, Nkusu [2004a] argues there might be imperfect substitutability between domestically produced manufactured goods and imported ones, i.e. domestic producers of import substitutes have some market power and are able to raise prices to some extent without consumers switching towards imports. Hence PT rises alongside PN which again reduces the appreciating pressure on the RER.
1.1.3
Empirical Results
Empirically, we are looking for a chain of causality from aid to a real exchange rate appreciation, to decreased exports and finally to decreased GDP growth. The econometric evidence on the first link, from aid to RER appreciation, is mostly done in reduced-form single equation regressions. In this research, heavily influenced by the theoretical work of Sebastian Edwards11 , aid is typically one of the right-hand side variables trying to explain the behavior of the equilibrium real exchange rate. An overview of this empirical literature can be found in Nkusu [2004a] or Bourguignon et al. [2005]. Results are ambiguous, with some papers even reporting RER depreciating effects in the face of an aid increase. One conclusion that seems quite robust however is that aid has a rather small quantitative impact on the real exchange rate compared to other determinants like trade policy or the terms of trade. The main problem with this type of research is that an aid-inflow which is at least partly spent on non-tradables will always appreciate the real exchange rate more relative 10 11
Bourguignon et al. [2005] See Edwards [1989a] or Edwards [1989b].
14
to a situation in which no aid was given. Obviously, this is a hard counterfactual to work with. Furthermore, the forces mentioned in the previous section also fall under this category, i.e. they can ease RER pressure, but will not reverse it. There is some evidence for LICs however that especially the second issue of supply-side response might well be more valid than is generally thought. Nkusu [2004b] documents this for Uganda, whilst Bourguignon et al. [2005] show some tentative evidence for post-conflict countries, notably Rwanda and again Uganda for the second half of the 1990s. There is little evidence on issues three and four, but these might well play a role in regional trade. It would be reasonable to argue that regionally tradables are more easily substitutable with domestic goods than with internationally tradables. Equally, domestic producers might have some market power over regionally tradables. These issues will be central to the development of the next chapters. Empirical results are somewhat clearer for the relation between the RER and exports. Research applied to African countries show clear incentive effects of real exchange rate movements on export performance12 . It is often the misalignment of the real exchange rate to its equilibrium value that gives the wrong signals to exporters13 . However, recent research also indicates that the real exchange rate, whilst often crucial, might well not be the binding constraint for exporters. Castel-Branco and Sulemane [2005] for example argues that in Mozambique, non-price factors such as transport and storage facilities might well be more important in limiting the expansion of the exportables sector. Finally, what does the empirical literature say about the link between exports and growth? The experiences of the East-Asian tigers and China has convinced most commentators of the benefits of an outward-led development path with active export promotion14 . In sub-Saharan Africa similarly, there is mounting evidence that productivity 12
See for example Bigsten et al. [1999] See Elbadawi [1999] or Sekkat and Varoudakis [2000]. 14 See for example World Bank [1993] or Kraay [1999]. 13
15
growth is higher in the exports sector and that learning-by-doing is responsible for this [Bigsten et al., 2004], and that export performance is positively linked to a country’s overal growth experience [Ndulu and Ndungu’u, 1997].
1.1.4
Short-run and Long(er)-run effects
A first intertemporal issue complicating the above analysis is whether the aid inflow comes in the form of a permanent or a temporary shock. Related to this is whether the interaction of supply and demand causes the real exchange rate to overshoot from its longrun equilibrium path. In most cases it is hard to disentangle these effects, and as such it is difficult to see whether the Government should try and take corrective action15 . This can have quite important effects on exporters, as the lack of depth in financial markets in developing countries means that many of them might find it hard to smooth over a dip in earnings. This kind of hysteresis effect might thus mean that temporary aid has long-run consequences for a country’s economic structure. So far, we have not really touched on the medium- and long-run effects aid could have on a country’s economy. The crucial question, reflected in the title of a recent paper16 is the effect of aid on the supply-side. The paper claims that ‘..supply side constraints are usually dominant and (..) these are often deep-rooted, unamenable to short-term solutions’. However, over time, at least some of the aid will be invested to help increase the receiving economy’s productivity, for example through funding public goods like infrastructure. The question then becomes how long this takes and how this productivity shock will be transmitted to tradable and non-tradable sectors. 15
This situation is not helped by the current aid environment in which aid flows have been known to be quite unpredictable. Aid pledges, for example, are a very bad predictor for actual aid disbursements. Further Bulir and Hamann [2005] show evidence that aid is on average pro-cyclical and more volatile than other sources of Government revenue. 16 ‘The Macroeconomics of Doubling Aid To Africa and the Centrality of the Supply Side’ [Killick and Foster, 2007].
16
There is a new urgency about this debate as, in the wake of some influential reports like the report of the Commission for Africa or the Millennium Project17 calling for more aid, the G8 promised at its 2005 summit in Gleneagles to double aid to Africa by 2010. This vast increase in foreign assistance might well mean that the future aid environment will be quite different from the historical patterns most of the empirical literature is based on18 . In the words of Adam [2005]: ‘..the evidence on the short- and medium-run macroeconomic effects of aid is still partial, often contradictory, generally ambiguous, and arguably of only limited relevance to the contemporary policy debate on aid management’. On the back of these political evolutions, research has moved into two new directions. Firstly, detailed country case studies of recent aid surges have been conducted. IMF [2005] and Killick and Foster [2007] look at 7 sub-Saharan African countries who have seen aid flows increase substantially at some point during the past decade. These studies found few instances of clear-cut aid-induced Dutch Disease. Ethiopia and Ghana chose to build their foreign exchange reserves, effectively saving the extra aid, whilst Uganda sterilized the foreign exchange inflow to try and keep their RER at a competitive level. Only Tanzania and Mozambique did seem to experience some Dutch Disease symptoms at the end of the nineties. Secondly, there are increasingly sophisticated efforts at analyzing the effects of increased aid flows in Computable General Equilibrium frameworks, e.g. Adam and Bevan [2006] or the World Bank’s work on MDG funding scenario’s [Sundberg et al., 2005]. The former paper builds a CGE for a fairly standard African post-stabilization country. The lessons from Adam and Bevan [2006] are well summarized by Adam [2005]:
‘First, when public infrastructure augments the productivity of private 17
See Commission for Africa [2005] and UN Millennium Project [2005] respectively. On current trends, it seems very unlikely the doubling will be achieved as aid to sub-Saharan Africa barely increased between 2005 and 2007 [OECD, 2007]. This does not mean however, that substantial increases might not materialize for individual countries. 18
17
factors, (..) there are potentially large medium-term welfare gains from aidfunded increases in public investment, despite the presence of some short-run Dutch disease effects, and (..) these are compatible with growth in the export sector of the economy. Second, however, when supply side responses to aid are important, real exchange rate overshooting may be an importance feature of the economys response to aid inflows. Third, (..) the actual evolution of the economy will depend crucially on the form of public investment, how powerfully (and how quickly) it feeds back onto private production capabilities, and the costs of any short-run contraction of the export sector. Importantly, though, export promotion and growth may be benefited as much, if not more, by public investment geared towards improving the productivity of domestic non-tradable goods production rather than directly towards improving productivity in the export sector itself.’
Concluding, it should be remembered there is no magical formula to avoid pressure on the real exchange rate in the face of increased aid inflows. The basic macroeconomic policy choice is whether to save the aid, let the RER appreciate or to sterilize with likely crowding-out effects for credit to the private sector. As van Wijnbergen and Edwards [1989] put it: ‘the conflict between substantial volumes of aid and export promotion is a dilemma policy-makers will need to face’.
1.2
The Transfer Paradox
Aid flows can generate what has become known in the international economics literature as the ‘transfer paradox’. The ‘transfer paradox’ was first mentioned in a debate on the endogeneity of the terms-of-trade between economists John Maynard Keynes and
18
Bertil Ohlin in the context of war reparation payments from Germany to France and England after World War I. It refers to a situation in which an unrequited transfer to a country might leave that country worse off. Samuelson [1952] shows that, in a standard two country trade model and given the stability of a Walrasian equilibrium, the transfer paradox cannot occur. However, when allowing for a multi-country setting19 and/or a distortionary trade environment20 , the paradox becomes a possibility as a recipient’s worsening terms-of-trade might overturn the positive income effect. More recently, the paradox was researched in the context of a small country, i.e. a setting where international prices are given and the terms of trade are thus independent of transfers. Chichilnisky [1980] looks at aid in a setting with a rich and a poor group in a high-income country and a (poor) group in a low-income country. She concludes that under certain circumstances an increase in aid will increase global inequality21 . More recently Yano and Nugent [1999] link the transfer paradox to the non-traded goods sector. They claim to ‘demonstrate that the expansion of the non-traded goods sector can change the domestic price of the non-traded good in such a way that the otherwise beneficial effect of aid may be offset’. This is kind of the opposite of the Dutch Disease effect which, as we have shown, emphasizes an increase in the demand for the non-traded good. Their model however, was found to be incorrect as pointed out by Tokarick [2006]. The empirics of the Yano-Nugent paper are not entirely convincing either. In their sample of 44 aid receiving countries, they find only six countries where aid flows are statistically and positively associated with the symptoms of a transfer paradox (an expansion of the non-tradable sector, contraction of the tradable sector and a decline in real GDP). The merit of these papers however lies in their highlighting of the distributional ten19
See for example Bhagwati et al. [1983]. See for example Johnson [1967]. 21 Her model and results were however shown to be slightly incorrect and quite contrived. See the discussion in the Journal of Development Economics volume 13, 1983. 20
19
sions of an economy adjusting to a new aid reality. Even though it is perhaps unlikely that welfare will go down on aggregate, when distributional effects are non-neutral, the ways and mechanisms in which ‘the spoils are divided’ becomes important. Adam and Bevan [2006] show for example that, when rural-urban migration is limited, farmers, which often face a low income elasticity of demand for their output, might well loose out in real terms in an economy adjusting to a positive aid shock22 . We will return to this type of analysis in the CGE model of chapters 3, 4 and 5. In the context of an aid-receiving regional grouping of countries that reduces internal barriers to trade, there might be similar inter-country and intra-country (i.e. inter-sectoral) distributional effects. In a way, we will be marrying the the two possibilities mentioned above to revive the transfer paradox: (i) a multi-country setting, (ii) a tariff-distorted environment.
1.3
Regional Trade Agreements
For a short overview of the literature on Regional Trade agreements (RTAs), we adopt a similar strategy as in section 1.1. This means we start with some theory (first two sections), move on to empirical results, discuss the relationship between aid and trade and end with a section covering aid and trade links. For more in-depth literature reviews see Baldwin and Venables [1995], Burfisher et al. [2004], Panagariya [2000], or Yang and Gupta [2005] for a focus on Africa. 22
This is related to the analysis of Matsuyama [1992]. He shows that for a SOE operating in an environment where the income elasticity of demand for agricultural goods is less than unity and with a learning-by-doing externality in the manufacturing sector, there will be a negative link between agricultural productivity and growth. This is in contrast to the widely held belief in development economics that an increase in agricultural productivity is necessary for an ‘industrialization’ phase of development to take off and for sustained longer run growth. See for example Rostow [1960] or Timmer [1988] for a more recent reference.
20
1.3.1
Some basic theory
Let us first of all clearly define RTAs. Baldwin and Venables [1995] distinguish between three types of ‘geographically discriminatory trade policy’. A Customs Unions (CU) is a trade configuration whereby all partner countries agree to reduce internal tariffs, with a common external tariff governing trade with third countries. A Free Trade Area (FTA) by contrast leaves every partner country to set its own tariff vis-`a-vis the rest of the world. A common market adds the free movement of factors of production to the free movement of goods and services between member states. The analysis in the following chapters is solely concerned with Customs Unions. Analysis of the welfare implications of regional trade agreements goes back to Viner [1950]. He used the concepts of trade creation and trade diversion to show that for small open economies, welfare effects of regional trade agreements are ambiguous23 . To understand Viner’s key result, consider a country entering an RTA, which aims to reduce intra-regional import tariffs. Trade creation is then the displacement of inefficient domestic production by imports from more efficient regional partners. Trade diversion is the displacement of imports from third countries by less efficient (relative to the thirdcountries) regional partners. It should be obvious that the higher the tariffs with respect to goods from third countries, the higher the probability that trade diversion will be significant.
1.3.2
The ‘New’ Regionalism
As early as 1985, Hamilton and Whalley [1985] argued that trade creation and trade diversion were not the overriding issues in regional trade integration, implying the Viner 23
When moving away from the SOE assumptions, one should also allow for terms-of-trade welfare effects
21
analysis must somehow be inadequate. As most economists in the 1950s, Viner worked within the Heckscher-Ohlin-Samuelson (HOS) trade modeling tradition. With respect to regional trade, the HOS mainly busied itself with investigating the static allocation effects of removing commodity trade barriers. Burfisher et al. [2004] dubs the standard HOS regional integration analysis the ‘old’ regionalism, claiming this type of research is ‘missing much of the action’. We can think of three main departures from the framework set out above to capture the notion of this ‘new’ regionalism. Firstly, we can relax the standard assumptions of the HOS-model, like homogenous goods, perfect competition or constant returns. Hence, researchers have built models to try and capture issues like differentiated products, market structure, international factor mobility, transport costs, scale economies and imperfect competition to analyze regional integration issues. Secondly, and closely related, is the need to incorporate dynamics explicitly into modeling endeavors. According to World Bank [2004] ‘a full welfare analysis of RTAs should also take into account potential dynamic gains, with Burfisher et al. [2004] arguing dynamic should include analysis of trade-productivity links and endogenous growth theory according
24 .
This would include technology and knowledge transfer, and technology
diffusion, especially from developed to developing countries, that increase productivity. The new trade analytics also encompass dynamic comparative advantage and learningby-doing efficiency gains through increased demand from expanded trade. The previous arguments can still mostly be linked to (regional) commodity trade. However, the ‘deeper’ integration Burfisher et al. [2004] talk about goes further than just commodity trade. Their analysis also incorporate services trade and moves towards political integration alongside economic integration. In the words of World Bank [2004]:‘RTAs 24 One should bear in mind the last paragraph of 1.1.1 which argues that for low-income countries positive externalities from increased regional market exposure might well be relatively low.
