Macroeconomic Models for Developing Economies: SAMs and CGE

A. Macroeconomic Models for Developing Economies: Some General Considerations

It is well known that the developing economies have special features that need to be recognized. Below, I first discuss some of these aspects of developing economies from the macroeconomic modelling perspective. Later in this section, a brief discussion of some "micro" institutional features that are also relevant for poverty reduction strategies are mentioned in order to round out the discussion. From the macroeconomic side, the following points are important⁶:

1. First, there must be an accounting framework and behavioral equations capturing some key aspects of macroeconomic modelling for developing countries. The most straightforward way of giving economic content to a set of aggregate accounting relationships is by adding appropriate behavioral equations and equilibrium conditions.
2. The accounting relationships that are relevant for a particular case depend on the structure of the economy. There could, for example, be a) "benchmark" accounting framework; b) particular features, modelling aspects such as alternative choices of disaggregation of production and consumption, structural features of labor market, degree of development of the financial system etc. c) behavioral functions, liquidity constraints on aggregate consumption; credit and foreign exchange rationing, debt overhang and its effects on production and private investment uncertainty and irreversibility effects on investment decisions; effects of financial repression, currency substitution, and informal financial markets on money demand etc.
3. Fiscal, monetary and exchange rate policies in developing countries are important features in most models. Data must cover a wide range of variables including industrial output, prices, wages, various monetary aggregates, domestic private sector credit, fiscal variables, exchange rates and trade variables.
4. Nature and implications of fiscal rigidities and the effect of fiscal deficits on a variety of macroeconomic variables are also important.
5. Developing country fiscal problems require special attention. Some of the most important features here are: high tax rates levied on a narrow tax base and heavy reliance on revenues from financial repression and multiple currency practices, on the inflation tax, and on excessive debt financing.
6. Exchange rates modelling require special attention as well. In particular, one needs to identify special features such as "fixed" with rationing and simultaneous transactions in parallel markets when these are present. Credibility and inflation under a fixed exchange rate regime also need particular attention. The association of quasi-fixed exchange rates with currency and financial crises makes this issue specially significant. Contractionary effects of devaluation may also destabilize the economy, and could be included as a theoretical possibility that may sometimes become a practical problem.
7. The role of labor markets in the context of short-run macroeconomic adjustment in developing countries is particularly important for analyzing the poverty reduction implications of macroeconomic policies. More specifically, labor market segmentation and sectoral wage rigidity need special attention.
8. It should also be recognized that by now there are both orthodox and "heterodox" programs and models of structural adjustment in developing economies. Alternative models of inflationary process are also available. However, the approach I have adopted here is intended to skirt unnecessary terminological (and at times, ideological) controversy. This 'ecumenical approach' --- to use Sherman Robinson’s felicitous term --- adopted here is more concerned with the real contents of the models and their real world policy relevance.
9. Macroeconomic dynamics associated with monetary and exchange rate policy rules in a context where international capital mobility is imperfect need to be emphasized.
10. Three important issues that models must focus on in the context of exchange-rate based disinflation programs in developing economies are: i) output, ii) interest rates, iii) real wages.
11. It should be pointed out that with humility that none of the modelling approaches that are widely used in developing countries is at present able to adequately address the complex dynamic interactions between stabilization, growth and distribution. This makes the intermediate and longer-term analysis of issues related to external debt, capital inflows, and currency crises particularly difficult.

    Trade and financial liberalization and macroeconomic performance likewise become issues where the dynamic aspects are often treated simplistically. Problems of short-run macroeconomic management during the liberalization process are also well known and need little commentary.

12. Political factors in the adoption and abandonment of stabilization and structural adjustment programs in developing countries effects are also of obvious importance, but are very difficult to incorporate in the standard macroeconomic models of applied general equilibrium variety. For example, it would obviously be important to include the effects of the presidential and parliamentary electoral cycle on the pattern of public spending in many Asian and Latin American countries. One could also make the same case for including an analytical framework for examining the linkage between exchange rate policy and electoral cycles.

Some questions that are relevant rise in light of the features discussed above are:

1. What structural changes need to be preceded by macroeconomic stabilization? Or, alternatively, can the two proceed concurrently?
2. What is the proper sequencing of the liberalization and reform measures?
3. What “structural” differences between developed and developing economies, and “structural” similarities among the latter. Are relevant to model?

In response to the third question above, the following shared structural characteristics may be important:

1. Many agents possess significant market power Macroeconomic causality in developing countries tends to run from "injections" such as investment, exports, and government spending to "leakages", such is imports and saving;
2. Money is often endogenous
3. The structure of the financial systems can influence macroeconomic outcomes in important ways
4. The role of imported intermediate and capital goods as well as direct complementarity between public and private investment are empirically important.

