Global models

Global models

403 GLOBAL MODELS A review of recent developments Sam Cole This article summarizes recent developments in computerized global modelling. It is lar...

2MB Sizes 0 Downloads 1 Views

403

GLOBAL

MODELS

A review of recent developments Sam Cole

This article summarizes recent developments in computerized global modelling. It is largely descriptive, and will be followed in further issues by articles dealing with more qualitative scenario approaches to global forecasting and an overall evaluation of current forecasting efforts. to describe the ‘Limits’ studies ( World Dynamics and Limits to as the first generation of global models and the studies which more or less immediately followed sponsored by the Club (such as the [email protected] forsurvival- later the World Integrated Model, Latin American and Japanese Club of Rome models) and as a reaction to the CoR studies (such as Leontief s The Future of the World Economy published by the United Nations, the UK government’s SARUM and the World Bank models) as second generation. Similarly, those models which appeared in the wake of critiques of these studies (such as the UN models in UNCTAD, UNITAR and DIESA, and the GLOBUS and Global 2000 studies) could be termed third generation models. However, because of the way the field has developed, such easy divisioes are misleading, especially if global studies are categorized in this way. Some global modelling exercises predate Limits to Growth or draw on studies of the 1960s and before. These various traditions have merged over the decade, and also within the various individual projects the emphasis attached to the various approaches has changed. Most models are regularly updated or spin off several variants (as with the SARUM and WIM models), Some models undergo a name change but show little change in content. The models reviewed and acronyms used in this article are listed in Table 1, which also shows the vintage of the models, ie when they were built or when their results were published (although the dates indicated are at best approximate). A bibliography of publications used is given at the end of the article. IT IS COMMONPLACE

Growth) of the Club of Rome (CoR)

An overview

of trends

in modelling

The original systems models (World Dynamics and the more detailed Limits to Growth) were executed and published by 1972; although the model has not Professor Cole is at the State University of New York at Buffalo, School of Architecture and Environmental Design, Hayes Hall, 3435 Main Street, Buffalo, NY 14214, USA. The authoracknowledges funding from the Swedish Parliamentary Committee for Futures Studies, who are not responsible for the opinions expressed.

FUTURES August 1987

~1~3287/87104~0~2~3.~~

1987

Butterworth&Co(Publishers) Ltd

404

Globalmodrls

TABLE 1. PRINCIPAL PROJECTS REVIEWED Study

Vintage

Principal publication

1970-75 1972-86 19761976-84 1972-86

Limits to Growth Strategy for Survival Latin American: Catastrophe or a New Society The future of the World Economy Future of Global Interdependence

or institution

Sys terns models

Limits WIM LA Leontief FUGI

Institutional models 1972-86 SARUM 1977-84 Global 2000 1975SIMIGDP

Systems Analysis Research Unit Report to the President System for Modelling Global Development Processes

World Bank UNIDISEA UNCTAD

1973-86 1968-86 1970-81

World Bank Development Report United Nations World Development Outlook Trade and Development Report

Deinstitutionalized UNITAD UNITAR

models 1976-86 1977-86

UNCTADIUNIDO Joint Project Project on the Future

Politico-economic SIMPEST GLOBUS

models 1975-86 1980-86

Simulation of Political Economic Strategic Interactions Generating Long- term Options by Using Simulation

Modelling consortia LINK 1988-86 INFORUM 1970-86 G-MAPP 1980-86 UNU 1982-85

Linked International Econometric Models International System of Input-Output Models Global Models and the Policy Process United Nations University

Note: The dates given cover the period when the model is being developed and updated and published.

developed further, some of the Limits philosophy has carried over into other studies, in particular the Global 2000 project and to a lesser degree the Soviet Global Development Processes model. The so-called second generation systems models were begun and completed by 1975. Unlike the first models, some of the second generation models (SARUM, FUG1 and Strategyfor Survival) have been systematically modified up to the present day. Unlike the earlier independently funded CoR models, the SARUM and FUG1 models were supported at first by national governments. However, other global modelling exercises such as project LINK were already under way in the late 1960s. The modelling programmes of the DRPA (then the CDPPP) and the World Bank also had started on a modest scale by the late 1960s and early 197Os, using simple econometric (‘two-gap’) models. During the 1970s and with the impetus of the Limits exercise, these developed with attempts to build rather detailed into large-scale modelling exercises, multi-country input-output and econometric models and also models of the international trade system. After 1980 these ‘institutionalized’ modelling programmes contracted considerably. A number of other global modelling projects were sponsored by national governments and within the UN system in the mid-1970s. When the climate for modelling changed in the early 198Os, some of these models became de-institutionalized, and now operate on a more independent basis (in universities and FUTURES August 1987

Globalmo&ls 405

research institutes). If the rise of institutional modelling was in good part stimulated by other global modelling exercises and an attempt to overcome the flaws in earlier models and to set up a reliable medium-term policy model, its decline appears to be partly a result of the failure of the models to meet this technical objective, but more a result of the changing policies of these institutions. Politico-economic global models (eg GLOBUS) are the most recent innovation, although they too come out of a political science tradition begun in the 1960s. Primarily these studies focus on the dynamics of the East-West arms race. Since 1980, partly because of curtailed funding and partly due to the need for peer group reinforcement, all projects have undergone a period of retrenchment or consolidation. Typically, consolidation has involved setting up consortia of ‘established’ models, sharing data and comparing (and also sometimes sharing) results. This process involves models which are still institutionalized as well as the second generation models such as FUG1 and SARUM and the third generation models (such as UNITAD and UNITAR), all ofwhich now function Some models (notably SARUM) have been on a more independent basis. involved in several consortia, whereas a smaller number of projects (such as GLOBUS) have maintained an isolated existence. To some degree, the situation can be characterized as one of inbreeding, cross-breeding and survival through a time of scarcity of research funds. Because of the way the various models have adapted, distinctions between generations are less important than the differences in their objectives, which determine the content, size, structure, behavioural assumptions, the way the models are used as well as the calculations and future alternatives explored with the models. Several trends in global modelling appear to be discernible-they increasingly focus on economic variables, they forecast a shorter timespan and they are getting bigger. Table 2 suggests a tendency for these computerized studies to concentrate on ‘economic’ issues (primarily comparative growth rates and economic structures) and to head for much greater detail (all major countries and sectors individually recognized and key behavioural assumptions endogenized). Unlike the Limits study, which was more truly ‘global’ in that it dealt simultaneously with ecological, demographic, social and economic variables, as well as dealing with the whole world, contemporary models are quite limited. Demographic trends are usually taken directly from UN statistics, and not related to economic trends. The reason for this trend is partly a belief that omitting the less secure areas of knowledge and providing extra detail will make the results of these models more convincing (or at least less open to criticism) and more relevant to specific policy. But this shift in emphasis also reflects a perceived trend in the way issues are interpreted and prioritized by academics, government and the general public. ‘Environment’, for example, was redefined as an aspect of ‘development’ , in turn to be viewed in terms of relative growth rates. Other novel concepts too, such as that of ‘basic human needs’ employed in the Latin American model, were redefined in such a way as to be dealt with in terms of conventional variables and theories. FUTURES August 1987