22
are often one component of a larger political effort to deepen economic relations with neighboring countries, trying to overcome institutional as well as policy barriers to trade. Trade agreements usually therefore go beyond slashing tariffs to include measures to reduce trade impediments associated with standards, customs and border crossings, and services regulations–as well as broader rules that improve the overall investment climate.’. These are the so-called ‘behind the border’ issues. This also brings issues like rent-seeking behavior and other political economy considerations, like the potential conflict between regionalism and multilateralism, to the fore. The issue at stake here is whether RTAs are, in the words of Jagdish Bhagwati, a ‘stepping stone’ or a ‘stumbling block’ towards multilateral free trade [Bhagwati, 1991]25 . However, even when including these post-HOS ‘new’ regionalism aspects of regional trade environments, [Burfisher et al., 2004]’s conclusion that ‘it is impossible to draw strong general conclusions about the desirability of forming an RTA from the perspective either of members or of the rest of the world, from theory alone’ still stands.
1.3.3
African RTAs: The record so far
According to World Bank [2004], every African country is member to on average 4 RTAs, often with overlapping membership. Hence the term ‘spaghetti bowl’. Yang and Gupta [2005] find that in early 2005, there were some 30 RTAs in Africa, ‘with a renewed push in recent years to broader and deeper preferential trade arrangements’. African RTAs have largely been motivated by the continent’s desire to promote growth, increase bargaining power in multilateral negotiations and reduce conflict through regional cooperation. 25
For a nice overview of the literature on this important literature, see chapter 6 of World Bank [2004]. Other key sources are Limao [2005], Saggi [2004] or Winters [2000]
23
What is the success rate of African RTAs as measured against these stated objectives? Yang and Gupta [2005] have a good discussion on this, their main points are:
• Trade intensity is considerably higher among African countries than between African countries and non-African countries. This is however, largely, due to Africa’s marginalization in the world market rather than to the performance of intraregional trade. • Econometric evidence on RTAs promoting intra-African trade is mixed. One might tentatively argue that RTAs have had positive but uneven effects on intra- regional trade. Over the longer run though, effects seem to have been small or insignificant26 . • RTAs do not seem to have had a significant impact on Africa’s export performance in the world market, with little indication that Africa’s export competitiveness has a positive relation with the presence of an RTA. • Africa’s poor record in attracting FDI also seems to indicate that RTAs have not significantly improved the region’s competitiveness. • Considering the the failure of African RTAs to increase the region’s overall trade, it is unlikely that dynamic welfare effects (e.g. from increased competition or learningby-doing) have been very significant.
On the non-economic objectives, it seems that RTAs have had some success in reducing conflict on the continent, e.g. ECOWAS in Sierra Leone. Such evidence is anecdotal however and should thus not be exaggerated. RTAs can even be bad for regional cooperation as the collapse of the East African Community in the 1970s shows27 . Finally, 26
See Elbadawi [1997] for a skeptical assessment or Carrere [2004] for more positive evidence. Trade integration among EAC members (Tanzania, Uganda and Kenya), led to manufacturing activities being concentrated in Kenya, upon which Uganda and Tanzania withdrew. The EAC has recently been revived, and now also include Rwanda and Burundi. This also shows that even though regional aggregate welfare might go up, some countries might loose out. Hence the need to set up regional institutions to handle redistribution. 27
24
on the ‘increase in negotiating power’ objective, evidence is again mixed, with African countries for example not always having a common negotiating position in multilateral talks. Overall, it seems that one of the main problems plaguing African RTAs has been poor implementation records. With low administrative capacity in most SSA countries, the current ‘spaghetti bowl’ combined with the increasing complexity of multilateral negotiations does not bode well for the future.
1.3.4
North-South RTAs and Economic Partnership Agreements
What about regional trade agreements between African countries and countries or groups of countries from outside Africa? Welfare results for such RTAs could well be more benign compared to the intra-African RTAs we just reviewed. Venables [2003] develops a model based on comparative advantage, which shows that if comparative advantage is associated with per capita income, CU membership leads to convergence of income levels within a union composed of high income countries, and divergence in a union composed of low income members. World Bank [2004] argues that North-South RTAs with small developing countries yield the biggest benefits for the latter: ‘When one partner is much smaller than the other, its participation has virtually no impact on prices, and any incremental exports it sells are a tiny share of the large country’s trade flows–so the resulting gains or losses for the larger partner are also small’. However, it has also been argued that the recent aggressive pursuit of RTAs by the EU and the US are not always benefiting the low-income RTA partner(s). World Bank [2004] is quite critical of these socalled ‘hub-and-spoke’ agreements. It claims the ‘hub’ benefits disproportionately because of differing rules of origin, product exclusions, nontrade issues, etc, reflecting the overall unequal negotiating clout in an agreement bringing together one of the world trading powers with (a) LIC(s).
25
The EU has a long history of trade agreements with the African continent. Since 1975, external relations between ACP countries and the EU, in which trade figures prominently, have been governed by the subsequent Lom´e conventions, and, since 2000, the Cotonou agreement. However, the trade preferences extended by the EU under these agreements, are to be phased out in the next couple of years, with new so-called Economic Partnership Agreements (EPAs) to be negotiated between the EC and 4 different African regions28 . All EAC countries but Tanzania are in the East and Southern Africa regional grouping, with Tanzania in the SADC-minus grouping. The EU also gives the group of LeastDeveloped Countries (LDCs, a group of countries defined by the UN) duty and quota-free access to EU market for all their exports. This preference scheme, called the Everything But Arms (EBA) initiative, will keep on running regardless of the EPA process, but is in a way less attractive to LDCs as it is not contractual and thus less predictable. All members of the EAC, except Kenya, are LDCs. It should be clear that there are a huge number of possible trade configurations that need to be kept track of in analyzing these issues. Cali and te Velde [2006] review the evidence on the expected welfare effects of EPAs, concluding that ‘[under EPAs] trade liberalization of goods trade is broadly positive though not as large as could be expected from multilateral trade’. For the effects of the EPA process for the EAC specifically, Milner et al. [2005] find that welfare effects will be small, whether positive or negative, but countries will experience short-run adjustment costs, especially in the form of revenue losses. 28
See Stevens [2006]. In general under article XXIV of the GATT/WTO, RTAs are ‘compatible’ with rules governing multilateral trade as long as they don’t violate one of the two following conditions: (i) they fail to eliminate barriers on ‘substantially all the trade’ among members, and (ii) external tariffs imposed ‘shall not on the whole be higher or more restrictive’ compared to the pre-RTA situation. Hence under the EPAs, trade relations will have to be guided by the reciprocality principle.
26
1.3.5
Aid and Trade
This brings us to the relationship between aid and trade. As LICs often have tax structures that are biased towards trade taxes
29 ,
trade liberalization often has negative effects on
domestic revenues. Even though the evidence is not entirely conclusive, it does seem that such losses could be non-trivial especially in LICs where structural features might impede governments from raising taxes elsewhere30 . For the EAC, Castro et al. [2004] estimate that under the initial 2004 set-up where the EAC comprised only Kenya, Uganda and Tanzania, revenue losses would be in the region of 10 percent of pre-CU customs revenue. A possible use for aid funds could well be to help liberalizing countries to smooth out these revenue losses, making the adjustment process easier to bear31 . The EU for example, in its regional programmes often has budget support agreements to deliver exactly this type of aid. Lauren Philips and Page [2005] distinguish between short-term and long-term aid for trade. The former would include grants to meet costs directly linked to liberalization negotiations (losses due to preference erosion, terms of trade losses, compliance with regulation and procedures, etc), whilst the latter refers to larger ongoing support to address infrastructure and supply-side constraints. Of course this brings us back to the discussion of section 2.1 on Dutch Disease. On top of the issue of whether aid should be spent on tradables or non-tradables, consumption or investment, we now add the regional dimension. I.e. is there a case to be made to spend aid at the regional level? Birdsall [2004] makes a passionate case for aid agencies to guide their funding more towards regional public goods. With the amount of small countries in SSA, there will definitely be externalities from regional infrastructure projects. For landlocked countries, there may even be externalities from a neighboring country investing in infrastructure 29
See for example World Bank and IMF [2004]. See Baunsgaard and Keen [2005] and Khatty and Rao [2002] for evidence on those structural impediments, and Agbeyegbe et al. [2006] for a more skeptical view on the links between liberalization and revenue. 31 This could also come under the guise of explicit or implicit conditionality. 30
27
[Yang and Gupta, 2005].
Conclusion
With the literature review in this first chapter we have set the stage for the following chapters. In turn, research on aid and Dutch Disease, the transfer paradox and regional integration were discussed, with the last section coming full circle to make the link with aid again. Despite the criticism and the move away from the neo-classical tradition in recent research covering these topics, the thesis will mainly stay within the confines of that tradition. However in the main conclusion to the thesis, I identify some interesting future research possibilities linked to issues such as dynamics, unemployment of production factors or ‘second generation’ RTAs.
28
Chapter 2
Theoretical Models
Introduction
We now proceed to set up a framework integrating the literatures on Dutch Disease and regional integration. The theoretical set-up is the standard neo-classical trade theory model. Section 1 shows how aid and trade policy effects can be analyzed in a 3-good model set up, demonstrating the well-known effects of an increase in aid on the real exchange rate in the presence of a tariff. Section 2 adds a regional dimension, increasing the complexity of the model substantially. Quantities and prices for a newly introduced regional good are endogenously determined, which means we move partially away from the standard small open economy assumption of all tradable prices given exogenously. As complexity increases, the model also become harder to solve ‘by hand’, and the exact adjustment mechanisms more difficult to interpret. The model is hence discussed in a more intuitive way, focusing on what we expect a move towards freer trade of a regional good can add to the framework set out in section 1. This includes discussion of effects on (i) real exchange rates, (ii) regional integration, (iii) welfare, including transfer paradox29
type issues and (iv) the regional market. Section 2 concludes by making the case for moving towards a computable general equilibrium model. We shortly discuss the added value of a CGE and make the bridge to the second, CGE part of the thesis.
2.1
2.1.1
Aid, Tariffs and the Real Exchange Rate
Model set up
In this chapter we use the standard neo-classical international trade model. This type of model is superbly covered in Dixit and Norman [1980] using a duality framework. That is also our approach here. The Ricardo-Viner set-up is used throughout, which means that we are looking at short to medium-run scenario’s in which labor is mobile across sectors, but not across countries, with capital sector-specific and fixed. This section shows the effects of tariffs and aid on aggregate welfare and the real exchange rate. Our low-income country consumes a non-tradable good and a good imported from the rest of the world. It produces a non-tradable and a good exported to the rest of the world. We thus have a 3-good economy, with an ad valorem tariff ‘t’ on imports and no distortions in the exports sector. Hence, the world price of these exports PX∗ is equal to the domestic price1 of exports PX . We set this price to one and use it as the num´eraire in the completely real model2 . We can now define the following prices: PX = PX∗ = 1 ∗ PM = PM (1 + t) 1
PN PX∗ P∗ T = M PX∗
Q=
(2.1)
A price with a star superscript indicates a world price. As this is effectively the set up of a general equilibrium model with only real effects, only relative prices matter. Hence we can choose one of the prices as the num´eraire. 2
30
Q is the Exportable Real Exchange Rate (ERER) and T is the inverse of the international terms of trade. Duality properties allow us to depict the economy-wide budget constraint through the use of Revenue R(.) and Expenditure E(.) functions. First derivatives with respect to prices of these functions give us supply and demand quantities respectively3 . ‘U ’ is aggregate utility and ‘L’ is the labor supply. ∗ ∗ E(PN , PM ; U ) = R(PN , PX ; L) + tPM E M + PM A
∗ E is the revenue accruing to the tariff collecting body4 , which we assume rebates tPM M
the full value of this revenue in a lump-sum fashion to households. This is a standard assumption in the literature. In other words, we will not be investigating the behavior of a government-type economic agent5 . Following our exposition on Dutch Disease in the previous chapter, it should be clear that Aid ‘A’ fills the balance of payments deficit. As we are assuming that imports bought with aid also face a tariff, the aid inflow is priced at the world price of importables. Dividing through by the num´eraire PX∗ = PX = 1, we can write:
E(Q, T (1 + t); U ) = R(Q, 1; L) + tT EM (Q, T (1 + t); U ) + T A
(2.2)
We further need to establish equilibrium in the tradables and non-tradables markets. For non-tradables, this boils down to:
EQ (Q, T (1 + t); U ) = RQ (Q, 1; L) 3
(2.3)
∂E Subscripts to functions indicate derivatives. Thus for example EPN = ∂P . N We know that total revenue in the economy must equal total expenditure at world market prices. However, when writing domestic prices as the arguments of the expenditure and revenue functions, we have to take care to add taxes (including tariffs) and deduct subsidies. This is why we are adding the terms on the right hand side involving the tariff revenue. 5 Edwards [1989c] correctly points this out as not being a very good way of depicting reality, but then continues using the same assumption nevertheless. 4
31
We get the equilibrium for the tradables good, by deducting the non-tradables equilibrium from the economy-wide equilibrium:
T E M = RX + T A
2.1.2
(2.4)
Comparative statics
We are ultimately interested in welfare and the real exchange rate, and by totally differentiating equations (2.2) and (2.3), we obtain two equations in dU and dQ which are our endogenous variables as functions of the exogenous variables dA, dt and dT.
(1 + t)EM dT + T EM dt + EU dU = h
i
tT EM Q dQ + tT EM U dU + T EM + tT EM M dt + h
i
(2.5)
EQQ dQ + EQM T dt + EQM (1 + t)dT + EQU dU = RQQ dQ
(2.6)
t EM + (1 + t)T EM M dT + T dA + AdT
Isolating dQ from equation (2.6), and plugging it into equation (2.5), we get: EQU )dU − T tEM U dU = EQQ − RQQ 2 h i tT 2 EM Q T dA + tT 2 EM M + ( ) dt + RQQ − EQQ 2 h i t(1 + t)T EM Q − EM + t(1 + t)T EM M + A + ( ) dT RQQ − EQQ
EU dU + T tEM Q (
Next, define, the ‘Marshallian’ income elasticities of demand for non-tradables and imports as DQ,y =
EQU EU ,
respectively DM,y =
EM U EU .