Partly as a consequence of these features some have questioned the wisdom and efficiency of orthodox short-run macroeconomic policy prescriptions, particularly "shock treatment" in the form of fiscal austerity coupled with devaluation and tight monetary policy.

Disagreements among modelers also exist with respect to the identification of the source of inflation. The key controversy is about whether one should ascribe an accommodative rather than a causal role to money supply growth. According to the nonmonetarist view frequently the source of inflation is slow relative productivity growth in agriculture (arising from poor land distribution and land tenure patterns) combined with administered prices (arising from noncompetitive market procedures and implying downward price rigidities) in industry, together with wage indexation. Monetary policy is perceived to be passive in the face of these already pervasive inflationary forces. Moreover, in part because of the roles of working capital and imported inputs, and in part because substitution possibilities are more limited than assumed by the proponents of orthodox macroeconomic management, a policy package combining devaluation with tight fiscal and monetary policies will result in stagflation in the short-run with little or no improvement in the external accounts. The alternative new structuralist policy prescription is not always clear, but it would in all likelihood contain a greater element of gradualism, direct intervention, and employ many of the means of medium term resolution of structural problems that are contained in traditional stabilization programs.

For the sake of parsimonious modelling, quite often a three good modelling approach is adopted. The three aggregated goods are non-traded domestic good, exportable good, and importable good.⁷Here, too, some important differences between the developed and developing countries need to be kept in mind. For example,

1. Developing economics, like small industrial countries, tend to be much more open to trade in goods and services than are the major industrial countries. In 1995 trade share of developing countries was 45% compared to G-7s trade share of 25 percent.
2. Developing countries typically have little control over the prices of goods they export and import. In particular, they often face exogenous terms of trade.
3. Over half of the exports typically consist of agricultural and primary commodities. Such an export structure needs to be modeled explicitly. The Mundell-Fleming model which has long been the work-horse of open economy industrial country model, assumes endogenous terms of trade determination, with the domestic economy completely specialized in the production of a good over which it exerts significant market power. The production structure most suitable for the analysis of developing country macroeconomic phenomena is instead likely to be the Salter-Swan dependent economic model or (as mentioned before) a three good model consisting of exportables, importables, and nontraded goods.

Such a production structure permits a distinction to be drawn between the exogenous terms of trade and an endogenous real exchange rate, which is the central intertemporal macroeconomic relative price in these economies.

In terms of the exogenous prices faced by the typical developing economy, both oil and non-oil commodities prices fluctuate a great deal. The extent of external trade in assets have tended to be more limited in developing countries than in developed countries although this situation has recently begun to change in dramatic fashion for an important group of developing economies. The resulting instabilities however have also caused serious dislocations. In particular the increase in poverty in the affected Asian economies after the Asian Financial Crisis from July 1997 on should be kept.

In particular, the macroeconomic consequences of pegging, of altering the peg (typically devaluation) and of the rules for moving the peg are of particular importance in macro-modelling in developing countries. It is also useful to remind ourselves in trying to model the financial sectors that financial markets in many developing countries have long been characterized by the prevalence of rudimentary financial institutions. This is of particular relevance in analyzing the impact of policies on poverty reduction in low-income countries.

In light of the above, it should be apparent that in the modelling of these economies some macro-behavioral relationships may need to be modified. For example, we may need to incorporate the implications of credit and foreign exchange rationing in private decision rules where such rationing is present. This will affect, for instance, private consumption, investment, asset demand, export supply and import demand functions.

Some Relevant Aspects of Public Sector Behavior

The Government Budget is another important segment of a macro-model requiring careful handling. In particular, we need to remember that the composition of the government budget differs markedly between industrial and developing countries. Pervasive role of the state in many developing economies is reflected through the following factors, among others:

1. nonfinancial public sector – central government, local governments, specialized agencies, and nonfinancial public enterprises;
2. financial institutions owned by the government; the central government absorbs a smaller fraction of output of developed countries than in developing countries;
3. the composition of spending differs between the two groups of countries. Developing countries spend proportionately more of their budget on general public service, defense, education and other economic services. Developed countries spend more on health and substantially more on social security.
4. Revenue: tax collection is hindered by limited administrative capacity and political constraints. This means that direct taxation plays a much more limited role than in developed countries. Direct taxes, taxes on domestic goods and services, and taxes on foreign trade account for roughly equal shares of total tax revenue in developing countries; in industrial countries income taxes account for the largest shares and taxes on foreign trade are negligible. In developing countries, the share of tax revenue raised from individuals is much higher than corporate income tax.
5. greater reliance on seigniorage (change in base money stock divided by nominal GDP). Seigniorage and inflation are positively related.