%

B

X

i

SIMPEST GLOBUS

LINK INFORUM G-MAPP

X X X

X X

X X

X X

X X

X

X

X X

X

X

X

X

X

X

xx

X

X

r

r

X

Ev

xx

X

X

X

U

a

X

X

Ed

X X

X

L

X

X

r

r

H

X

X

X X

X

X

X

X

X

X

X

X

X

X

X

X

i a

X

x i

X

i

X

I

X

X

X

X

X

X

W

W

W

X

X

i

W

i

X

W

W

X W

B

X

X/i X

l-r

X

T

X

X

a

I

X

X

M

X

i

X

X

G

International

:: P

f P

f B

f f P

f/P o/P Q

P

f I

P o/P f

M

(Xl

(Xl

X

X

X

F

Method of solution-p

Importance of feature-x

price mechanism, o optimization,

f fixed proportion, w world trade pool, I linear programming,

included, r recursive only, X major feature, a accounted for, i introduced exogenousiy. g general equilibrium.

- - Notes: Subsystems incorporated-Ec (Economy), D (Demography), A (Agri~uiture), R (Raw Materials), En (Energy), Ev (Environment), iJ (Urbanization), Ed (Education), H (Household distribution), L (Domestic politics), G (Government sector), M (Military sector), I (international politics), T (International trade), ll (Terms of trade), B (Bilateral trade), M (international markets), F (international finance).

i

i

(x)

i

i

i

X

X

X X

X X

X X

i

X

X

X X

X X

i

X

X

X

X X

II

X

X X

X X X X

X X X

En

R

A

0

UNITAD UNITAR

iMF3 UN/DIES& IO GEM WTM UNCTADlSlGNA

World Bank: SIMLINK WDR

SARUM Global 2000

X

Limits WIM LA Leontief FUGI: GIOM GMEM

X

EC

Study

lntrareglonal

TABLE 2. SUBSYSTEMS DETAIL IN MODELS

By the late 1970s too the world ‘crisis’ was reduced to an ‘economic’ phenomenon, at least by agencies able to support global modelling. In addition, many of the more imaginative proposals for tackling long-term international reform and environmental issues promoted in the 1970s became passe’ as most governments struggled with the more immediate symptoms of crisis. With an emphasis worldwide on international markets and national competitiveness in the past decade, it has been a short step for models to become increasingly devoted to these issues. Shift

in aim

On b&nce, global modelling appears to have shifted from being an attempt to describe the long-run evolution of the global system to an attempt to describe the present global economy. There are several approaches to this endeavour, principally input -output, econometric models and general equilibrium modelling. Input-output models emphasize the structure of transactions within the production side of an economy at a given point in time, whereas econometric models emphasize behavioural relationships of an economy in the light of statistically observed historical behaviour. In practice, the two approaches have considerable overlap; some projects maintain two distinct models, whereas the majority have an input-output structure at their core. The so-called two-gap models are used to calculate investment requirements to meet a given growth rate (or vice versa): by contrast, general equilibrium models model all market behaviour in order to determine all behaviour endogenously. Politico-economic models introduce additional equations representing some non-economic behaviour (such as measures of international tension or militarization). The modelling techniques are shown in Table 3. There appears also to be a trend to enormity. Certainly the size of models (ie the number of sectors, countries, actors and complexity of behaviour) also varies considerably. Some indication of this is given in Table 4. Models intended to project mid-term trends, and to be more or less directly relevant to policy, tend to be very large (up to 50 countries and sectors). This level of detail is felt necessary in order to make the results ‘policy relevant’ and so provide the precision needed for the individual sectors and nations. The time horizon for forecasts appears to have shortened. No other models have extrapolated to the year 2 100 as did the Limits models. Most models now appear fixated on the end of the millennium as a target. Even so, the LINK exercise has always focused on the short run, while other current models are used in the context of long-run scenario exercises to forecast up to 2025. Perhaps the most important question is whether the forecasts themselves are changing--’ m particufar, whether the long-term outlook and prospects for improvement as a consequence of changes in policy are seen to be improving or worsening. There are several problems in assessing this: models differ in their structure, applications, forecasting periods, what is meant by a ‘trend’ and also the policies and policy objectives considered. Table 5 shows forecasts which are indicative of those reported from the various models. The Limits study remains the most pessimistic for the long run, although the SARWM (~~~~~~~~e~)and UNITAR studies both present relatively low growth rate scenarios, Previous studies by this author and others suggest that the trends forecasts FUTURES

August 1987

408

Global mo&ls

TABLE 3. METHODOLOGICAL Study

Systems

Limits WIM Latin American Leontief FUGI

xxxxx xxxx xx

SARUM Global 2000 SlMlGDP

xxx

X

EMPHASIS Input-output

Econometric

Politico

X

xx xxxxx xx

xx xx xx xxx

xxx xx

World Bank UNlDlSEA UNCTAD

X

UNITAD UNITAR

xx xxxx

xxx

xxxx xx xxxxx xxx X

xx xx

SIMPEST GLOBUS LINK INFORUM G-MAPP UNU

IN STUDIES

xxx xxx

xxxxx X

xxxx

X

X

xxxx xxx

xx

X

Note: An indication of the methodological emphasis in each study is given. In principle, systems studies emphasize the dynamic feedback between variables, input-output demonstrates the importance of the underiying economic structure and econometric models stress the importance of historical trends and relationships.

reflect the world view of the forecaster almost as much as the reforms suggested. With modelling studies reviewed here, the institutional distance of a project from government agencies sometimes appears to provide as good an explanation of its forecasts as either its methodology or the historical data used (hence the distinction between institutionalized models and non-institutionalized models). With some models this distinction is not straightforward, since while they started life in government or international agencies they are now largely independent exercises (eg SARUM and UNITAD), whereas others are nominally independent exercises but depend heavily on government agencies for support. Even though the distinctions between content, forecasting horizons, detail and methodology alone cannot explain differences between the forecasts actually presented, they do often determine the technical problems that confront modellers. With varying emphasis, the models discussed below combine these techniques, each with their own agenda and more or less useful to futures studies. Keeping in mind these different approaches to modelling, in the sections that follow it is most straightforward to discuss individual projects, in order to clarify the role of modelling in each as well as the variety of problems faced, and to give an indication of the issues addressed and forecasts made. The models discussed in this article are essentially ‘ballistic’, ie the scenario parameters are set up at the outset and internal relationships determine the growth trajectory. Other models and applications described in a later article are less quantitative or use a series of scenarios to determine the path traced by the economic model. FUTURES Auguet 1987