Both of these will be positive as
we are assuming both non-tradables and importables are normal goods. Also, define
32
Φ = RQQ − EQQ . Φ captures the slopes of economy-wide supply and demand, and will be positive in the standard case of upward sloping supply and downward sloping demand curves. Under normal circumstances, imports and non-tradables are substitutes, which implies a price increase for one of the two goods will drive up the demand for the other one (EQM > 0). Finally, as we are mainly interested in what effect aid and tariffs have on welfare and the RER, we suppress the terms of trade term effect (dT = 0). We then get an expression for dU: h
1−
2 i h i EM tT EM Q Q DQ,y − tT DM,y EU dU = T dA + tT 2 (EM M + ) dt Φ Φ
h
Defining γ = 1 −
tT EM Q DQ,y Φ
i
− tT DM,y , which is the term in square brackets on the
left-hand side of the previous expression and re-arranging, we get: 2 h i EM 1 Q T dA + tT 2 (EM M + ) dt dU = γEU Φ
"
#
(2.7)
As γ > 0 in stable models6 , we get our standard welfare results: (i) welfare unambiguously increases with an increase in foreign aid, (ii) a trade tax in the form of a tariff will normally decrease welfare. Indeed, the first ‘dt’ term is negative (as EM M < 0 for a negatively sloping demand curve) showing the distortionary effect of a tariff. This term gives us a deadweight-loss-triangle for consumers only, as producers are not producing importables in this stylized economy. Logically in a price-taking economy, the more price-elastic demand for imports (i.e. the higher the absolute value of EM M ), the greater is the distortionary impact of the tariff and the more welfare goes down. This negative relationship between the trade tax and welfare is muted through a term involving the cross-price elasticity between the demand for the two consumed goods 6
See for example similar expressions in Tokarick [2006] and Edwards and van Wijnbergen [1987], or Deaton and Muellbauer [1980] for a more in-depth discussion.
33
(remember that Φ > 0). In a world without non-tradables this term would be zero. However, in the presence of non-tradables, and assuming some degree of substitutability in consumption, consumers can decide to move demand away from importables towards non-tradables if the tariff becomes too high. The tariff will not have any effect on welfare if EM M +
2 EM Q Φ
2 . So if there is perfect = 0, or equivalently if (EQQ − RQQ )EM M = EM Q
substitutability between tradables and non-tradables (EM Q = ∞) and there is a perfect elastic supply of non-tradables (RQQ = ∞), then the tariff will not have any effect on welfare, as consumers will just move all their consumption towards the non-tradable good from the moment the domestic price of the importable goes marginally up. Next, we find an expression for dQ, by substituting out dU in equation (2.6) and using (2.7)–still holding dT=0: 2 i h EM 1 1 Q ) dt = (ΦdQ − T EQM dt) T dA + tT 2 (EM M + γEU Φ EQU
#
"
Isolating dQ from this expression is straightforward:
dQ =
i 1 h DQ,y i T h DQ,y EM M DQ,y EM Q T dA + tT + EQM ( tT + 1) dt Φ γ Φ γ Φγ
(2.8)
This means that, again invoking the stability condition γ > 0, an increase in aid will appreciate the (exportable) real exchange rate. This is the standard Dutch Disease effect. The appreciation will be more pronounced (i) the lower are elasticities of supply and demand7 (Φ), as price effects will then dominate quantity effects, (ii) the higher is the share of spending directed towards non-tradables (DQ,y ). The effect of an import tariff on the real exchange rate is more complicated, with an income effect and a substitution effect pulling in different directions. The first ‘dt’ term 7
The latter in absolute value.
34
captures the income effect. As EM M is negative, an increase in the tariff will depreciate the real exchange rate. We can think of this as the equivalent of a negative transfer-type income effect linked to the distortionary effect of the tariff. The second ‘dt’ term captures the substitution effect, which pulls consumption away from the relatively more expensive tradables towards non-tradables. This implies demand for non-tradables going up, pulling resources away from the exportables sector in favor of producing more non-tradables. To maintain equilibrium, the real exchange rate appreciates. Hence, ex-ante we cannot tell which way the real exchange rate will go.
2.2
Adding the Regional Dimension
2.2.1
Adding a regional good to the model
Next, we introduce a regional dimension by adding a regionally tradable good that is not tradable with the rest of the world. Hence, for this good the SOE of price-taking in international markets does not hold. We obviously have to introduce a second country, with both countries together making up the region. We name these two countries ‘Home’ and ‘Foreign’, with ‘HOME’ variables designated by capital and ‘foreign’ variables by lower case letters8 . It is assumed that ‘Home’ has a cost disadvantage over ‘Foreign’ in producing regional goods. ‘Home’ will thus be a net importer of these goods. Initially, ‘Home’ applies an ad valorem tariff ‘τ ’ to such trade. The external ad valorem tariff ‘t’ vis-`a-vis the rest of the world is common, so we are dealing with a customs union rather than a free trade area. In general, the relation with the rest of the world is the same for both countries. Both countries’ balance of payments’ deficits are filled through an exogenous aid flow. 8
‘International’ variables, like prices determined in world markets, are also depicted in capital letters.
35
2.2.2
Equilibrium in ‘Home’ and ‘Foreign’
Both countries consume a non-tradable good, a regional good, and a good imported from the rest of the world. Both produce a non-tradable, a regional good and a good exported to the rest of the world. All prices for ‘Home’ are still valid from section (2.1). We add PR and PR∗ , which are the domestic and regional price of the regional good, and Z, which is the ratio of the price of the regional good to the price of the internationally exportable good. PR = PR∗ (1 + τ )
Z=
PR∗ PX∗
(2.9)
Similarly, we have the equivalents of all the ‘Home’ prices for ‘Foreign’, remembering that prices governing relations with the rest of the world are in common with ‘Home’. pR = PR∗ (1 − φτ )
q=
pN PX∗
(2.10)
Note from the expression of pR that the tariff τ applied by ‘Home’ induces a tax on the production of regional goods for ‘Foreign’. We will have more to say about this interdependency in the next section. As we did in the previous section, we can again write the economy-wide budget constraints, first for ‘Home’:
E(PN , PR , PM ; U ) = R(PN , PR , PX ; L)+ ∗ ∗ PM A + tPM EM + τ PR∗ (ER − RR )
Dividing through by the same num´eraire, PX∗ , we can write:
E(Q, Z(1 + τ ), T (1 + t); U ) = R(Q, Z(1 + τ ), 1; L)+ 36
(2.11)
T A + tT EM + τ Z(ER − RR )
(2.12)
By definition, the quantity of non-tradables consumed (EQ ) needs to be equal to the quantity of non-tradables produced (RQ ):
EQ = R Q
(2.13)
Deducting the non-tradables equilibrium from the overall budget constraint gives us the overall trade balance for ‘Home’:
Z(ER − RR ) + T EM = RX + T A
(2.14)
We can write the equivalents of equations (2.11) to (2.14) for ‘Foreign’, with ‘a’ the aid inflow into ‘Foreign’.
e(pN , pR , pM ; u) = r(pN , pR , pX ; l)+ ∗ ∗ eM + PM a + PR∗ (rR − eR )φτ tPM
(2.15)
e(q, Z(1 − φτ ), T (1 + t); u) = r(q, Z(1 − φτ ), 1; l)+ T a + tT eM + Z(rR − eR )φτ
(2.16)
eq = rq
(2.17)
Z(eR − rR ) + T eM = rX + T a
(2.18)
37
2.2.3
Regional links
We have now set up all the equilibrium equations for both countries individually, but we still need to link the two together through the regional market. First of all, we need an expression saying that trade in regionally tradables only takes place between the two countries making up the regional trade agreement. As we assumed that ‘Foreign’ has a cost advantage, it will be a net exporter of regional goods and the following expression will be positive. ER − RR = rR − eR
(2.19)
Adding both countries’ trade balances (equations 2.14 and 2.18), we get the trade balance between the region and the rest of the world, which is a similar expression as equation (2.4) from section (2.1):
T (EM + eM ) = RX + rX + T (A + a)
(2.20)
As (2.19) is positive, ‘A’ will be larger than ‘a’, with the aid flow derived in section (2.1) in between. What about the links between both countries through the setting of the tariff τ ? Appendix A shows that ‘Home’ decreasing τ implies a lower tax on the production of regional goods for ‘Foreign’. With the help of Figure A1, the appendix derives an expression for φ: φ=(
ERR − RRR ) eRR − rRR
(2.21)
Under normal behavior of supply and demand elasticities, this expression will be positive.
38
2.2.4
Interpreting the model: regional integration and aid
We are now at the stage where we can take total differentials and start interpreting the model. However, by adding the regional market we have substantially increased the complexity of the model. Indeed, we now have a 5x5 model9 which is quite cumbersome to solve ‘by hand’, with on top of that little hope of gaining much intuition. This is a first reason for moving towards a computable general equilibrium in the next chapters. We can still analyze our regional set-up in an informal way however, comparing the analysis with the model of section (2.1), thus paving the way for the interpretation of the results of the CGE model. We discuss (i) the changing concept of the real exchange rate, (ii) substitution effects of lowering regional tariffs, (iii) welfare and income effects of lowering regional tariffs and increasing aid, and (iv) effects on ‘Foreign’. Firstly, how do we think about real exchange rates for ‘Home’ in this model? As we have added a regional good, the exportable real exchange rate is not equal to Q anymore, but is now the ratio of PN to a price index of international and regional exports. It is crucial to point out that even though ‘Home’ is actually a net importer of the regional good, its price is included in this producer price index!10 . The two relative prices (as we N = Q) and ( PPNR ). have 3 goods produced) actually faced by producers in ‘Home’ are ( PPX
We can think of the former as the RER with respect to internationally exportables and the latter as the RER with respect to regionally exportables. An aggregate exportable real exchange rate could then be defined as: PN
ERER =
!
(2.22)
PX RX +PR RR RX +RR 9
5 independent equations are (2.12), (2.13), (2.16), (2.17), and (2.19), for 5 unknown variables, U, u, Q, q and Z, with exogenous variables t, τ, A, a and T , and φ implicitly determined by τ (see appendix A for the last result). 10 This illustrates the difference between exports and exportables, the latter including those goods that face international forces in determining their equilibrium prices and quantities.
39
N Equally, we can think of two relative prices that govern consumption incentives, ( PPM )
and ( PPNR ). It does not make sense to define an aggregate importable real exchange rate as the main exogenous shock we are interested in is a change in τ , which implies the regional terms-of-trade change. Secondly, what happens when ‘Home’ lowers its regional tariff τ ? On the production side, ‘Home’ will loose market share in the regional good market to more efficient ‘Foreign’ producers. This will move factors of production out of the regional sector, towards the non-tradable and the internationally exportable sectors. In the latter the response will be purely through an increase in quantities produced (as PX = PX∗ is given in world markets), whilst for the former, at least part of the adjustment will be through a price decrease. What happens to the ERER? In terms of definition (2.22), PR goes down, RR goes down, RX goes up and PN goes down. Movements in the ERER will hence depend on information not given in this highly stylized set up, i.e. on assumptions made about income and price elasticities. On the consumption side, consumers will spend relatively more on buying regional goods (ER up) , decreasing demand for non-tradables and importables (EM and EN down). Thirdly, what about welfare? In the previous paragraph, we only analyzed the equivalent of the substitution effect of the tariff decrease. What about the income effect? We first need to know whether income is going up or down. This brings us to standard regional trade welfare effects. If trade creation dominates trade diversion, then welfare and income go up, as discussed in section (1.3). However, as the regional good is not tradable with the rest of the world, we cannot really talk meaningfully about trade diversion in this model. There is definitely trade creation however, as the more efficient ‘Foreign’ country now supplies more of the total production of the regional good. Hence there is aggregate net wealth creation in this set up. We can say more, in that welfare goes up in both ‘Home’ and ‘Foreign’. This can be seen from standard welfare analysis as in figure 40
A1 of appendix A. When τ is reduced to zero, consumer surplus in ‘Home’ goes up by areas c+d+e+f, whilst producer surplus goes down by area ‘c’, and tariff revenue goes down by area ‘e’. Net welfare goes up by areas ‘d’ and ‘f’, which are standard deadweight loss triangles. Through similar analysis, we can see that ‘Foreign’ welfare goes up by area k+l+m. Assuming all goods are normal, the positive income effect translates into increased demand for all 3 consumer goods, with the relative demand increases determined by the respective income elasticities. This will put upwards pressure on PN and PR . As in section (2.1), we again have inflationary pressures through the income effect and deflationary pressures through the substitution effect. Once again, we need a more detailed model to be able to sign the movements in our key relative prices11 . Fourthly, we have seen that welfare in ‘Foreign’ goes up, but what happens to its relative prices? The reduction in the value of τ increases the price of regional goods in ‘Foreign’, inciting foreign producers to increase production, and consumers to decrease consumption of the regional good. We can perform the same analysis as for ‘Home’, with similar conclusions, i.e. we cannot sign the movements of the relative prices we are interested in.
2.2.5
Towards a CGE modeling strategy
This last section serves as a bridge between the current chapter that uses highly stylized models of international trade and the next chapters in which I develop a computable general equilibrium model. In the context of regional trade integration, CGEs have ‘helped us to understand theoretical interactions in models that are too complicated to study analytically’ Baldwin and Venables [1995]. A CGE allows a researcher to put more structure and detail into her model. In our case this takes the form of some key elasticities which 11 Equally, this type of static theoretical model does not address supply-side responses, which play an important role in the dynamics of relative price movements.