Three other dimensions of the budget institutions that have relevance for both growth and poverty reduction have been much discussed recently:

1. The nature and credibility effects of the constitutional rules that can be implemented to impose constraints on the size of the fiscal deficit, e.g., the balanced budget rule
2. The procedural rules that guide the articulation and elaboration of the budget by the executive branch, its approval by the legislative branch, and its execution.
3. The type of rules (whether collegial or hierarchical) that may enhance the transparency of the budgetary process e.g., Debt/ GDP upper limit constraint.

Aggregate Supply and Labor Markets: some further issues

Aggregate supply and the labor market are aspects that need some further attention before we close our discussion of institutional and macroeconomic aspects of macro-modelling. Here it is important to point out that through the cost of intermediate inputs that are imported; the exchange rate has an important influence on the position of the economy’s short-run supply curve (SRSC).

SRSCs in developing countries may be significantly affected by working capital considerations. Many have claimed that costs of working capital tend to give interest rates and credit availability an important short-run supply-side role, although this is controversial and the empirical evidence is mixed.⁸

Although labor market institutions vary substantially across developing countries, the informal sector continues to play an important role in the determination of wages and employment in many of them. The modelling of short-run wage-setting behavior represents one of the key differences between some of the major schools of modern macroeconomics, but most participants in the disputes acknowledge that country-specific institutional differences (such as the prevalence of staggered overlapping contracts in the U.S. or synchronized wage bargaining in Scandinavia) are important in determining the economy’s SRSC. In this context, the role of economy-wide backward indexation mechanism in the context of disinflation programs has been studied extensively. Developing countries, as is well known, often have disguised unemployment. What is less well known is the prevalence of flexibility in many of the developing country labor markets as well. It would appear from the available evidence that many developing country labor markets have a high degree of real wage flexibility (Horton et al., 1994). Thus, for proper modelling of these markets in developing economies, a properly nuanced mix of flexibility and rigidity in specific labor markets is called for, rather than following one specific characterization for all labor markets.

As a result of the foregoing, the macroeconomic environment in developing countries is often much more volatile than that in industrial countries. The fundamental causes of the macroeconomic instability in developing countries are both external and internal. Small developing countries are price takers in the international markets for goods and services as well as financial assets. Therefore, these countries are directly affected by volatility in international markets. If we add to this the inflexibility and paucity of domestic macroeconomic instruments and we then face get a situation that is not easily amenable to control. There is also political instability in many countries resulting in frequent jumps in policy regimes. Such regime switch in a weak institutional environment creates the unfortunately typical developing countries scenario of a macroeconomic trajectory punctuated by a series of crises.

To sum up, economic boom and bust are much more prevalent in developing countries than developed countries. Such a history of macroeconomic volatility has serious economic costs-sometimes reaching into double-digit percentage points (See Khan, forthcoming 2004).

The above discussion is intended to give a fair summary of what is special about development macroeconomics. However, in order to link poverty analysis to the macromodels, more explicit recognition of the nature of market imperfections and of informal institutions that arise to fill the gaps is necessary. The literature in this area has experienced a tremendous explosion drawing on advanced work in game theory and the economics of information.⁹ These new approaches try to explain empirical institutional features that include the following (Mookherjee and Ray, 2001).

1. Fragmented credit markets;
2. Segmented labor markets;
3. Lack of market clearing manifested in unemployment and credit rationing;
4. Co-presence of different types of contracts, e.g., tenancy contracts of both fixed rent and share cropping varieties;
5. Pervasive long-term relationships between borrowers and lenders, employers, and employees, or farmers and traders;
6. Dual labor markets in which some workers enter into long-term contracts while others are employed to carry out similar task without such contracts at a lower level of wages;
7. Interlinked transactions and exclusive dealing between specific groups of agents across many markets for instance, credit and tenancy may be bundled together. Likewise, credit may also be bundled with employment or marketing contracts;
8. Asset ownership is the key to access to credit, tenancy or employment markets. Thus the poor have limited or no access to credit because they lack collateral assets. The poor also have limited or no access to employment owing to malnutrition, debilitating diseases or low levels of human capital.
9. Small farms show higher yields even when the large farms have better access to credit and technology.
10. Some markets such as the market for land sales are quite thin, leading to the persistence of tenancy and unequal land ownership in spite of the superior productivity of owner cultivated small farms.
11. Informal cooperatives and kinship networks are significant determinants of access to credit, insurance, technological information, water and common lands.

As Stiglitz (1994) and others have pointed out, the standard Arrow-Debreu model with a complete set of markets and optimizing agents cannot explain these phenomena. However, models using game theory and the approach of information economics largely pioneered by Stiglitz have amassed an impressive analytical record in explaining these features. While macromodels cannot be expected to accommodate all these features, in detail, at least the labor and credit markets need to be modeled carefully. This point is beginning to be recognized by development macroeconomists of virtually all persuasions (Agenor and Montiel, 1999).