Global mohls

Revisions

of systems

409

models

Club of Rome models The Limits to Growth, Strategyfor Survival and Latin American World models were each more or less ‘fostered’ by the CoR. Together with the FUGI, SARUM and Leontief models, these models roughly represent the ‘systems’ tradition in global modelling. Each took a more or less systemic view of the world (as a set of interrelated economic, social and ecological processes) and also the philosophy of model construction drew on the disciplines of systems analysis and control engineering as much as on economics. All these models have been reviewed in detail elsewhere (see BibIiography and earlier issues of Futures) and this critique will not be detailed here. Although sharing a systems philosophy, the CoR models each adopted somewhat different TABLE 4. LEVEL OF DETAIL IN MODELS Study

Horizon

Regions

s8ctors

Factor8

2100 2025 2060 w2000

1 12 4 15

6 15 9 45

-1990 1990

Z (62)

14 3

3

SARUM

2020

12 (13)

11

World Bank: SIMLINK WDR IMF UNIDIESA:

1990 1990 1966

7 (30) 15 (30) 45

2000 2000 2000 2000

15

22

g 15 (6)

2; 4 12 3

Limits WIM LA Leontief FUGI: GIOM GMEM

lGOEM WTM UNCTADlSlGMA UNITAD UNITAR

2000 2020

x

SIMPEST GLOBUS

1996 2015

3 5 (25)

LINK INFORUM

1990 1990

-

25 15

6 (10) 6

-

Trade

D8t&Sfz8

0 15 3 45

500 5000 3000 5000

10

3

2000 lOOff0

3

11

4000

&17)

:

1000 3000

2 2

22

3000

2

2; 4

2090 3000 1500

i(4)

11 3

4000 300

: ; (16)

4 (35)

3 2 (6)

3 3

4 50-190

2 2

3 2 (6) 4 119

1000 4000 10000 10000

Notes: The data give an approximate guide to the level of detail In the models. Where there are discrepancies between sources as to the number of sectors etc, the largest figure reported is taken or alternatives given in parentheses. Forecasting horizons given are as used in scenarios or projections. Regions may be single nations, politico-economic blocs, developmental groups or geographic areas. The number of sectors and actors suggests thedetail for each region. The estimates of model size give a very rough comparison of the amount of data items introduced into the models after aggregation from the initial database. These are estimated from the other data given and evidence in the cited references. Approximate size equals: number of elements in IO matrix (which is fairly empty when there are many sectors), plus the number of factors, plus the number of trade sectors, all times the number of regions. If the sectors are treated as subsystems, or the trade matrix is bilateral, or if the parameters are statistically estimates (as opposed to point estimates), or if a sophisticated solution method is used, the figures are adjusted accordingly. Typically, the figures are around 2-5 times the number of parameters or equations in the model and lo-20 times the number of subsystems. FUTURES Augu8t 1987

4 10

Global madels

TABLE 5. PER CAPITA INCOME FORECASTS TO THE YEAR 2000 ‘Worst’ Study Limits WIM Latin American Leontief FUGI

LDC

DC

2.6

2.5

2.7

DC (1973) (1975) (1975) (1976) (1974,1984)

‘Beet’

1.2

LQC 1.1

3.7 (2.8)

2.7 1.8 2.8 6.8 (3.6)

SARUM Global 2000

(1978) (1980)

1.0 1.2

0.4 1.0

f:80

5::

World Bank UNIDISEA UNCTAD

(1985) (1985) (1985)

2.0 2.3 2.2

2.7 2.8 2.8

3.8 3.8 3.5

z 4:o

UNITAD UNITAR

(19&Q) (1984)

2.6 0.0

2.6 0.8

2.0

2.1

Note: The figures for average per capita income growth rates presented here are indicative at best. The projections on which they are based refer to different publication dates, years and time horizons, different regional aggregations and different assumptions about policies and trends. The forecasts from individual models differ between applications and publications. In some cases (eg World Bank and Global 2000), the low forecast is also the trend forecast; in others (eg SARUM-Interfutures) the tow forecast Implies adverse conditions beyond those expected for the trend projection. in some instances, the beneficial effects of the favoured scenario are disguised because of aggregation (eg UNITAD) or are not apparent until beyond the year2000 time horizon (eg Limits and UNITAR). Results for FUGI in parenthesis are for 1984. Results are not relevant or available if no figures are given.

approaches to modelling and the way that the global ~~u~~~g~~q~ should be addressed. Limits to Growth, for example, used a continuous simulation method to emphasize the ‘Malthusian’ interrelationships between economic, demographic and ecological parts of the world system as a whole. The Latin American model, built as a critique of Limits, used a programming approach to show how four regions of the world (Africa, Asia, Latin America and the industrial world) might, through the adoption of more egalitarian societies, independently solve their problems in the shortest possible time. The Strutqyfor Sumival model by contrast used an interactive model to demonstrate the need for coordinated planning at a global scale. Between these early studies there was a fairly clear-cut methodological as well as ideological dialectic. The conclusions of the CoR models cannot, however, be considered ‘proven’: the models were an attempt to describe a clearly specified problem of the kind that futures research is obliged to address, and to use a methodology deemed appropriate to the problem. Each of the models then was innovative in method and in the issues dealt with. Each was an attempt to go beyond the bounds of contemporary economic modelling, and to develop a system which was global, not only in the sense of describing the world economy, but also including some non-economic social, political or environmental variables. Some of the difficulties with the early models were evident; data were not available and the theories underlying key relationships were not established. There appears to have been little further development of the Limits model, which may have suffered overly from strong criticisms. Some of the caution in present exercises also may be put down to an over-reaction to the complaints FUTURES August 1987