41
enrich our understanding of the regional integration mechanisms at work. As such elasticities of substitution governing consumption and production in domestic, regional and international goods are introduced through specification of constant elasticity of substitution (CES) and constant elasticity of transformation (CET) functions. These create a ‘partial insulation of the domestic price system from changes in world prices of sectoral substitutes. Through choice of substitution elasticities, the CET and CES functions provide a continuum of tradability at the sector level’ [Devarajan et al., 1993]. This is a much better approximation of reality, and allows for two-way trade within sectors. Also, the introduction of substitution elasticities makes the analysis of Vinerian trade diversion possible. Further, elasticities of substitution give us an alternative way of thinking about regional integration. As we have seen in section (1.3), the removal of commodity trade barriers is only one way of thinking about regional integration. One way of capturing these ‘deeper’ integration issues is through differentiating substitution possibilities for goods traded within the region vis-`a-vis goods traded with the rest of the world. Other elasticities such as price and income elasticities can be specified, and we can see how changing these can alter sectoral distributions and welfare effects. The CGE will help us to understand better distributional effects as we add substitution possibilities and specify wage and other income sources. Relatedly, the introduction of extra sectors allows the analysis of transfer paradox issues. Finally, a CGE can introduce dynamics and productivity effects into our model, for example to analyze the supply-side issues specified in chapter 1. We will not go into more detail in how our trade model translates into the CGE, as that is exactly what the next chapter does12 . 12 It is of course perfectly possible to analyze all the issues mentioned in this section from purely theoretical perspectives. Researchers have done this, but mostly in partial equilibrium frameworks. So, in a way, one could argue that the argument still boils down to one of ‘solvability’ and intuition/interpretation of results.
42
Conclusion
This theoretical chapter helped us in dealing with the questions at stake through the use of duality-based neo-classical models of international trade. The first section showed the well-known effects of aid on welfare and the real exchange rate in the presence of a tariff. Aid unambiguously increases welfare, whereas tariffs decreases it under reasonable parameter behavior. Aid appreciates the real exchange rate, whilst the effect of a tariff on the real exchange rate is undetermined ex-ante as substitution and income effects go in different directions. The second section added a regional dimension to the model. Hence, we included another country and a regional good, tradable between the regional partners, but nontradable with the rest of the world. The model’s increased complexity precluded pursuing the analytical strategy of section 1 (taking total differentials and solving for endogenous variables). However, we made a start with the analysis of the model by informally describing the forces at work. This showed that a lot of the effects in which we are interested are specific to the values the model’s key parameters (mainly elasticities) take. This is both an advantage, in that one can generalize results across a vast simulation space, but when a theoretic model becomes too complex, it becomes hard to grasp the intuition behind the algebra. We hence made the case for pursuing a computable general equilibrium strategy, in which values for these key parameters can be chosen, and robustness checked through sensitivity analysis.
43
Chapter 3
Computable General Equilibrium model: Structural Equations
Introduction
We now construct a highly stylized Computable General Equilibrium (CGE) model to be able to answer questions that were intractable through pure theoretical modeling in previous chapters. To reduce data requirements, the model is set up for only one country. This means we will have to put some structure on this country’s relations with both the region and the rest of the world. In this chapter the structural relations governing the CGE are developed, whilst in the next chapter the CGE is fitted on data for Rwanda. Chapter 5 then consists of running simulation experiments we are interested in and discussing the results. In the following sections of this chapter, we go over the details of the structural
44
relations of the model1 . We start with a general outline after which we work through a price block, a quantity block, an income block, an expenditure block and finally a market clearing and closure block, which together make up the model. A seventh section discusses ways of analyzing real exchange rate issues, bearing close resemblance to section (2.2.4) of the previous chapter. The final section then introduces welfare analysis, showing how we can improve our understanding of distributional issues through the CGE.
3.1
Set up of the CGE
The stylized social accounting matrix in the accompanying table gives an overview of the relations between activities, commodities, factors, institutions and the external world. Payments are made from columns to rows. For example, households spend (trace the household column) all their money on ‘household consumption’2 , which is paid to the ‘commodities’ row, as households’ money is spent on agricultural goods and services. Investment and savings are abstracted from as the model is essentially static, implying a pure endowment economy. This also means that foreign savings, which equal foreign aid, wholly accrue to Government. Following section (2.1), the Government sector receives tariff revenue and foreign aid and passes this on to households in a lump-sum non-distortionary way. Households own all of the two factors of production, capital and labor, land being lumped together with capital. Full employment of those production factors is assumed throughout, with households spending their factor income on services and agricultural goods3 . Services are completely non-tradable, which means most of the structural characteristics in this chapter are only applicable to the agricultural sector. There are no intermediate inputs. The model is completely real, so there are no mone1
This is standard practice for outlining CGEs, see e.g. Devarajan et al. [1994]. A corollary of there being no saving in our model. 3 Sectors are indexed by the letter ‘i’. 2
45
46
HoH
Gov’t
Regional
RoW
Total
I
P
T
S
Factors
C
E
Commodities
E
R
SE
Stylized Social Accounting Matrix Activities AG
Activity
Payment for input use in resp. sectors
Imports from the region Imports from RoW Supply Exp. AG
Import Tariffs
Supply Exp. SE
Domestic Sales SE
Factor Expenditure.
Factor Payments to HoH
HoH Expenditure
Total HoH Cons.
Gov’t transfers
Gov’t transfers to HoH
Table 3.1: Stylized Social Accounting Matrix (SAM) EXPENDITURES Activities Commodities Factors HoH Gov’t AG SE Domestic Sales AG
Extraregional BoP
Forex Inflow
Intraregional BoP
In-
Outflow
Forex
Gov’t come
In-
HoH Consumption
Factor come
AG GDP at market prices SE GDP at market prices Absorption
Regional Exports
Exports to RoW
Total
Regional RoW
tary effects4 . Appendix B contains a full list of variables and parameters, whilst appendix C gives an overview of the model’s equations5 .
3.2
Price Equations
Reflecting the set up in chapter 2, imports from and exports to the rest of the world are governed by the SOE assumption, i.e. world prices are determined exogenously. Prices for regionally tradable goods are determined endogenously within the model. Details regarding the derivation of these prices are specified in the next section (the quantity block). Here it suffices to state that these regional prices P Rimr and P Rier need pinning down elsewhere in the model. It is assumed that the region and the rest of the world are governed by the same nominal exchange rate EXR. Tariffs are levied on all imports, but can differ according to origin. There are no export subsidies or taxes. Equations (3.1) to (3.4) below give us four domestic prices Pimw , Pimr , Piew and Pier in terms of world prices (exogenous), regional prices (endogenous), tariffs (exogenous) and the nominal exchange rate (endogenous).
Pimw = pwimw (1 + tmw i ) ∗ EXR
(3.1)
Pimr = P Rimr (1 + tmr i ) ∗ EXR
(3.2)
Piew = pwiew ∗ EXR
(3.3)
Pier = P Rier ∗ EXR
(3.4)
4 It should be noted that the symbols used in chapters 3, 4 and 5 do not bear any resemblance to symbols used in previous chapters. E.g. ‘Q’ was the real exchange rate in chapter 2, but in the CGE set-up, ‘Q’ is used for domestic absorption. 5 As a convention, endogenous variables appear in roman capital letters, whilst exogenous variables appear in one of the following three guises: (1) small roman letters, (2) Greek letters or (3) acronyms in roman capitals starting with an ‘A’–for shift parameters, and also for the exogenous AID variable.
47
As discussed in (2.2.5), we use CET and CES specifications to capture imperfect substitutability between domestic, regionally tradable and internationally tradable goods. The first level of the CES/CET functions distinguishes between international and local goods. The second level splits up the local goods, differentiating between regional and domestic goods. The first order conditions for the CET and CES functions are specified in the next section in equations (3.15), (3.17), (3.19) and (3.21). As CES and CET aggregation functions are linearly homogenous, we can replace the original cost function of the respective optimization problems with the following accounting identities6 : Piq =
Piql QLi + Pimw M Wi Qi
(3.5)
Pid Di + Pimr M Ri QLi
(3.6)
Pixl XLi + Piew EWi Xi
(3.7)
Pid Di + Pier ERi XLi
(3.8)
Piql = Pix =
Pixl =
Equations (3.5) and (3.6) describe the price of the composite goods emanating from the 2 aforementioned CES functions. Equations (3.7) and (3.8) do the same for the 2-level CET functions. Finally, equations (3.9) and (3.10) define price indices. P P I is a Producer Price Index (effectively the GDP deflator), and CP I is a Consumer Price Index, where βP,i and βC,i are production and consumption shares respectively.
PPI =
X
βP,i ∗ Pix
(3.9)
βC,i ∗ Piq
(3.10)
i
CP I =
X i
6
This is derived in detail in Devarajan et al. [1990].
48
3.3
Quantity Equations (including the regional market)
This section describes the supply side of the model. The structure governing production ¯i is the fixed capital stock and Li the labor force active is of the Ricardo-Viner type. K in sector i. Hence the following production functions: ¯i 1−λi ) Xi = APi (Lλi i K
(3.11)
To allow for possible wage discrepancies between different sectors, a wage distortion (wdisti ) is introduced. The W below can thus be interpreted as the average wage across sectors. The total wage bill is then P
P
i wdisti Li W ,
with total capital income equal to
¯i . Standard first order conditions give: Ri K
wdisti ∗ W =
Ri =
λi Xi Pix Li
(1 − λi )Xi Pix ¯i K
(3.12)
(3.13)
(3.11) to (3.13) specify 6 equations (as we have two sectors), from which 6 unknowns XAG , XSE , W , LAG , RAG and RSE are derived. These are solved out as a function of ¯ 7 , sector-specific capital stocks (K ¯i ), labor and capital shares in total labor supply (L) production (the λi ’s) and the shift factor in the production function (APi ). Even though introducing the wdisti parameter manages to drive a wedge between sectoral wages, this is obviously a very crude way to model the labor market. Once we start experimenting with exogenous shocks (see chapter 5), labor will move between sectors to equalize (wdisti ∗ W ). One should think of this as the exogenous shock moving labor at the margin, with the equilibrium depicted here accounting for labor market characteristics ¯ see also equation (3.31) below), Note that as labor is mobile between sectors, the total labor supply (L, and one of the two sectoral labor supplies (here we chose LAG ) determine the other sector’s labor supply (here LSE ). 7
49
like risk premia when moving sectors and skill differentials between sectors8 . Next, equations (3.14) to (3.17) describe the two level CET functions, whilst equations (3.18) to (3.21) describe the two level CES functions. These functions were introduced in the previous section. h
ρ1 Xi = AT 1i αi XLρ1 i + (1 − αi )EWi
"
Piew αi EWi = XLi Pixl (1 − αi )
i
1 ρ2
(3.17)
h
M Wi Piql (1 − γi ) = QLi Pimw γi h
"
σt =
1 ρt−1
(for t=1,2), and σt =
1 1+ρt
i −1 ρ3
(3.18)
#σ3
QLi = AC2i δi Di−ρ4 + (1 − δi )M Ri−ρ4 M Ri Pid (1 − δi ) = Di Pimr δi
(3.16)
#σ2
Qi = AC1i γi QL−ρ3 + (1 − γi )M Wi−ρ3 i "
(3.14)
(3.15)
h
"
1 ρ1
#σ1
XLi = AT 2i βi Diρ2 + (1 − βi )ERiρ2 ERi Pier βi = Di Pid (1 − βi )
i
(3.19) i −1 ρ4
(3.20)
#σ4
(3.21)
(for t=3,4) are elasticities of substitution, Xi is
domestic output, XLi is the composite good for second level CET, Qi the composite good for first level CES (absorption), and QLi is the composite good for the second level CES. EWi , ERi , M Wi and M Ri are exports to the rest of the world, exports to the region, imports from the rest of the world and imports from the region. All these quantities have prices assigned to them which were determined in the previous ‘price-block’ section. Thus 8
For more sophisticated modeling of the labor market in a CGE context, see for example Adam and Bevan [2006] or Decaluw´e et al. [2004]
50
the 12 equations (3.5) to (3.8) and (3.14) to (3.21) determine 8 quantities and 4 prices: XLi , EWi , Di , ERi , Qi , QLi , M Wi , M Ri , Piq , Pix , Piql , Pixl . For the determination of Pid see the ‘closure’ section below. The introduction of substitutability between regional and international exportables, also makes trade diversion possible, which was not the case with the model in section 2.2 (see chapter 5 for more on this). Prices governing regional trade, i.e. P Rier and P Rimr , are now the only prices left undetermined. Regional demand for domestic goods (=exports to the region) is determined through the specification of a log-linear downward sloping demand function:
ERi = AERi ∗ (GN I R ∗ EXR)1 ∗ (
P Rier η1 er ) P Rsub
(3.22)
Import demand by the domestic economy for goods that are produced in the region is similarly governed by a log-linear specification, again giving us a downward sloping demand curve: M Ri = AM Ri ∗ (Y h )2 ∗ (
P Rimr η2 mr ) P Rsub
(3.23)
GN I R and Y h are total income in the region and the domestic economy, expressed in foreign and domestic currency respectively9 . The ’s are income elasticities, which are positive as we are assuming normal goods, whilst the η’s are own price elasticities and er and P Rmr are prices of close substitutes of the respective regionally hence negative. P Rsub sub
tradables. For the former we use an index including prices of domestically produced goods d ) and goods produced for the rest of the world (P ew ). For the latter, we use a price (PAG AG d ) and internationally imported index comprising the prices of domestically consumed (PAG mw ). goods (PAG
Equations (3.22) and (3.23) are effectively the CGE equivalent of equation (2.19). Both regionally tradables are now partially substitutable with international tradables 9
In the domestic economy, total income is equal to total household income Y h , see (3.24).
51
through the CES and CET specifications. We have not put any structure on the regional partner(s), but it should be clear that domestic agents can now influence regional prices. The biggest consequence (and drawback) of this 1-country specification is that ER and M R are de facto determined in different markets, such that we loose the interdependency in the regional market of the chapter 2 model. AERi and AM Ri are shift parameters.