In surveying the macromodels attempting to link macropolicies to poverty reduction, we will need to ask how well some of these features are modelled in particular instances. Before turning to a discussion of some relevant economy wide CGE models, it is useful to discuss the economy wide data base for such models in the form of Social Accounting Matrices (SAMs). In the following sub- section, I present a brief discussion of the relevant issues and a particular case of SAM-based modelling as a background to the flexible price CGE models discussed from the following section on.

B. Social Accounting Matrices as Consistent Economy wide Data Bases and Fixed Price Multipliers

In this section the Social Accounting Matrix is presented as a data gathering framework as well as an analytical tool for studying the effects of various macroeconomic policies as well as the impact of sectoral growth on poverty alleviation. The origins of social accounting can be traced as far back as Gregory King’s efforts in 1681, but more recent work stems from the attempts by Richard Stone, Graham Pyatt, Erik Thorkbecke and others. ¹⁰

In the methodological framework of this particular study of CGE models, the SAM is viewed as a tool for mapping production and distribution at the economy wide level. In this sub-section, first a general SAM is described. Then it is shown how the method for studying the short-run effects of economic growth within this framework follows logically from its structure. The model used is a simple version of a class of SAM-based general equilibrium models.¹¹ It summarizes succinctly the interdependence between productive activities, factor shares, household income distribution, balance of payments, capital accounts, etc. for the economy as a whole at a point in time. Given the technical conditions of production the value added is distributed to the factors in a determinate fashion. The value added accrued by the factors is further received by households according to their ownership of assets and the prevailing wage structure. In the matrix form the SAM consists of rows and columns representing receipts and expenditures, respectively. As an accounting constraint receipts must equal expenditures.

As is elaborated further in Khan and Thorbecke (1988), the SAM framework can be used to depict a set of linear relationships in a fixed coefficient model. For deciding the question of determination, the accounts need to be divided into exogenous and endogenous ones. For instance, in the South African SAM used by Khan (1989) to analyze the impact of economic sanctions on the South African economy, there are three endogenous accounts. These are factors, households and production activities, leaving the government, capital and the rest of the world accounts as exogenous.¹²

In examining the poverty profiles in any country, one particular set of accounts assume special importance. These are the household accounts. The proper flow of income and expenditures need to be recorded for these accounts if an accurate picture of poverty as inadequate income/ consumption is to emerge out of a given SAM. For this reason, the classification of households needs special care. There are at least six aspects that need careful attention.

These six aspects are:

1. to classify households by socio-economic characteristics;
2. to understand the income generation process by which the households receive their incomes;
3. to pinpoint the distributional mechanisms;
4. to understand the household consumption patterns;
5. to link household income and consumption to social capabilities and functionings; and
6. to estimate the resource generating capacity and resource absorbing capacity of the households.

If items 1-6 can be investigated systematically by combining economic and social modes of inquiry in a SAM, proper policy intervention for poverty reduction will become a more tractable exercise than it is at present. In particular, if disaggregated SAMs can be constructed at the local, sub-national levels, then intervention at the local levels may be much more effective than it has been historically in many cases. This is yet to be realized, but clearly is an important goal to pursue. I now turn to a discussion of another particular strength of the SAM framework for data gathering. SAMs have the consistency features that one needs in capturing economic flows for use in a general equilibrium framework.

The following tables illustrate in the aggregate the consistency requirements for building a SAM.

See Table 1. SAM-Format OF SNA-Aggregates, Kenya, 1982 [PDF 88kb | 1 page]

See Table 2. Modular omposition of the SAM [PDF 67kb | 1 page]

In terms of the usefulness of the SAM information base, one can argue that not only is the National SAM a tool for the overall poverty reduction analysis, perhaps even more importantly, the building of local and regional SAMs will help the field-worker to understand the interrelations between households characteristics, the immediate causes of poverty and the best way to help specific types of households out of poverty. I now turn to the discussion of a particular type of modelling exercise that can be carried out with both the national and regional SAMs.

Fixed Price Multipliers for National and Regional SAMs

In what follows, a national framework with distinct regions where the poor may be located is assumed. Suppose there are n regions indexed by i = 1, 2, ......., n. For each region i, there are intra-regional transactions as well as inter-regional transactions. Then, the national SAM can be disaggregated into 'n' Regional or RSAMs. The typical RSAM for region i can be schematically described as in Table 3[PDF 47kb | 1 page]. Table 4[PDF 87kb | 1 page] divides up the regional accounts according to whether these are endogenous or exogenous for the purpose of modelling.

The above SAM framework can be used to depict a set of linear relationships in a fixed coefficient model. This is the essential point behind fixed price multiplier modelling approach based on a SAM. For deciding the question of determination of the equilibrium quantities, the accounts need to be divided into exogenous and endogenous ones as in Table 4[PDF 87kb | 1 page] below.