Global

modrls

4 11

levelled at the Limits models. The Latin American model also suffered from political developments in the Argentine. Nevertheless, both models left their mark. World Dynamics and Limits to Growth (and the publicity given to them by the CoR) undoubtedly stimulated new efforts in global modelling, even though other global economic modelling (Project LINK and GEM) had been under way since the mid-1960s. The Latin American model was used to explore an alternative approach to Brazilian development and the central idea of ‘basic needs’ satisfaction used in the model to define targets for development was taken up by the International Labour Office (ILO) in the Bachue series of models. The basic needs idea was also taken up by the World Bank (although not used in a model) and adopted for the UNITAR model. The Strategyfor Survival model of Mesarovic and Pestel was designed around the idea of interacting strata (or hierarchy) of variables, some of which would be internalized in the model, whereas others would be manipulated externally. In practice, the principal results were trajectories determined by the initial parameters, although the idea was partially implemented in the later World Integrated Model (WIM) and again in the International Futures (IF) derivative of the model, and possibly in the FORECASTS version (built for the US Department of Defense to evaluate international political and military strategies). The IF model also draws on the development of several earlier models (SARUM, Latin American, Leontief). The level of detail is comparable to those earlier models, and as distinct from the present trend to exclusively economic models maintains some appearance of the earlier systems models. Nevertheless, the model appears rather recursive and contains few of the ‘feedbacks’ which characterize the systems philosophy. With this variety, however, the model is used to simulate a variety of widely contrasting development paths (in this case based on the assumptions in published global futures studies, from Limits to Kahn). In effect, these are rather wide-ranging sensitivity tests. Like most models, IF contains a good many exogenous parameters (ie parameters that have to be fixed by the model user, either because theory is weak or in order to characterize an historic or future situation). Like its predecessors, the IF model contains a great many ‘user changeable’ parameters which permit experiments to be set up relatively easily. This model and the SARUM model will both be considered again in the context of the linking of models to scenarios. SARUM and FUGI SARUM and FUGI, like the Strategyfor Survival and Latin American models, were both a response to the Limits model but begun somewhat later and both are still being developed. Increasingly, however, these models have focused on economic issues, and paid relatively less attention to those more contentious sectors and issues of the first models. There are indeed few exceptions to this among the major modelling studies. The Global 2000 project (discussed later) viewed the issue of environmental constraints from a Limits perspective. The input-output model constructed by Leontief for the United Nations also provided estimates for specific environmental pollutants. Although the Leontief model is not reviewed in detail here, the original model provided the basic data for the modelling exercises ofthe World Bank and the United Nations (DIESA), FUTURES August 1987

412

Global

models

and although superseded in this respect the model has been used also to explore the consequences of redirecting military expenditures worldwide. The SARUM global model has led a rather chequered career and has been employed in a variety of future studies. It was initially established by the UK government in an attempt to review the Limits to Growth controversy. Interest in the UK ceased when “it was realised that the problems of limits to growth were institutional and not physical”. However, the model was novel in its treatment of international trade, technical change and product substitution. Initially devised as a ‘systems’ model, it has now become more or less an economic simulation model, yet still takes explicit account of scarcities of resources in the agriculture and energy sectors. SARUM was adopted, in preference to other models then available, by the OECD Interfutures study (described in a later article) and subsequently revised for use in Australia and some ‘political dynamics’ variables added (as part of the G-MAPP project). The original model described only three regions (high-, medium- and low-income countries) and included a fairly detailed agriculture sector. The detail has now been generally extended (eg to 12 regions). The FUG1 model which was supported initially by MITI, the Japanese Ministry of Trade and Industry, has become part of the G-MAPP project and associated with some UN Secretariat forecasting exercises. The approach reflects the systematic development of two models, GMEM and GIOM, based respectively on econometric and input-output methods. In this combination the less detailed GMEM model provides external inputs to the much more disaggregated GIOM model. The two models have developed in parallel, and although each may be used independently, it is possible to use the input-output system to elaborate the ‘policy relevant’ detail. Despite the relative simplicity in the breakdown of production, even the GMEM model details more than 60 major countries and regions. It also employs a sophisticated optimal control system for balancing bilateral finance and commodity flows between all economies (again by contrast with the ‘pool’ approach). This probably makes the combined FUG1 system (if fully implemented) the prime contender for the title of the world’s biggest model (previously only rivalled by the LINK project), with about 100 times as many equations as the Limits model, and which many people found rather awe-inspiring at the time. However, some new efforts in the United States aim to link several large existing models together on a powerful computer. Like other models, FUG1 aims to project current and future possible trends and to explore alternative development strategies. The high and low projections of the model differ by at most a few per cent, suggesting either an unrealistic degree of certainty in a world where growth forecasts are constantly revised, or inflexibility in the model. In so far as the agenda of economic modelling is to construct a world economic model that is policy applicable at all levels, then FUG1 demonstrates some of the possibilities for such an exercise. Nevertheless, an issue still to be resolved here is whether the vast amount of detail provides better forecasts of world economic growth, or even of the growth of individual countries. From the point of view of the future studies agenda, an equally relevant issue is whether the very detail of the model does not prohibit exploring ‘alternative’ strategies in anything but a naive way, by simply changing whole FUTURES August 1967

Globalmodrls 413

blocks of parameters in the same way (in which case the exercise might be better carried out in a suitably simple model). Institutionalized

models

World Bank (and IMF) Institutionalized models are those which are primarily expected to respond to policy needs. The World Bank modelling effort is the archetype of ‘institutionalized’ global modelling and demonstrates many of the difficulties of carrying out research and pragmatic policy making simultaneously in a constantly changing political environment. For much of the 1970s the Bank committed extensive resources to modelling exercises, data accumulation and the preparation of the World Development Report (WDR). This work was more or less supportive of an indicative approach to planning at the global level. The modelling programme passed through several stages, moving into the third generation by about 1980. From 1980 a process of consolidation has begun. Broadly, the modelling objective has been to have a ‘reliable’ model for the preparation of projections. Early discussion of the modelling approach at the Bank at this time exposes a dilemma of modelling in a policy-making agency, namely that while the construction of comprehensive global models requires a centralized team of specialists, this cloistered arrangement risks losing touch with current events, and paying too much attention to academic niceties. This dilemma was partly resolved by the preparation of separate country models and models dealing with commodities and sectors of interest in the relevant departments, but with enough overlap for some degree of linkage to be effected. In the event, most exercises have been collaboration between the Bank’s extensive research departments and universities (mainly in the USA), In the early 1970s a simple ‘two-gap’ SIMLINK model was developed. Briefly, this related the deficit in domestic investment and foreign exchange requirements, to an assumed’overall growth rate). It became evident (especially in the light of the first oil crisis) that other factors had to be accounted for and a more detailed model was developedthe WDR model. This was considerably more complex, consisting of a linked dynamic linear programming model, an elaborate system of capital and debt flow equations and a bilateral trade matrix. In turn, the need to account better for the effects of international markets provoked research into a general equilibrium type of model (one in which the interactions of different markets are treated simultaneously). This model (called M3) was built at the University of Louvain. Modelling work at the Bank had also been influenced by earlier research, termed ‘Redistribution with Growth’, which attempted to address wider economic development issues and led to a programme to construct social accounts (displaying, for example, urban-rural and household income distribution) for about 20 developing countries. Although not used in the global modelling, this and similar work at the Bank contributed to its thinking about development objectives. By 1980, the Bank was drawing on three sets of models to prepare long-term projections (the WDR, 35 country models and the experimental M3 model) but with tremendous problems in achieving consistency in growth, investment and FUTURES August 1997