3.4
Income Equations
This section describes the sources of income for the institutional actors (the household sector and the government sector). Equation (3.24) defines total household income, which is made up of factor income and government revenue. Equation (3.25) shows total tariff revenue which is made up of revenue from tariffs levied on imports from the rest of the world plus tariffs levied on imports originating in the region. Finally, equation (3.26) shows that government revenue is made up of tariff revenue and foreign aid.
Yh =
X
¯i Ri + wdisti Li W ) + G (K
(3.24)
X
(3.25)
i
T ARIF F = EXR ∗
mw mr mr (tmw i pwi M Wi + ti P Ri M Ri )
i
G = T ARIF F + (EXR ∗ AID)
3.5
(3.26)
Expenditure Equations
Domestic demand is taken up by the household sector10 . Final demand for good ‘i’ (CDi ) is specified by a Stone-Geary Linear Expenditure System (LES) as detailed in 10
As there is no investment demand, no intermediate goods demand, and no government consumption.
52
equation (3.27)11 . It is assumed there is a subsistence level in agricultural consumption (CDmin,AG 6= 0). One could think of this as the minimal food intake necessary to be able to perform standard labor tasks, but not in services (i.e. CDmin,SE = 0). βLES,i are P
marginal budget shares, which sum to 1 (
i βLES,i
= 1), by the properties of the LES.
Equations (3.28) and (3.29) define nominal and real GDP respectively. Piq CDi = Piq CDmin,i + βLES,i [Y h −
X
Piq CDmin,i ]
(3.27)
i
N GDP =
X
Pix Xi
(3.28)
i
RGDP =
X
Xi
(3.29)
i
3.6
Market Clearing and Closure
Finally, we come to the last block of equations, detailing how all markets clear and discussing how the model is closed. Again following Devarajan et al. [1994], we can link an equilibrating variable to each one of the market clearing conditions below. Firstly, as there is no investment or government demand, supply of the composite good Qi equals final demand by the household sector CDi (3.30). Pid is the equilibrating variable (it is easy to see this through substitution of equations in the ‘price’ and ‘quantities’ blocks). Secondly, full employment is assumed for both factors of production. Capital has ¯i ’s). Labor is mobile and allocated exogenously given sector specific endowments (our K between the agricultural and services sector through (3.12), with total labor supply again given exogenously through (3.31). 11
For more details on Linear Expenditure Systems, see Chung [1994], chapter 2.
53
Thirdly, (3.32) shows the balance of payments of the domestic economy. Foreign aid fills the BoP gap as it is the sole external capital flow and equals (imports − exports) at world prices. Foreign aid being determined outside of the model, it is the nominal exchange rate EXR which moves to close the BoP in the face of exogenous shocks. Changes in EXR will affect prices of imports and exports relative to domestic goods prices. This is of course equivalent to movements in the real exchange rate. In practice, as we are working with a general equilibrium model, without any monetary aspects, we can fix one of the prices as num´eraire. For this we will choose the nominal exchange rate EXR. Qi = CDi
(3.30)
¯= L
(3.31)
X
Li
i
M Wi pwimw + M Ri P Rimr = EWi pwiew + ERi P Rier + AID
3.7
(3.32)
Real Exchange Rates in the CGE
Similarly to what was done in section (2.2.4), we want to derive expressions for the key relative prices in the model. As simulated shocks will include changing trade environments, in the form of changing import tariffs, it does not make sense to construct aggregate price indices for an overall real exchange rate or even for an importable real exchange rate. We can however define an aggregate exportable real exchange rate (ERER). In addition, as tradables with the rest of the world have constant prices throughout, we can define an aggregate real exchange rate governing relations with the rest of the world (RERrow). ! P d D P i i i P i
ERER = P i
Di
P
Piew EWi + i Pier ERi P P EWi + i ERi i
54
!
(3.33)
PN T
RERrow = P i
P
Piew EWi + i Pimw M Wi P P EWi + i M Wi i
!
(3.34)
To make notation a little less heavy, PN T was used in (3.34) to replicate the nominator of equation (3.33). An increase in either ratio implies an appreciation of the respective real exchange rate. Further, we need expressions for the relative prices governing incentives within the agricultural sector. These are given below, with T oT the regional terms of trade and H the price governing relative consumption of domestically produced versus regionally produced goods. H=
d PAG mr PAG
T oT =
3.8
er PAG mr PAG
(3.35)
(3.36)
Distributional aspects of the CGE
To be able to research welfare consequences in our simulations, we need some way to distribute total income across the working population12 . Total income is made up of wages, capital income and government transfers. Two scenario’s are contemplated. (i) Households are fully aligned with sectors, i.e. there are ‘AG’ and ‘SE’ households. Transfers are distributed equitably across the population. This is the ‘communist’ scenario. (ii) There is an ‘AG worker household’ that receives ‘AG’ wage income only and there is a capitalist-urban’ household that receives its own (urban) wage income plus all capital and transfer income. We call this the ‘capitalist’ scenario. People moving between sectors in response to an exogenous shock are assumed to keep their initial capital and transfer endowments. 12
As we are assuming that the unemployed and children do not have any income, we only look at welfare effects across the employed population.
55
Conclusion
In this chapter, a CGE model has been set up for a typical low-income country. The different sections detail the model’s core equations. Here, I reiterate the CGE’s main features. There is no saving, nor investment, no intermediate inputs, no dynamics and no uncertainty. Government’s only function is to rebate tariff revenue to the private sector, with households the only domestic spending force in the economy. Households also own all factors of production, which are fully employed. There are two sectors, agriculture and services, the former comprising a domestic good, a regionally tradable good and an internationally tradable good, whilst the latter is completely non-tradable. A tariff is applied to imports from both the region and the rest of the world. There is a subsistence level of consumption for agricultural goods. Consumption and production for agricultural goods are governed by a two-level CES and CET respectively. Prices for internationally traded goods are determined outside the model, whilst prices for regionally traded goods are endogenous. Production is governed by a Ricardo-Viner framework with labor mobile across the two sectors and capital sector-specific. The trade balance, which is assumed to be in deficit, is closed by an external aid flow. The model is closed by Pid moving to equilibrate the goods market and by choosing a value for EXR .
56
Chapter 4
Calibrating the CGE to Rwanda
Introduction
In this penultimate chapter, the CGE model is calibrated, with data for the SAM loosely based on Rwanda for 2003. Section 1 describes Rwandan macroeconomic data, section 2 shows how values for the factors of production are calibrated, section 3 identifies initial values for the model’s various elasticities and section 4 discusses remaining issues to do with the calibration exercise. This chapter hence makes the bridge to the next one in which we simulate aid and regional integration shocks hitting our stylized economy. (As this is a short descriptive chapter, there is no ‘conclusion’ section at the end).
4.1
Macroeconomic and trade data
Table 4.1 shows the Social Accounting Matrix introduced in the previous chapter, but here with values filled out for Rwanda, whilst table 4.2 shows major macroeconomic 57
58
R E C E I P T S
HoH Gov’t Regional RoW Total
Factors
Commodities
Activities
Table 4.1: Social Accounting Matrix (SAM) for Rwanda (in Rwf billion) EXPENDITURES Activities Commodities Factors HOH Gov’t AG SE AG SE CAP LAB AG 358.41 SE 325.91 AG 457.84 SE 325.91 CAP 207.85 234.84 LAB 204.06 91.07 442.69 295.13 45.93 12.44 IMP 37.28 IMP 49.71 411.91 325.91 457.84 325.91 442.69 295.13 783.75 45.93
12.21
49.71
37.28
RoW EXP 37.50
21.28
Regional EXP 16.00
411.91 325.91 457.84 325.91 442.69 295.13 783.75 45.93 37.28 49.71
Total
Table 4.2: Key Macroeconomic 1999 GDP (bio Rwf ) GDP at market prices 644.9 Agriculture 270.4 Industry 124.9 Services 249.6 GDP at factor cost 600.3 BoP (mio USD) Exports (fob) 62.0 Coffee 26.5 Tea 17.5 ‘Other’ 15.9 Imports (fob) 248.8 Trade Balance -186.7 Int’l Reserves (eop) 174.2 PRICES (percent change) GDP Deflator -3.5 CPI -2.4 Nom. Eff. xr (’90=100) 43.1 Real Eff. xr (’90=100) 93.1
Indicators for Rwanda 2000 2001 2002
2003
705.7 292.3 144.4 269.0 658.9
754.3 305.2 163.0 286.2 699.3
825.0 341.6 175.9 307.5 760.7
905.3 373.9 192.7 338.7 823.9
89.8 22.5 24.3 41.6 239.8 -150.1 190.6
93.5 19.4 22.7 47.8 237.9 -144.4 212.1
67.3 14.6 22.0 25.6 234.7 -167.4 243.6
63.0 15.0 22.5 16.0 244.0 -181.0 214.2
3.3 3.9 37.5 83.2
0.2 3.4 36.8 80.2
0.0 2.0 31.1 70.2
8.7 7.4 29.9 67.6
aggregates for Rwanda from 1999 to 20031 . To construct the SAM, only information from the last column was taken. Previous years are included to put things in perspective and to show that 2003 was an average year for Rwanda’s economy. Thus, possibly biased results caused by year-specific events can be avoided. In 2003, GDP at market prices was 905 billion Rwandan francs (Rwf). With a baseline nominal exchange rate of 500 Rwf/USD, this equals 1810 million USD. Agriculture accounted for 45 percent (412 billion Rwf) and services for 36 percent of GDP (326 billion Rwf), industry making up the balance. It should be noted that the data in table 4.2 refers to the whole of the Rwandan economy, i.e. including the industrial sector, with our model only involving the agricultural and services sectors. In our calibration exercise, we always deduct from the macroeconomic aggregates (like GDP, the labor force, exports, etc.) the 1
Data taken from IMF [2004].
59
share that pertains to the industrial sector. Data on tariffs is taken from Ng [2007], who gives average tariff rates disaggregated by sector for more than 100 countries since 1980. In 2003, Rwanda applied an average tariff rate of 14.3 percent on agricultural imports2 . Export and import data are taken from IMF [2004]. It is assumed that the only international exports in agriculture are coffee and tea, whilst regional agricultural exports are ‘other’ exports, which are mainly foodstuffs. For agricultural imports, shares in total imports were taken from Ng [2007] and applied to total imports as appears in table 4.2. According to this data, Rwanda was running a heavy trade balance deficit during the period 1999-2003.
4.2
Factors of production
According to the 2002 census, there are 8.1 million Rwandans (MINECOFIN [2003]), of which 22.5 percent are younger than 7. MINECOFIN [2002] shows that of all Rwandans aged 7 or above, 62.5 percent are employed and of those, over 90 percent work in agriculture. For the labor force employed in services, I use the share in GDP to split the remaining 10 percent pro rata between industry and services3 . Assuming perfect markets, the wage and capital bill need to add up to equal GDP at factor prices. Adding compensation for land use to the capital bill, a capital bill-wage bill ratio of 3 to 2 looks reasonable4 . Applying this gives us a wage bill of 295 billion Rwf and a capital bill of 443 billion Rwf. How is the wage bill divided between agriculture and services? MINECOFIN [2002] 2
In the baseline run, where tariffs are applied to both international and regional imports, we do not attempt to differentiate the tariff rates between destinations. 3 This also implies that I do not include people that work in industry into welfare considerations which are discussed below. 4 This is for example what Adam and Bevan [2006] end up using in their CGE fitted on neighboring Uganda.
60
calculates a consumption poverty line to be Rwf 64000 per adult equivalent. We set the agricultural wage at Rwf 57,600, which is 90 percent of that poverty line. This means that a wage laborer in the agricultural sector, who possesses no capital, falls into the ‘poverty’ category by this poverty indicator5 . Using employment figures from the previous paragraph, the average wage in the services sector can be calculated (350,241 Rwf), as can the average wage across the economy (77,610 Rwf), which is above the 64,000 Rwf poverty line6 . In the last core experiment of chapter 5, subsistence consumption in agriculture is introduced. This reflects the importance of consuming food in household expenditure and implies that income elasticity in agriculture falls below 1. We use values of 50 and 90 percent of total agricultural consumption to simulate this subsistence consumption. The residual obtained from subtracting total labor income from total sectoral value added (GDP) is an amalgam of payments to land and capital. We do not distinguish between these so we call this total ‘capital income’. Without depreciation, we only need ¯i ’s in equation (3.13). We to set a real profit rate, after which we can solve for the K choose the same profit rate in both sectors, setting it at 10 percent. Finally, we need an estimate for regional income to use in equation (3.22). To this end, Gross National Income (GNI) of the regional partners is added together, with the data coming from the World Bank’s World Development Indicators. Regional partners are defined as all Rwanda’s neighbors (Burundi, DRC, Uganda and Tanzania), plus Kenya. This gives us GN I R = 30.8 billion USD. 5
This is logical as a wage laborer without any capital is the poorest economic agent in our model. The ‘Quantity Equations’ section in the previous chapter explains how introducing a labor market distortion (wdisti ) allows for a wage differential between sectors even when labor is mobile. The equations from that section are used to calculate the numbers presented here. 6
61
4.3
Elasticities
In this section, values for the various elasticities used in the model are derived. This is achieved mostly by looking at existing literature in which these are estimated. These values will profoundly impact on our economy, so we need to ensure we set these parameters at reasonable initial values. Firstly, we need baseline values for the elasticities of substitution for the 2-level CES and CET functions. These are inspired by Decaluw´e et al. [1994] and Adam and Bevan [2006], who make assumptions about similar elasticities for countries that are part of the West African Economic and Monetary Union (WAEMU) and for East African countries respectively. For our baseline calibration, substitution elasticities are set as shown in table 4.3. How do we justify these values? In particular, why do we set international elasticities lower than regional ones? On the production side (σ1 = 0.5 and σ2 = 1.5), we expect more difficulty in switching production between cash crops (coffee and tea) and regional goods than between domestic and regional goods. The latter are mainly foodstuffs which we assume are not very variable between neighbouring countries. Also, with ongoing trade liberalization in many Low-Income Countries, including Rwanda, production substitution possibilities will become larger7 . Again, this might favour production substitution possibilities relatively more in the regional market, at least in the short term. 8 . A further reason is that exports to international (read:western) markets are much more heavily regulated compared to regional markets in terms of standards. We can think of quality thresholds, packaging requirements, sanitary and phytosanitary measures, etc. On the consumption side (σ3 = 0.5 and σ4 = 0.7), tastes might be more alike between regional 7
Adam and Bevan [2006] for example argue that the export share in current output is relatively low as a result of a long history of anti-export bias in trade policy. They set their 1-level CET elasticity of substitution equal to 2. 8 One could argue the opposite in the longer run as it can be expected that incentives will move producers towards internationally exportables as regional goods probably have lower income elasticities. On the other hand increased ‘deeper’ regional integration might well increase σ2 disproportionably.