414

Global models

trade between the models’ results. Despite the apparent technical progress then, it is evident that some modelling complexities had yet to be resolved. At this time, however, there was a considerable change of direction in the philosophy of the Bank, stemming from a change in the US administration, which appears to have affected modelling and development activities. The objectives shifted from the somewhat narrowly defined basic needs and development targets to the even narrower one of measuring the ‘fiscal soundness’ of potential borrowers. At the same time, the role of the models in the WDR annual projections has diminished, and estimates of economic growth depend mainly on the evaluations of the various country desks (on the basis of short-run economic and political risk assessment) together, once again, with the help of two-gap models. Briefly, the approach now adopted by the Bank is to project potential output in industrial countries from given assumptions about factor supply (labour and capital) and technical progress. With this information and considerations of the local situation in developing countries, the regional forecasts are prepared independently and then reconciled to comply with the Bank’s aggregate projections of capital availability, trade and debt servicing. This reorientation in the Bank in effect resulted in a period of reappraisal and consolidation, and some considerable refocusing of activity towards less elaborate models, within a more piecemeal, case-by-case approach to the developing countries. This change appears to have had little effect on the actual forecasts, even though there is some evidence that they are systematically biased. Typically, the Bank prepares a ‘high’ and a ‘low’ forecast for the remainder of the decade (and until 1995 since the 1984 Report). The low forecast is characteristically a projection of a continuation of present policies; eg the 1984 Report reads: Governments of industrial countries would find it hard to control inflation, and their budgetary deficits and unemployment would remain high. Protectionist sentiment would be strong, threatening the exports of developing countries and their ability to service their debts (but) with some increase in the penetration of markets in industrial countries. The high forecast, by contrast, assumes that a package of policy measures to stimulate growth is adopted by the industrial countries. According to the 1984 Report, this would wishfully: offer industrial countries a path of steady and sustained expansion . . . lower domestic interest rates and smaller budget deficits, investment would increase. As unemployment eases, protectionist measures would subside, so developing countries would find it easier to expand exports and to ease their debt service burden. Investment confidence would rapidly improve, which along with larger aid programmes would lead to an expansion of the flows of capital to developing countries.

This package of recommendations differs rather little from year to year. The Bank’s ‘low’ forecasts for the industrial countries are shown in Figure 1. These forecasts “indicate what will happen if the industrial countries do nothing to improve their performance of the last ten years”. They are typically around 2.5 % and show little variation, in contrast to the data the Bank itself presents on past growth rates (also shown in Figure 1) which show a steady downward FUTURES August 1987

Global models

4 15

trend. Applying common sense or simple extrapolation to these data presented by the Bank suggests that their ‘low’ forecasts are always optimistic and should be discounted by at least one or two per cent. This bias appears to be a reflection of the particular institutional pressures on the Bank and on its forecasting department. The ‘high’ forecasts are also shown in Figure 1. United Nations Secretariat The United Nations modelling research in the Department of International Economic and Social Affairs (DIESA) began in the early 196Os, and has seen a similar pattern of rise and decline as that in the World Bank. Most of the other UN agencies have also been involved to some degree in global modelling and were involved with several of the models mentioned above. UNEP sponsored the input-output model of Leontief and UNESCO and the IL0 provided some support for the Latin American model. In addition, UNCTAD has been involved for several years in the preparation of international trade models and the regional agencies (especially ECE and ECLA) in the preparation of linked national models. The main programme was otherwise greatly stimulated by funds from the Dutch and Swedish governments in preparation for the UN

6

I4 I

I

I

I

I

I

I965

I 970

1975

1980

1985

1990

Year 0

Current

annual growth

rate of industrial

economies

x

Trend predicted by least squares regression

+

World Bank ‘high’ projections

A

World Bank ‘low’ projections

at midpoint

of decade

Figure 1. Annual growth rates and projections for industrial countries Source: World Bank Development Reports (1978-85)

FUTURES August 1987

I 1995

4 16

Global malls

Third Development Decade. In this programme, agencies worked together on modelling exercises. For example, UNCTAD and UNIDO together developed the UNITAD model, while UNITAR, ILO, ECLA and UNESCO carried out independent exercises (around the common theme of basic needs), as did the FAO and the CDPPP (the principal agency involved, now the DPRA). Three of these exercises will be commented on below. The activities of the DIESA (carried out in the CDPPP, later renamed the DPRA-Department of Policy Research and Analysis), provide the major input to the annual World Economic Survey and World Economic Outlook of the UN and the International Development Strategy (IDS) proposed in 1982 by the UN for the Third Development Decade (1980-90). Until the joint exercises noted above, the Global Econometric Model (GEM) and the collection of global statistics were the major activities in the CDPPP. From the mid-1970s, a new major effort in the CDPPP centred on the building of a large input-output model (developed in part out of the Leontief global study). This approach presented some problems. In particular, the absence of behavioural relations in the model meant that projections were quite unconstrained, either by raw materials or environmental constraints (which are not included), and even by the availability of invested capital. Despite this, the very detail of the model made it cumbersome and time consuming. An attempt was made to remedy these problems through a less detailed model, DYNAMICO, which used a sophisticated approach to determine optimal paths of economic resource allocation, first at a regional level and then at an integrated world level (a version of this model was also constructed at the Siberian Academy of Sciences). Despite the effort to introduce more elaborate models, the major model used by the DRPA for long-term projections is still the GEM, together with a world trade matrix. GEM consists essentially of a set of relatively simple ‘two-gap’ country models. Although the details and data have improved significantly, the model is essentially the same as that used by the UN for the last 18 years. The main data and projection activities of the UN now fall under the Macroeconomic Data System (MEDS) of the DPRA. The main purpose is to create a detailed and consistent set of economic, demographic and social statistics for the projection activities of the UN, to estimate econometric relationships for individual countries and to link national and regional models into a world model. The database and the modelling are treated as an integrated computerized activity so that both data and models can be relatively easily updated. The most recent data (in 1985) were for 1981, so the database lags behind events by around four years (a problem when the data are to be used for short- or mediumterm projections or when economic performance is erratic). For this reason the data are used in conjunction with the projections of other models. GEM provides the projections for the annual economic outlook-reporting on the prospects of developing, market and centrally planned economies, the long-term prospects for the world economy and monitoring the progress of developing countries towards the goals set for the UN Third Development Decade (the IDS). The GEM model is typically used to calculate investment requirements and trade levels using given growth assumptions. For example, to set up the baseline (ie trend) long-term projections to the year 2000, the FUTURES August 1987