62
Table 4.3: Initial Values for key Elasticities Parameter Initial Value Name Initial Value σ1 0.50 1 1.00 σ2 1.50 2 1.00 σ3 0.50 η1 -1.00 σ4 0.70 η2 -1.00 partners than between the region and the rest of the world. Also, certain imports, e.g. intermediate inputs, might well only be available in international markets. Next, we need own price elasticities and income elasticities for regionally tradables (η’s and ’s in equations 3.22 and 3.22). For the own-price elasticities, the η’s, we assume a reasonably neutral value of -1 for both equations. Income elasticities 1 and 2 are initially set at 1. This implies that extra income will be spent proportionally on regionally tradables.
4.4
Calibration
For the base-run of the model to work, some variables and parameters still need to be assigned initial values. First of all, all domestic prices (Pimw , Pimr , Piew , Pier ,Piq ,Pix ,Piql ,Pixl and Pid –and hence P P I and CP I), are set to 1 in the baseline run. Recalling the values in the SAM, we now can, through the structural equations derived in the previous chapter, solve for the parameters that have not been assigned an initial value yet. Effectively, this amounts to solving the model in reverse [Devarajan et al., 1994]. The parameters that are solved in this way are listed in appendix B under ‘Parameters calculated through Calibration’. To close the model, we fix the nominal exchange rate EXR at 500 Rwf/USD and use domestic prices for equilibrating goods markets as explained in section (3.6). GAMS software was used to implement the CGE (see www.gams.com)
63
Chapter 5
Aid and Regional Integration: Simulating Exogenous Shocks
Introduction
With our baseline CGE up and running, we are now in a position to analyze the general equilibrium effects of regional integration and increasing aid flows. This last chapter describes the adjustment process of essentially 5 external shocks: (i) an increase in aid that is fully consumed, (ii) a decrease in regional trade barriers through abolishing the tariff on regional trade, (iii) the two previous shocks taken together, (iv) part of the aid is not consumed, but is used to increase productivity in the different sectors, (v) part of the aid is spent on increasing productivity in the non-tradable part of the agricultural sector (DAG ). This last experiment also introduces subsistence consumption in agriculture, to look at the effect of diverging sectoral income elasticities on distributional outcomes. Table 5.1 gives an overview of the 5 main simulations and reports the behavior of key variables.
64
The reader is reminded of the fact that, as we do not have an investment function nor any growth in the labor supply, output changes will be reallocations around the Production Possibility Frontier (PPF). The original PPF will then be shifted outwards through the productivity increases simulated in experiments 4 and 5. Nowhere however, do we drop the assumption of full employment1 .
5.1
Simulation I: Aid increase fully consumed
Our first experiment is an aid increase that is fully consumed. Like this, we isolate the typical Dutch Disease demand-side effects. To facilitate comparisons with the next set of experiments, the increase in aid is set equal to the tariff revenue collected from regional trade (at baseline prices). This implies an extra 10.7 million USD in aid, or a 16 percent aid increase. Firstly, as expected, all the relative prices with ratio’s of non-tradables to tradables (ERER, RERrow and H) go up, i.e. appreciate. The appreciation that would materialize if all tradable prices were fully exogenously determined, is given by RERrow. However, when adding endogenous regional prices to tradable price indices, the appreciation is muted (ERER). As explained in previous chapters, the increase in aid gives domestic agents incentives to produce relatively more non-tradables and consume relamr decreasing, tively more importables. In the regional market, this translates into PAG er increasing, moving promoving consumption towards regionally importables, and PAG
duction towards non-tradables. These movements improve the regional terms-of-trade (T oT goes up).
1
And hence the point made by Bourguignon et al. [2005] about idle capacity (see section 1.1.2) is not taken up in this analysis.
65
.
66
.
67
mr and P In theory, one could even think of a very inelastic PAG N T and a very elastic er , to get the overall real exchange rate to depreciate in the face of an aid increase2 . PAG
This is however an unlikely prospect as it is reasonable to assume that, at least in the short-run, consumers can switch their spending patterns more easily than producers can change their production processes. This is exactly the supply-side response issue we talked about at length in section (1.1). The exact quantification of these effects is dependent on the values of several parameters. If we increase elasticities of substitution for example, the adjustment process becomes easier, i.e. falls less on prices and more on quantities, with a more muted RER appreciation. Secondly and relatedly, spending in both import categories goes up and earnings in both export categories go down, as expected. Even though regional import quantities increase by 21 percent, regional imports in value terms (at international prices) increase by mr goes down. For international imports of course the percentage less than 1 percent, as PAG er doesn’t increase is the same in both cases, as prices stay put. On the export side, as PAG
increase very much, the decrease is more or less the same for both export destinations (both in value and in quantity terms). Interestingly, in absolute value both regional exports and international exports go down by about the same amount, even though regional exports are less than half of international exports. This is important bearing in mind our argument in section 1.1.1) that regional exports might be less important for our competitiveness story. These effects also make the regional trade balance deteriorate less than the international trade balance. Thirdly, looking at quantities, the expected increase in real absorption (‘Q’) clearly materializes, going up by 3.0 and 0.1 percent in the agricultural and services sectors respectively–and of course by even more in value terms (as the consumer prices increase). 2
Our table of results does not show an aggregate real exchange rate index, but the movements in the relative prices reported show how this is a possibility.
68
The increase in the price of non-tradables (DAG and DSE ) means that factors of production will move towards the services sector and the non-tradable part of agricultural production, with the exportable part of agricultural production facing contractionary pressures. As sector specific capital stocks are fixed, we see that labor moves from agriculture to services to increase production in services and decrease it in agriculture. Fourthly, looking at welfare effects, real household income goes up by 1.8 percent. As households are the only consumers of final goods in the domestic economy, aggregate household welfare increases unambiguously as real consumption of both agricultural and service goods go up. What happens to income distribution according to our two scenario’s? Under both scenario’s, everyone is better off, with the percentage increase in real wages the same in both sectors as everyone has similar consumption patterns. The winners are obviously those households that move from the agricultural to the services sector.
5.1.1
Simulation II: Regional free trade
What are the effects of abolishing the tariff governing regional trade? We can apply the theoretical analysis from section (2.1), with the added twist that the tariff reduction now takes place in the presence of other trade distortions (i.e. there is a tariff on international trade in place when we reduce the regional tariff). What are the main results of this experiment? Firstly, what happens to overall welfare? In the agricultural sector, there is trade cremr below P d , switching consumption ation as the decrease in the regional tariff brings PAG AG
from domestically produced goods to regional imports. Hence regional imports are up by 40 percent, even though in value terms spending on regional imports decreases. The mr falls below P mw . There will be some trade diversion as tariff reduction also ensures PAG AG
69
the regional partners have upward sloping supply curves for their exports to the domestic economy, whilst the rest of the world has completely elastic export supply curves. Lastly, tariff revenues go down, but less than expected as international imports went slightly up. The net welfare effect is positive as can be seen from real income going up by 1.7 percent. As all goods are normal in this model, we would expect consumption of all goods to go up. However, because of easier substitution possibilities within the agricultural sector3 , there is a (very small) shift in production towards agricultural goods, which makes consumption of services slightly go down4 . Similar conclusions as to the previous experiment can be made with respect to our different distributional scenario’s, i.e. everyone shares equally in the wealth creation in scenario 1, with the income distribution in the capitalist scenario actually becoming less skewed because the transfer income of the capitalists went down through decreased tariff revenues. Secondly, what happens to relative prices? The reduction of the regional tariff rate is clearly deflationary, and both exportable real exchange rates depreciate as the substitution effects outweigh income effect5 . Prices of exports hardly change at all6 , with all the ‘action’ coming from the decrease in prices of non-tradables, induced by the even bigger fall in mr . H does go up however, as P mr increases by more than P d . the price of PAG AG AG
Thirdly, the regional trade balance improves by the same amount as the international trade balance deteriorates, leaving the overall trade balance untouched. This is of course because we have not changed the aid flow, and hence the trade balance needs to stay constant. 3
This is linked to our Linear Expenditure System which leaves us with constant consumption shares in both goods 4 See Panagariya [2000] for a more stylized welfare analyis of a similar set up. 5 This is by no means a generalisable outcome. Indeed, if we for example lower substitution elasticities, this can be overturned into a depreciation 6 A consequence of our import and export regional markets not really being linked up. See also comment in section 3.3.
70
5.1.2
Simulation III: Aid increase fully consumed, with regional free trade
Our third experiment combines the two previous ones. This means abolishing the regional tariff, plus increasing aid by 10.7 million USD. The aid increase is still fully consumed. It is also equal to the loss of (original) tariff revenue and one could think of this as conditional aid ‘buying’ trade reform or an aid-for-trade package smoothing out revenue losses. The joint effects are an adding up of the results we discussed in the previous experiments. We should keep in mind that the aid inflow had mainly inflationary and the tariff reduction mainly deflationary effects, as expected. Exportable real exchange rate effects of both individual experiments pull in different directions, but we still have an overall appreciation, but more muted compared to the aid-only story. Regional imports mr reinforce each other, with increase now by almost 70 percent, as the decreases in PAG
international imports going up more modestly. Export quantities suffer with an overall decrease of 5 percent. Balance of trade effects are again halfway between the previous two experiments. Overall production is slightly down due to the displacement effects of the aid inflow. Welfare is going up all around (household income up 3.4 percent), as we concluded that both regional integration and aid had positive effects on aggregate consumption.
5.1.3
Simulation IV: Aid to increase sectoral productivity, with regional free trade
How do we introduce increases in productivity into our model? The CGE structure doesn’t really allow us to investigate the exact channels through which such productivity growth might happen, so we content ourselves with simulating this through changing shift and share parameters of the production functions (equations 3.11) and CETs (equations 3.14 71
and 3.16). In this section, half of the aid is used for consumption purposes, whilst the other half is assumed to go into some kind of investment that increases productivity in one or more (sub)sectors instantaneously. This is of course a very crude way of modeling productivity as the actual channel through which aid is transformed into productive investment is not specified. An alternative productivity story, more closely related to the ‘new’ regionalism described in section (1.3), can link the productivity increase to the reduction in the regional tariff and the opening up of competition and other dynamic effects. It should be kept in mind that the 50 percent of foreign aid spent on investment goods, will have their own marginal propensities to spend on international, regional and non-tradable goods. This in turn will trigger a sectoral adjustment process like any other aid inflow. Here, we make abstraction from these effects as the amount of aid going into productive investment actually never enters the economy, but affects it in the form of increases in productivity parameters. Concretely, the two simulations increase sectoral productivity through changing the shift parameters APi in equation (3.11). Experiment (4a) looks at productivity increases of 2.5 percent in both the agricultural and services sectors. Real GDP in both sectors logically goes up, with the productivity increase also having a marked impact on the exportable real exchange rates which appreciate much less compared to simulation III. In quantities as well as in value terms, compared to the baseline case, regional exports only slightly go down, with international exports actually going up, overturning the negative effects of the aid increase, and hence muting Dutch Disease effects. As the increase in productivity also falls on D(AG), the increase in production this entails moves importable consumption towards agricultural non-tradables. In value terms, importables don’t change that much. Predictably, welfareimprovements become more impressive across the board.
72
Experiment (4b) looks at a services productivity bias, i.e. we increase productivity by 2.5 percent in the services sector, but nothing happens in agriculture. We get very similar results as in (4a). There is again a smaller real exchange rate appreciation, with exports still going up (relatively to simulation III), and Dutch Disease effects again muted.
5.1.4
Simulation V: Non-tradable productivity, subsistence consumption and the transfer paradox
Our final experiments look at productivity increases in the domestic (non-tradable) agricultural sector. This is modeled through an increase in the share parameter β (experiment 5a). Experiments (5b) and (5c) keep the β increase, introducing a consumption subsistence level in the agricultural sector. (5b) sets this level at 50 percent of baseline agricultural consumption and (5c) sets it at 90 percent. (5a) is very similar to (4b) with again an increase in the productivity of a non-tradable (sub)sector. The reason for these simulations is to show how our experiments relate to a dynamic set up, as in for example Adam and Bevan [2006]. In their experiment 4, they also have a productivity bias towards non-tradables, which means that the real exchange rate appreciation is more muted compared to the time path when the productivity rise does not discriminate between goods. In our simpler model, we also pick up these effects that reduce the importance of Dutch Disease. (5b) and (5c) introduce subsistence consumption in the agricultural sector. The immediate effect is to make the income elasticity of demand for agricultural goods fall below one7 . It implies that, once a household has consumed this ‘subsistence’ amount of goods, it will spend relatively more of its income on services than on agricultural goods. Hence, 7
In our simple Linear Expenditure System, this implies the income elasticity of demand for services is now above one.
73
when we move from (4a) to (4b), we see the sectoral terms-of-trade move against the producers of agricultural goods. Further moving to (4c), income elasticities fall even lower, and households spend almost nothing of their supra-subsistence income on agricultural goods. The intersectoral terms-of-trade now move so much against agricultural producers that agricultural wage workers actually become worse off (their real wage falls by 1.5 percent) compared to the baseline case. This is, remember, despite the productivity increase in DAG . This worsens the income distribution under both distributional scenario’s. This is a type of intersectoral transfer paradox reminiscent of Matsuyama [1992] or Adam and Bevan [2006].