medium-term growth 1990 Ijrojections of the FUG1 model and the short-term (3-year) projections of the LINK system for individual countries are inserted into the model and projected forward. The model then indicates the capital requirements of developing countries. To calculate trade levels, imports are either treated as a fixed share of domestic consumption, and then exports calculated as a residual after domestic production, or exports and imports are simply calculated separately (using the UN bilateral trade matrix, predicting a balance of payments gap for each country and a world imbalance on supply and demand). Three further projections are made, an ‘optimistic’ projection based on the assumption of a recovery to the average growth trends from 1965 to 1982, a ‘pessimistic’ trend without recovery based on a continuation of 1975-82 averages, and a trend based on the growth and investment targets in the IDS. The IDS strategy itself is intended to exploit each region’s ‘carrying capacity’ -national governments adopt increasingly expansionary policies to escape from present stagnation and, as time elapses, exploit more effectively the growth potential of their economies. The IDS “postulates growing diversification in the developed market economies, along dynamic comparative advantage lines”, to achieve their pre-1973 performance. This is found to export economic recovery to the open developing countries. The task for the developing countries then is to poise themselves for the upsurge in external possibilities. For example, “National policies must eliminate major distortions and allow the most productive use of inputs and choice of output composition within the framework of ongoing global structural change. ” UNCTAD/SIGMA UNCTAD has been involved in the construction of commodity and trade models since the 1970s and contributed to the UNIDOWNCTAD-the UNITAD model mentioned above. In addition, this agency has also embarked on a new global modelling exercise, System for Interlinked Global Modelling which has provided forecasts for the 1985 Trade and and Analysis (SIGMA), Development Report. The SIGMA model is similar to the FUGI-GMEM model, with a rather straightforward domestic sector but with an elaborate trade and payments system. It is an econometric model consisting of up to 15 single sector regional growth models, an exchange rate model and a model to determine the (spot) price of traded commodities. The share of output in agriculture, mining, manufacturing and other activities is determined for each region from behavioural equations, and international trade is determined from a bilateral trade matrix for food, raw materials, energy and manufactures, such that exports depend on regional production capacity and competitiveness. Exchange rates, inflation and commodity prices are calculated jointly from trade imbalances. Some attempt is under way to integrate a financial sector and a demographic sector into the model (the latter is significant because demographic change would respond to economic change, and vice versa, as in the original CoR models). The UNCTAD report provides “a medium-term outlook on present trends”, and a “positive alternative”. In the latter, global macro-economic variables are “assumed to move in a positive direction, creating a much more favorable FUTURES August 1987

418

Global

models

environment for debt services and development”. The differences from the World Bank and IDS strategy are seen in the magnitude of the economic transfers from industrial to developing countries, rather than in the character of the variables acted upon. In particular, a hypothesized increase in economic growth in industrial countries, together with a relaxation of trade restrictions and an easing of financial pressures on developing countries, lead to a renewed expansion of their economies. Relative to comparable strategies (such as that in the Latin American model); rather less emphasis is placed on domestic growth factors, intra-South trade and economic self-reliance in the developing countries. The UNCTAD study emphasizes the use of ‘scenario analysis’ in its methodology. The meaning of this is somewhat restricted from the standpoint of futures studies but idealizes the modelling process. It consists of four stages: l l

0 0

identify key factors important to the medium-term evolution of the economy; construct a flexible analytic framework which shows how these factors interact and impact on the evolving situation; use the framework to simulate trends in these factors under a range of different (scenario) assumptions; examine the results to ascertain the plausibility of the scenario as a whole (without necessarily placing any probability on its occurrence).

De-institutionalized

models

Several modelling efforts initially sponsored by international agencies (UNITAD and UNITAR), like those initiated by national governments (SARUM and FUGI), h ave subsequently transferred to universities and made more independent arrangements with a variety of sponsors. Several reasons account for the de-institutionalization: the end of the major DIESA funds, personnel retirements, a cutback in research funding generally, a failure to keep up with the changing complexity of world events, and shifts in priorities which mirrored those in some of the major industrial nations, or simply a refusal to concur with the dominant forecasts. The transformation has provided an opportunity for studies which have had good working access to the extensive databases of the UN to be elaborated in a direction which is less constrained in its adherence to an imposed external wisdom than the major agencies, and to pay more attention to the domestic and distributional dynamics of development. Generally, the more independent forecasts are more pessimistic in their trend projections, examine a wider range of options, are more radical in the reforms they suggest, and are more optimistic about the outcome. UNZTAD The opportunities posed by this institutional defrocking are typified by the UNITAD model, which brought together some of the experience of UNIDO in the area of industrialization in developing countries and of UNCTAD in international trade (the exercise is nevertheless independent of the SIGMA model described above). Although the team was largely disbanded when the original FUTURES August 1907

Global

mo&ls

419

funds terminated, the model has been systematically modified and updated using ad hoc funds from various UN agencies (including UNCTAD). The UNITAD model is somewhat less detailed than the majority of models described above. In large part this was because of the realization that great detail does not guarantee accuracy any more than it does relevance. It was argued that because input-output tables with more than about 10 sectors fluctuate significantly from year to year, this represented an upper limit on sectoral detail (in all but short-term models). In any case, the most significant results of this model come from the way the traditional and modern sectors are treated in the input-output core of the model. While the same results could be inferred from much more detailed input-output studies, the amount of computation would be much greater, and the reliability probably less. Despite this smaller level of detail, the model nevertheless demonstrates some fairly general propositions about global development. Since its de-institutionalization, the UNITAD model has been used to explore a wider range of trajectories than hitherto possible, and to highlight additional features of the ‘standard’ projections. For example, the results suggest that if developing countries rely on modernization and the pull effect of the industrial countries (ie export-led policies) to achieve economic growth, this will have little impact on their present situation of massive unemployment. By contrast, the results show that policies which take account of the internal structure of developing economies as represented by the model structure are needed in LDCs (in effect by internal subsidy of the small industry sector). The most recent experiments with the model (1985) start from the basic assumption that the world economy will remain for another lo-15 years on a low growth path, thus leaving bleak prospects for industrialization strategies in developing countries based on exportation to the North. Instead, the model is used to simulate industrialization strategies fuelled by the domestic market, in which systematic complementary links are established between lowproductivity (rural) activities and high-productivity activities. The strategy differs from ‘import substitution’ type strategies proposed in earlier decades, in that the strategy demands income redistribution, land reform and governmental action in the fields of education, nutrition and other basic needs. While not all these variables are explicit in the model, the general argument is made that a coordinated action “ will trigger an equalitarian spiral, thus increasing welfare of the poorest . . . and substantially improving the employment balance”.

UNITAR The UNITAR model, in addition, chooses a higher level of regional and sectoral aggregation but is more explicit in its treatment of distributional issues and market mechanisms. The model grew out of the Science Policy Research Unit (SPRU) critique of global models and the scenario study World Futures (discussed later) and some of the ideas developed from the Latin American (Bariloche) model, with funding from the UNITAR Project on the Future. The UNITAR model focuses on the relationship between international and intrunational income distribution within different global economic strategies (described by alternative arrangements for trade, aid, technology etc). With six types of economy described (based on their industrial and political developFUTURES August 1987

420 Gloklnwdds

ment) and only three production sectors, the model nevertheless has an equal level of detail for both the supply and demand side of each economy. Economies and sectors are characterized by types of technology, workforces by skills, and households by wealth and income. To establish the parameters, the model uses the detailed Social Accounting Matrices previously collected by the World Bank and the trade data of the CDPPP, which are simply aggregated in a particular way into a much less detailed form, running on a microcomputer. The model is used in two ways. The first simply evaluates sharply polarized alternative international arrangements (eg autarchy u free trade) or technology (eg appropriate technology u modernization through transfer), and then a more balanced combination of policies sought. In this mode the model is an attempt to bridge the gap between the large-scale global models and the ‘stylized’ results of economic theory. This required sufficient detail to allow some useful insights into the major proposals for internation~ economic development, yet also allowing the results to be contrasted with the gains from trade and factor equalization theories, which are the main abstract tools used to interpret the results of the neo-classical models. Essentially this is a comparative static application of the model. The second use of the model is as a component of a more comprehensive scenario analysis (discussed in a subsequent article) in which the model experiments are integrated with non-quantitative variables, describing changes in the international and political order. For this, growth-related factors are assumed external to the model. The UNITAR model has been used to explore the implications of a wide range of strategies and the trade-offs between the objectives of economic growth and social redistribution (and, by implication, for political independence). These are reviewed later in the context of linked scenario-modelling studies. Some of the results are relevant in their own right to the earlier discussion. The results suggest that unless there is a rather careful coordination of technology, trade and social policies, some well meaning development strategies (including those of the World Bank and the Brandt Commission) may do as much harm as good. For example, the World Bank has recommended the use of development aid and small-scale production units to promote increased employment and economic growth in the rural sector, while simultaneously exhorting these countries to liberalize their markets. According to the UNITAR model, the effects of using small-scale and low-productivity technologies in developing economies can be undermined if those economies are also required to open their markets to the full force of international competition. By contrast, if the destabilizing effects of international markets are brought under control, then strong multipliers can operate in the domestic economy. This promotes the spiral of growth observed in the UNITAD model. A revised version of the model (S&c/&g the Chips) has also been used to examine the impact of an ‘info~ation technology revolution’ on world and domestic distribution and employment. The results suggest that in the industrial and richer developing countries (the ‘newly industrialized countries’), the future distribution of income and employment depends more on trends in the characteristics of the new technology than on trends in its location. (There is currently little consensus about either trend.) Further, the indirect effects of the FUTURESAugwt 1987

Globalmoakls421

new technology may be greater than its direct effects, especially countries, and do little to relieve unemployment there. Politico-economic

in the poorest

models

Politico-economic global models are broadly defined here as world models in which both economic and political relationships are formalized and given roughly equal status. There are rather few such models-perhaps only GLOBUS and some preliminary work with the SARUM model qualifies for inclusion in a review of global models-but other models such as SIMPEST deal with the same central issue of global interdependence and conflict. These last variables represent the central concerns of this class of models, particularly as they affect the militarization and tendencies to domestic and international conflict. Attempts to build these models undoubtedly reflect the importance of these issues at the present time. Nevertheless, these have been a major concern of political scientists generally and theie are numerous examples of game theory, bargaining and negotiation models, used to analyse these issues. Thus the advent of global models in the field of international political economy may largely be the result of a trend within the discipline of political science (as well as frustration that most global modelling exhibits a careful neglect of the more contentious subjects). These are not to be reviewed in detail here, but some additional approaches will be given in a later section. Although there are a fair number of examples of stylized politico-economic models going back over a number of years, and somewhat abstract prescriptions for their elaboration, there are relatively few examples of empirical multiregional models. In some cases the political model is added to an existing economic model. In particular, a version of the SARUM model (as part of the G-MAPP project discussed below) has been modified to include some government allocation processes, a military production and trade sector and a strategic raw material stockpile. These variables are determined by a new set of equations describing the political dynamics of the arms race according to the theory of Richardson which assumes arms expenditures to vary according to the level of ‘fear’ induced by a potential enemy’s current expenditures, fatigue arising from the extent of other commitments, and either ‘ambition’ or ‘grievance’ which determines the threshold expenditure. This simple concept originally based on a twocompetitor model, is extended to explain the dynamics of the arms race from the bilateral relations between many nations. The level of military expenditures then affects the allocation of public and private resources in the modified SARUM model, so leading to an interrelated economic and militarization projection. Depending on the parameters employed to quantify the political dynamics, the model may lead to an upward spiralling arms race, a more or less status quo, or a steady build-down. The SIMPEST model describes the political relationships between ‘the three military powers’the USA, USSR and China. These states are each distinctively characterized by their economy, internal polity and government. According to contrasting sets of rules deemed to reflect the political systems described, the various governments extract resources from their respective economies and allocate them according to prevailing conditions such as the level of support FUTURES August 1987

422

Global models

from domestic elites and the expenditures of other governments. Although the economic model of SIMPEST is relatively simple, it is sophisticated. The model is set up as a continuous simulation and the parameters are more or less simultaneously estimated, which is generally taken to be a rather difficult procedure.