Conclusion
In the final chapter, the CGE developed in the two previous chapters, was used to simulate 5 exogenous shocks. As such we showed the standard effects of aid, regional integration, productivity increases and subsistence consumption in 1 sector of the economy. Dutch Disease effects are present in our CGE calibration, but become more bearable when aid and / or regional integration have positive productivity effects. Following an opening up of the regional market, net welfare goes unambiguously up, with trade creation effects dominating trade diversion and revenue losses. When introducing subsistence consumption, we found a version of the transfer paradox, as inter-sectoral terms-of-trade moved against agricultural producers.
74
Conclusion
This thesis studied aid and Dutch Disease and regional integration effects in a unified framework. The first chapter gave us a tour of the literature on aid and Dutch Disease, the transfer paradox and regional integration. It should be clear that recent research on these issues have left behind the neo-classical tradition of perfect markets and full employment to quite a large extent. However, in this thesis, we stayed within the confines of the neo-classical framework. The second chapter used the standard international trade model to shed light on the interactions between aid and regional trade. Adding a regional dimension to the standard aid and trade model, which was discussed at the beginning of the chapter, proved to be difficult, with the added complexity making the model intractable. Hence, the aid-cum-regional trade framework was discussed rather informally. This ‘solvability’ issue was our major incentive to pursue the construction of a computable general equilibrium in the second part of the thesis. Chapter 4 constructed the CGE, the main elements of which are: (i) no investment nor savings, (ii) households the only consuming force, (iii) CES and CET specifications governing consumption and production respectively, (iv) two sectors, (v) Ricardo-Viner production functions, (v) internationally tradables prices of which given exogenously, and (vi) regionally tradables prices of which given endogenously. Chapter 5 then calibrates the CGE to data for Rwanda, with chapter 5 running 5 core simulations. The results of these
75
experiments show aid-induced Dutch Disease effects which are however muted if there are positive productivity effects. Net welfare increases when regional trade barriers are reduced, implying that trade creation effects dominate trade diversion and tariff revenue losses. Finally a transfer paradox is unearthed when allowing for subsistence consumption in the agricultural sector. This takes the form of real agricultural wages going down in the presence of a positive productivity boost in the domestic production part of the agricultural sector, There is a vast number of directions in which this research can be taken forward. First of all, there is a need to introduce dynamics explicitly into the set-up, such that productivity effects of both the aid increase and the regional integration can be studied. Productivity effects can also be linked to externality effects between countries. This brings regional aid schemes into the picture, which for many landlocked African countries, might well be the only way to finance much-needed regional public goods. Ultimately, one would like to devise aid mechanisms and projects that could increase productivity differentially between countries, so that more backwards countries can ‘catch up’ (Think of Kenya and Rwanda). Distributional effects and political economy considerations must be addressed so that the integration project becomes a success. Again, this opens new research alleys. Further, stochastic elements can be added, especially to deal with the unpredictability of aid flows. There is a possibility also that delving deeper into pure theoretical modeling and the CGE, we might be able to unearth a transfer paradox at the national level, i.e. one can conceive of two aid-dependent countries with an RTA project, whereby because of productivity, aid and trade effects one of the two actually becomes worse off. Of course, this would need a more general set-up with two countries, for which a CGE is probably the best option. We could also introduce the issues on the ‘new’ regionalism agenda into the CGE, and put more structure on the labor market. 76
Appendix A
Deriving An Expression for φ
When ‘Home’ unilaterally sets a tariff for the regional good, it can be shown that this implicitly determines a tax on Foreign’s production of the regional good. Further, we can get an expression for this tax as a function of the tariff set by ‘Home’ and elasticities of supply and demand in both countries. That is the subject of this section. It should be stressed that this result is dependent on (i) third countries neither consuming nor producing the regional good, and (ii) on ‘Foreign’ having a cost advantage vis--vis ‘Home’ in producing the regional good. Figure A.1 shows how we can derive the implicit tax producers of regional exportables in ‘Foreign’ face when ‘Home’ sets an ad valorem tariff equal to τ . The left panel describes equilibrium in ‘Home’ in the presence of a tariff, whilst the middle panel shows equilibrium in ‘Foreign’. The right panel shows the equilibrium under full integration, i.e. τ = 0. As the two countries make up ‘the world’ in terms of the regional good, net imports destined for ‘Home’ are equal to net exports originating in ‘Foreign’. Hence, the distance between ‘x’ and ‘y’ in the left panel (net imports to ‘Home’), must equal the distance between ‘v’
77
and ‘w’ (net exports from ‘Foreign’). This implies that the price of regionally tradables in ‘Foreign’ will be lower than what it would be under free trade, but higher than under autarky. Hence, we are looking for an expression like (1 − z)PR∗ = pR > paR , with pR the price of the regional good in ‘Foreign’, paR the price of the regional good in autarky and PR∗ the price of the regional good under free trade. Assume further that the share of regional goods produced in ‘Home’ under free trade is α and the share of regional goods consumed β. The respective shares for ‘Foreign’ are then 1 − α and 1 − β. Finally, production and I = RI consumption of the regional good under free trade are designated: ER R
In ‘Home’, we then get: RRR =
I x − αRR τ PR∗
(A.1)
Re-arranging, we get an expression for y: I x = τ PR∗ (RRR ) + αRR
(A.2)
Doing the same for y: ERR =
I y − βER τ PR∗
(A.3)
and: I y = τ PR∗ (ERR ) + βER
(A.4)
Hence, we get, for the imports of non-tradables: I I y − x = τ PR∗ (ERR − RRR ) + βER − αRR
(A.5)
We can do the same for w and v, to come up with the following expression for w-v: I w − v = zPR∗ (eRR − rRR ) + (1 − α)rR − (1 − β)eIR
78
(A.6)
Finally, we set y-x equal to w-v to end up with z as a function of τ and elasticities of demand and supply in both countries:
z=(
ERR − RRR )τ eRR − rRR
(A.7)
The first term on the Right Hand Side is what we called φ in the previous section. If both countries have exactly the same demand and supply elasticities, then φ will be equal to one, and autarky prices will lie at symmetrical distances around the free trade price.
79
.
80
Appendix B
List of Variables and Parameters
B.1
Endogenous Variables
1. Pimw = Domestic price of imports from RoW in sector i 2. Piew = Domestic price of exports to RoW in sector i 3. Pimr = Domestic price of imports from region in sector i 4. Pier = Domestic price of exports to region in sector i 5. P Rimr = Regional price for imports from region in sector i 6. P Rier = Regional price for exports to region in sector i 7. Piq = Price of the level 1 CES Composite good in sector i 8. Piql = Price of the level 2 CES Composite good in sector i 9. Pid = Price of domestic good in sector i 10. Pix = Price of domestic output (level 1 CET composite good) sector i 81
11. Pixl = Price of level 2 CET composite good in sector i 12. P P I = Producer Price Index 13. CP I = Consumer Price Index 14. Qi = Level 1 CES composite good in sector i (absorption) 15. QLi = Level 2 CES composite good in sector i 16. Di = Domestic good in sector i 17. Xi = Domestic output (level 1 CET Composite good) in sector i 18. XLi = Level 2 CET composite good in sector i 19. EWi = Exports to RoW in sector i 20. ERi = Exports to the region in sector i 21. M Wi = Imports from RoW in sector i 22. M Ri = Imports from the region in sector i 23. W = Average wage rate 24. Li = Labor supply in sector i 25. Ri = Rental rate of capital in sector i 26. N GDP = Nominal GDP 27. RGDP = Real GDP 28. Y h = Total household income in domestic economy 29. G = Total government revenue 30. T ARIF F = Total tariff revenue (accruing to government) 82
31. CDi = Total household consumption in sector i 32. EXR = Nominal exchange rate
B.2
Exogenous Variables
1. AID = Total foreign aid (accruing to government, in US Dollars) 2. pwimw = world price of imports from the RoW 3. pwiew = world price of exports to the RoW ¯ = total amount of labor available in the economy 4. L 5. Ki = Fixed capital stock in sector i 6. GN I R = Total income of region (in US Dollars) 7. CDmin,i = subsistence consumption for good i 8. tmw = tariff rate for imports from rest of the world i 9. tmr = tariff rate for imports from the region i 10. σ1 (ρ1) = CET Elasticity of Substitution between exports to RoW and regional composite good
1
11. σ2 (ρ2) = CET Elasticity of Substitution between exports to the region and domestic good 12. σ3 (ρ3) = CES Elasticity of Substitution between imports from RoW and regional composite good 1
The 4 σ’s uniquely determine the 4 ρ’s in our system. See section (3.3).
83
13. σ4 (ρ4) = CES Elasticity of Substitution between imports from the region and domestic good 14. 1=income elasticity of region (for regional exports to the region) 15. 2=income elasticity of domestic economy (for regional imports) 16. η1=own price elasticity of regional demand for domestically produced goods (exports to the region) 17. η2=own price elasticity of domestic demand for regionally produced goods (imports from region)
B.3
Parameters calculated through Calibration
1. wdisti = Parameter measuring labor market distortion in sector i 2. APi = Production function shift parameter 3. AT 1i = CET Function shift parameter (level 1) 4. AT 2i = CET Function shift parameter (level 2) 5. AC1i = CES Function shift parameter (level 1) 6. AC2i = CES Function shift parameter (level 2) 7. λi = Production function share parameter 8. αi = CET function share parameter (production for domestic+regional market vs RoW) 9. βi = CET function share parameter (production for domestic vs regional market)
84
10. γi = CES function share parameter (consumption for domestic+regional market vs RoW) 11. δi = CES function share parameter (consumption for domestic vs regional market) 12. AERi = Shift parameter for regional demand for exports (domestically produced) 13. AM Ri = Shift parameter for domestic demand for imports (regionally produced) 14. βLES,i = Marginal budget share in the Linear Expenditure System (LES) 15. βP,i = Weights for PPI (avg. share of i in total domestic production) 16. βC,i = Weights for CPI (avg. share of i in total domestic consumption)
B.4
Other Parameters used in the context of CGE simulations
er = price index for substitutes for regional exports 1. P Rsub mr = price index for substitutes for regional imports 2. P Rsub
3. ERER = Real Exchange Rate for exportables 4. RERrow = Real Exchange Rate governing trade between domestic economy and rest of the world 5. H = Price ratio of domestic agriculture price to regionally importable agricultural price 6. T oT = Regional terms of trade
85
Appendix C
List of Model Equations
C.1
Price Block Pimw = pwimw (1 + tmw i ) ∗ EXR
(C.1)
Pimr = P Rimr (1 + tmr i ) ∗ EXR
(C.2)
Piew = pwiew ∗ EXR
(C.3)
Pier = P Rier ∗ EXR
(C.4)
Piq =
Piql QLi + Pimw M Wi Qi
(C.5)
Pid Di + Pimr M Ri QLi
(C.6)
Pixl XLi + Piew EWi Xi
(C.7)
Pid Di + Pier ERi XLi
(C.8)
Piql = Pix =
Pixl =
86
PPI =
X
βP,i ∗ Pix
(C.9)
βC,i ∗ Piq
(C.10)
i
CP I =
X i
C.2
Quantity Equations ¯i Xi = APi (Lλi i K wdisti ∗ W = Ri =
1−λi
)
(C.11)
λi Xi Pix Li
(C.12)
(1 − λi )Xi Pix ¯i K
(C.13)
h
ρ1 Xi = AT 1i αi XLρ1 i + (1 − αi )EWi
"
Piew αi EWi = XLi Pixl (1 − αi )
i
1 ρ2
(C.17)
h
M Wi Piql (1 − γi ) = QLi Pimw γi
(C.18)
i −1 ρ4
(C.20)
#σ4
ERi = AERi (GN I R ∗ EXR)1 (
87
ρ3
(C.19)
h
"
i −1
#σ3
QLi = AC2i δi Di−ρ4 + (1 − δi )M Ri−ρ4 M Ri Pid (1 − δi ) = Di Pimr δi
(C.16)
#σ2
Qi = AC1i γi QL−ρ3 + (1 − γi )M Wi−ρ3 i "
(C.14)
(C.15)
h
"
1 ρ1
#σ1
XLi = AT 2i βi Diρ2 + (1 − βi )ERiρ2 ERi Pier βi = Di Pid (1 − βi )
i
(C.21) P Rier η1 er ) P Rsub
(C.22)
M Ri = AM Ri (Y h )2 (
C.3
P Rimr η2 mr ) P Rsub
(C.23)
Income Equations Yh =
X
¯i Ri + wdisti Li W ) + G (K
(C.24)
X
(C.25)
i
T ARIF F = EXR ∗
mw mr mr (tmw i pwi M Wi + ti P Ri M Ri )
i
G = T ARIF F + (EXR ∗ AID)
C.4
(C.26)
Expenditure Equations Piq CDi = Piq CDmin,i + βLES,i [Y h −
X
Piq CDmin,i ]
(C.27)
i
N GDP =
X
Pix Xi
(C.28)
i
RGDP =
X
Xi
(C.29)
i
C.5
Market Clearing and Closure
Qi = CDi
(C.30)
¯= L
(C.31)
X
Li
i
M Wi pwimw + M Ri P Rimr = EWi pwiew + ERi P Rier + AID
88
(C.32)
Bibliography Chris Adam. Exogenous Inflows and Real Exchange Rates: Theoretical Quirk or Empirical Reality? in Peter Isard, Leslie Lipschitz, Alexandros Mourmouras, and Boriana Yontcheva (eds.), The Macroeconomic Management of Foreign Aid: Opportunities and Pitfalls, 2005. Chris Adam and David Bevan. Aid and the Supply Side: Public Investment, Export Performance, and Dutch Disease in Low-Income Countries. World Bank Economic Review, 20(2):261–290, 2006. Terence Agbeyegbe, Janet Stotsky, and Asegedech WoldeMariam. Trade Liberalization, Exchange Rate Changes and Tax Revenue in Sub-Saharan Africa. Journal of Asian Economics, 17:261–284, 2006. Richard Baldwin and Tony Venables. Regional Economic Integratiotion. Chapter 31 in Handbook of International Economics by Gene Grossmann and Kenneth Rogoff (eds.), pp 1597-1643, 1995. Thomas Baunsgaard and Michael Keen. Trade Revenue and (or?) Trade Liberalization. IMF Working Papers No 05/112, International Monetary Fund, Washington, DC, June 2005. Jagdish Bhagwati. The World Trading System at Risk. Princetion University Press, Princeton, New Jersey, 1991. 89
Jagdish Bhagwati, Richard Brecher, and Tatsuo Hatta. The Generalized Theory of Transfers and Welfare: Bilateral Transfers in a Multilateral World. American Economic Review, 73(4):606–618, September 1983. Arne Bigsten, Paul Collier, Stefan Dercon, Marcel Fafchamps, Bernard Gauthier, JanWillem Gunning, Jean Haraburema, Abena Oduro, Remco Oostendorp, Catherine Pattillo, Mans Soderbom, Francis Teal, and Albert Zeufack. Exports of African Manufactures: Macro Policy and Firm Behaviour. Journal of International Trade and Economic Development, 8(1):53–71, 1999. Arne Bigsten, Paul Collier, Stefan Dercon, Marcel Fafchamps, Bernard Gauthier, JanWillem Gunning, Abena Oduro, Remco Oostendorp, Catherine Pattillo, Mans Soderbom, Francis Teal, and Albert Zeufack. Do African Manufacturing Firms Learn from Exporting? Journal of Development Studies, 40(3):115–141, 2004. Nancy Birdsall. Underfunded Regionalism in the Developing World. Center for Global Development Working Paper No. 49, 2004. Fran¸cois Bourguignon, Alan Gelb, and Bruno Versailles. Policy, Aid and Performance in Africa: The G11 and other Country Groups. Mimeo, World Bank, Washington, DC, 2005. Ales Bulir and Javier Hamann. Volatility of Development Aid: From the Frying Pan into the Fire? in Peter Isard, Leslie Lipschitz, Alexandros Mourmouras, and Boriana Yontcheva (eds.), The Macroeconomic Management of Foreign Aid: Opportunities and Pitfalls, 2005. Mary Burfisher, Sherman Robinson, and Karen Thierfelder. Regionalism: Old and New, Theory and Practice. MTID Discussion Paper No. 65, February 2004. Massimiliano Cali and Dirk-Willem te Velde. The Potential Effects of Economic Partnership Agreements: What Quantitative Models say. ODI Briefing Paper 5, June 2006. 90
Celine Carrere. African Regional Agreements: Impact on Trade With or Without Currency Unions. Journal of African Economies, 13(2):199–239, 2004. Carlos Castel-Branco and Jose Sulemane. Country Study of the Macroeconomic Impact of Scaling Up Aid: Mozambique. Background paper for the DFID/ODI study on the scaling up of Aid, 2005. Lucio Castro, Christiane Kraus, and Manuel de la Rocha. Regional Trade Integration in East Africa: Trade and Revenue Impacts of the planned East African Community Customs Union. Africa Region Working Paper Series No. 72, World Bank, Washington, DC, 2004. Graciela Chichilnisky. Basic goods, the Effects of Commodity Transfers and the International Economic Order. Journal of Development Economics, 7:505–519, 1980. Jae Wan Chung. Utility and Production Functions. Blackwell, Oxford UK and Cambridge USA, 1994. Commission for Africa. Our Common Interest. London, 2005. Max Corden. Booming Sector and Dutch Disease Economics: Survey and Consolidation. Oxford Economic Papers, 36(3):359–380, November 1984. Max Corden and Peter Neary. Booming Sector and De-industrialization in a Small Economy. The Economic Journal, 92:825–848, December 1982. Angus Deaton and John Muellbauer. Economics and Consumer Behaviour. Cambridge University Press, 1980. Bernard Decaluw´e, John Cockburn, and V´eronique Robichaud. Fiscal Policy, Balance of Payments Constraints and External Shocks: A Regional CGE for the CACEU. Paper presented at AERC Conference, April 1994.
91
Bernard Decaluw´e, Yazid Dissou, and V´eronique Robichaud. Regionalism and Labor Market Structure: A CGE Analysis of UEMOA Customs Union. Journal of African Economies, 13(2):302–332, 2004. Shantayanan Devarajan, Jeffrey Lewis, and Sherman Robinson. Policy Lessons from Trade-Focused Two-Sector Models. Journal of Policy Modeling, 12(4), 1990. Shantayanan Devarajan, Jeffrey Lewis, and Sherman Robinson. External Shocks, Purchasing Power Parity, and the Equilibrium Real Exchange Rate. World Bank Economic Review, 7(1), 1993. Shantayanan Devarajan, Jeffrey Lewis, and Sherman Robinson. Getting the Model Right: The General Equilibrium Approach to Adjustment Policy. Unpublished Manuscript, May 1994. Avinash Dixit and Victor Norman. Theory of International Trade. Cambridge Economic Handbooks. Cambridge University Press, 1980. Sebastian Edwards. Exchange Rate Misalignment in Developing Countries. World Bank Research Observer, 4(1):3–22, 1989a. Sebastian Edwards. Real Exchange Rates, Devaluation and Adjustment: Exchange Rate Policy in Developing Countries. MIT Press, 1989b. Sebastian Edwards. Tariffs, Capital Controls and Equilibrium Real Exchange Rates. The Canadian Journal of Economics, 22(1):79–92, February 1989c. Sebastian Edwards and Sweder van Wijnbergen. Tariffs, the Real Exchange Rate and the Terms of Trade: On Two Popular Propositions in International Economics. Oxford Economic Papers, 39(3):458–464, September 1987. Ibrahim Elbadawi. The Impact of Regional Trade and Monetary Schemes on Intra-sub-
92
Saharan Africa Trade. in A. Oyejide, I. Elbadawi and P. Collier (eds.), Regional Integration and Trade Liberalization in sub-Saharan Africa, 1997. Ibrahim Elbadawi. External Aid: Help or Hindrance to Export Orientation in Africa? Journal of African Economies, 8(4):578–616, 1999. Robert Hamilton and John Whalley. Geographically Discriminatory Trade Arrangements. Review of Economics and Statistics, 67(3):446–455, 1985. IMF. Rwanda: Selected Issues and Statistical Appendix. Country Report 04/383, IMF, December 2004. IMF. The Macroeconomics of Managing Increased Aid Inflows: Experiences of LowIncome Countries and Policy Implications. IMF, Washington, DC, 2005. Harry Johnson. The Possibility of Income Losses from Increased Efficiency or Factor Accumulation in the Presence of Tariffs. Economic Journal, 77:151–254, 1967. Barsha Khatty and Mohan Rao. Fiscal Faux Pas?: An Analysis of the Revenue Implications of Trade Liberalization. World Development, 30(8):1431–1444, 2002. Tony Killick and Mick Foster. The Macroeconomics of Doubling Aid to Africa and the Centrality of the Supply Side. Development Policy Review, 25(2):167–192, 2007. Aart Kraay. Exports and Economic Performance: Evidence from a Panel of Chinese Enterprises. Revue d’Economie du D´eveloppement, 2:183–207, 1999. Dirk-Willem te Velde Lauren Philips and Sheila Page. Aid for Trade: What does it mean? Why should aid be part of WTO negotiations? And how much might it cost? ODI Opinions 61, December 2005. Nuno Limao. Preferential Trade Agreements as Stumbling Blocks for Multilateral Trade Liberalization: Evidence for the US. CEPR Discussion Paper No. 4884, January 2005.
93
Kiminori Matsuyama. Agricultural Productivity, Comparative Advantage, and Economic Growth. Journal of Economic Theory, 58(2):317–334, December 1992. Mark Melitz. The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity. Econometrica, 71(6):1695–1725, 2003. Chris Milner, Oliver Morrissey, and Andrew McKay. Some Simple Analytics of the Trade and Welfare Effects of Economic Partnership Agreements. Journal of African Economies, 14(3):327–358, 2005. MINECOFIN. A Profile of Poverty in Rwanda. Technical report, Ministry of Finance and Economic Planning, February 2002. MINECOFIN. 3rd General Census of Population and Housing of Rwanda: Statistical Tables. Technical report, Ministry of Finance and Economic Planning, November 2003. Benno Ndulu and N Ndungu’u. Trade and Growth in sub-Saharan Africa. Paper presented at the IMF-AERC Seminar on Trade Reforms and Regional Integration in Africa, Washington DC, December 1-3, 1997. Francis Ng. Trade Data. accessed via www.worldbank.org/trade on 20 March, 2007. Mwanza Nkusu. Aid and the Dutch Disease in Low-Income Countries: Informed Diagnosis for Prudent Prognoses. IMF Working Papers No 04/49, International Monetary Fund, Washington, DC, March 2004a. Mwanza Nkusu. Financing Uganda’s Poverty Reduction Strategy: Is Aid Causing More Pain Than Gain?
IMF Working Papers No 04/170, International Monetary Fund,
Washington, DC, 2004b. OECD.
Aid dropped in 2006 for the first time since 1997.
http://www.oecd.org/dataoecd/56/51/38348577.htm on 9 May, 2007.
94
accessed via
Arvind Panagariya. Preferential Trade Liberalization: The Traditional Theory and New Developments. Journal of Economic Literature, XXXVIII:287–331, June 2000. Walt Whitman Rostow. The Stages of Economic Growth: A Non-Communist Manifesto. Cambridge University Press, Cambridge UK, 1960. Kamal Saggi. Preferential Trading Arrangements and Multilateral Tariff Cooperation. Southern Methodist University Working Paper, 2004. Paul Samuelson. The Transfer Problem and Transport Costs: The Terms of Trade when Impediments are Absent. Economic Journal, 62:278–305, June 1952. Khalid Sekkat and Aristomene Varoudakis. Exchange Rate Management and Manufactured Exports in sub-Saharan Africa. Journal of Development Economics, 61:237–253, 2000. Christopher Stevens. Economic Partnership Agreements: Where we are. ODI Briefing Paper 4, Overseas Development Institute, London, UK, June 2006. Mark Sundberg, Hans Lofgren, and Fran¸cois Bourguignon. Absorptive Capacity and Achieving the MDGs: The Case of Ethiopia. in Peter Isard, Leslie Lipschitz, Alexandros Mourmouras, and Boriana Yontcheva (eds.), The Macroeconomic Management of Foreign Aid: Opportunities and Pitfalls, 2005. Peter Timmer. The Agricultural Transformation. Chapter 8 in Handbook of Development Economics by Hollis Chenery and T.N. Srinivasan (eds.), pp 1481-1533, 1988. Stephen Tokarick. Immiserizing Foreign Aid: The Roles of Tariffs and Nontraded Goods. IMF Working Paper 06/129, International Monetary Fund, Washington, DC, May 2006. Ragnar Torvik. Learning by Doing and the Dutch Disease. European Economic Review, 45(2):285–306, February 2001. 95
UN Millennium Project. Investing in Development: A Practical Plan to Achieve the Millennium Development Goals. Oxford University Press, 2005. Sweder van Wijnbergen. The Dutch Disease: A Disease After All? The Economic Journal, 94:41–55, 1984. Sweder van Wijnbergen. Aid, Export Promotion and the Real Exchange Rate: An African Dilemma? CPD Working Paper 54, World Bank, Washington, DC, 1986. Sweder van Wijnbergen and Sebastian Edwards. Disequilibrium and Structural Adjustment. Chapter 28 in Handbook of Development Economics by Hollis Chenery and T.N. Srinivasan (eds.), pp 1481-1533, 1989. Tony Venables. Winners and Losers from Regional Integration Agreements. The Economic Journal, 113:747–761, October 2003. Jacob Viner. The Customs Union Issue. Carnegie Endowment for International Peace, New York, 1950. Alan Winters. Regionalism vs. Multilateralism. in R. Baldwin, D. Cohen, A. Sapir and A. Venables (eds.), Market Integration, Regionalism and the Global Economy, 2000. World Bank. The East Asian Miracle: Economic Growth and Public Policy. Oxford University Press, 1993. World Bank. Global Economic Prospects 2005: Trade, Regionalism and Development. World Bank, Washington, DC, 2004. World Bank and IMF. Global Monitoring Report. World Bank, Washington, DC, 2004. Yongzheng Yang and Sanjeev Gupta. Regional Trade Arrangements in Africa: Past Performance and the Way Forward. IMF Working Paper 05/36, International Monetary Fund, Washington, DC, February 2005.
96
Makoto Yano and Jeffrey Nugent. Aid, Non-Traded Goods, and the Transfer Paradox in Small Countries. American Economic Review, 89(3):431–449, 1999.
97
FIGURE A.1: THE MARKET FOR THE REGIONALLY TRADABLE GOOD ‘HOME’ Has a cost-disadvantage in producing regional goods vis-à-vis `Foreign’. It sets an ad-valorem tariff at rate τ. If τ goes down, imports from `Foreign’ go up.
a
b
c *
P
d
R
e
f
P*R k *
p R =(1-φτ)P
x
REGIONAL INTEGRATION Two markets are fully integrated with the equilibrium price in between the two old prices.
(1+τ)P*R
g
αRIR
‘FOREIGN’ Has a cost-advantage in producing regional goods vis-à-vis `Home’. When τ goes down, exports to `Home’ will rise and pR is dependent on how high `Home’ sets τ.
R
v
y βEIR
m
l
P*R
(1-β)EIR
w (1-α)RIR EIR = RIR
`HOME’ y-x = `Home’ net imports from `Foreign’ pre-integration I βE R - αRIR = `Home’ net imports from `Foreign’ post-integration αRIR = `Home’ share of regional good production under free trade βEIR = `Home’ share of regional good consumption under free trade
`FOREIGN’ w-v = `Foreign’ net exports to `Home’ pre-integration (1-α)RIR -(1- β)EIR = `Foreign’ net exports to `Home’ post-integration (1-α)RIR = `Foreign’ share of regional good production under free trade (1-β)EIR = `Foreign’ share of regional good consumption under free trade
Table 5.1: Simulation Results for Main Experiments