The GLOBUS project (at the Wissens~haftzentrum) is an attempt to construct a new politico-economic model. Although this led to several technical and conceptual problems, the study of the economic system in GLOBUS has some innovative features: with respect to international trade, in particular, the model includes a bilateral trade matrix which takes account of political variables and also-the lags between shifts in policy or comparative advantage and the response of the international economy. Of more immediate interest here is the system of mutual action-reaction relationships which govern the pattern of hostility and cooperation in response to fluctuations in international and domestic political and economic factors. The GLOBUS model apportions roughly equal weight between the economic and political relationships. The main actors of the political model are national governments who monitor their bilateral relationships with others, and adjust foreign and domestic policy in the light of this. This may consist, for example, of tariffs to shield domestic producers from foreign competition, political preferences for trade partners, and changes in government expenditures on civilian (education, health, social security, foreign aid) and defence budgets. Civilian expenditures do not appear to affect demographic and labour force characteristics, but do determine the domestic political support for government (and hence the trade-offs to be made). In politico-economic models such as GLOBUS and SIMPEST, model experiments are set up, as in regular economic models, to reflect an initial or ongoing situation and the model runs forward ballistically through time. The scenario then unfolds as determined by the relationships in the model. The behaviour of these models is internalized, an approach in which the indicative variables of the model are repeatedly assessed and appropriate adjustments made to the model parameters. The GLOBUS model is used to study the impact of increased military expenditures by the Western powers and the effect on further escalation. The ‘freeze’ option appears to lead to a systematic reduction in overall levels of expenditure (to half the present level over a period of 30 years), with an especially dramatic increase in social expenditures in the East. By contrast, a 6% increase drives both the West and East into the situation of superpower mutual reaction behaviour. There is as much controversy about this mechanism as about the functioning of world markets, and, as in most global economic models, the main conclusions of the results of a particular set of assumptions may be self-evident. For example, if it is assumed that the principal goal of the Eastern Bloc is to match the West’s capability, then it is evident that an end to the ‘arms race’ would have a much greater boost to the economic growth and social expenditures of the socialized East than the democratic West, given the present relative economic size of the two Blocs. Thus, as in more straightforward FUTURES August 1987

Globalnwdcls 423

models, the methodological debate surrounds the adequacy of data and theory, the relevance of the details, third-party relationships and feedbacks. Modelling

consortia and the development of peer groups

One of the distinctive processes of global modelling has been the formation of in which several models are brought together in a joint ‘consortia’ -groups project. One reason for this, as in the two efforts (LINK and INFORUM) considered next, is that the arrangement is an integral part of the construction of global models which are to be built up from the separate economic models of a select group of countries. Another reason is an attempt to establish like-minded peer groups, as in the setting up of the subgroups between the UN agencies. A third reason is as a response to the consolidation activity as funds for ud hoc research outside the major agencies became limited. The G-MAPP project and the United Nations University (UNU) Food-Energy study were set up in this way. In this case no actual linking of models took place; data were exchanged, common trajectories explored, and the results compared. The Global 2000 also brought together the sectoral models of various US government agencies and regional projections based on the SIMLINK model, in a single report. The methodologic~ interest in these efforts lies in the various technical (and organizational) problems of interrelating the studies, and the impact that this has on the forecasts. Inter-agency collaboration Through the UN Secretariat (DIESA), UN agencies meet on a regular basis to compare forecasts. Given the acknowledged dif~culties in forecasting and the great uncertainty in present international economic affairs, there is a surprising degree of uniformity. For example, the ACC 1985 Report states that: Despite contrasts in approach and purpose, Agency assessments of the prospects for world economic growth all point to the same conclusion. Although baseline forecasts presented at-the meeting suggest a spreading recovery in output growth in all regions, the extended outlook for the developing countries . . . is uniformly bleak. Notwithstanding the possible mild upturn in the immediate future (according to the short-term IMF and DIESA/LINK projections), all agencies expect that growth in each major world region during the rest of this decade will remain considerably below the pace attained in the 1960s and 1970s. For the trend projection at least there appears to be agreement (see Table 5). Generally, the models’ high and low variants differ by a few per cent-the low projection reflecting ‘present trends’ and the high projection based on pre-1973 growth rates. The models alone do not project the trend, since parameters affecting growth are usually adjusted so as to give plausible trend projections. Rather, the models are used to interpret the consequences of assumptions about major growth variables. Essentially what happens in this situation is that any differences in internal structure of the models shows up in other relatibnships such as the links between industrial and developing countries (through trade, aid, finance, etc). Comparing the low and IDS strategies of the DIESA, for example, suggests a linkage of about 1.6 (measured as the ratio of developing to industrial country

FUTURES August 1997

424

Global

mo&ls

growth), whereas a comparison of the World Bank high and low projections suggests a linkage of only 0.7. This last figure possibly indicates considerable pessimism that the level of protection against developing countries will be reduced, even if higher growth is achieved. The linkage figure suggested by the Bank’s forecasts is also markedly different from that implicit in the UNCTAD exercises, when the low and high projections of the two agencies are compared. The UNCTAD study suggests a linkage of around 3.7, since the policies they recommend lift industrial country growth from 3.0% to 3.3% and developing country growth from 4.0% to 7.0%. In this case, substantial and ahistorical real income transfers from rich to poor countries are assumed. Like other forecasters, the agencies place the usual caveats on the precision of their forecasts: the World Bank uses its high and low perspectives to “illustrate the range of possibilities” (although the trend forecasts are optimistically biased); UNCTAD cautions that “it is, of course, not possible to gauge the likelihood that the future might resemble this scenario”; the DIESA projections the apparent are “grim, but not without hope”; and so on. Nevertheless, authority of agency forecasts is not lessened by their apparent mutual reinforcement . LINK and INTERFOR

UM

The LINK project is one of the oldest of the global modelling efforts. The study has been a systematic attempt over nearly two decades formally to integrate econometric models of a significant number of countries (currently 13 OECD countries, including Sweden, seven CMEA countries and live developing areas). The OECD models are set up separately in their respective countries, the CMEA models based in the CDPPP and the developing country models on UNCTAD and DRPA two-gap and trade models. Despite the relative detail of several of the country models (more than in the FUGI/GIOM project), the international transfers have in the past been via a rather highly aggregated trade matrix-a reflection of the fact that it is very difficult to compare product categories across countries and decipher the various ‘standard’ classifications employed. As with the observation about the variability of input-output coefficients earlier, this casts doubt on the value of very detailed model descriptions. The LINK project is a short-term projection exercise, and in this sense is less relevant to futures studies. On the other hand, it does set a standard in terms of systematic long-term approach to methodological goals, as technical problems are identified and overcome. In particular, since the major feature of the model is the trade relations, it is again instructive to note the acknowledged limitations of the model: exchange rates shifts are assumed, not calculated and no account is taken of changes in the trade matrix (which described the propensity of different countries to import goods from each other). A similar project, INTERFORUM, demonstrates what may be an even more ambitious approach in an effort to impose sufficient uniformity onto a set of similarly detailed input-output models so that the whole system can be manipulated by a single user. A principal objective of the LINK system is to show how internal multipliers in domestic economies are stimulated by international trade. With respect to this, some of the experiments carried out with the model also clarify differences FUTURES August 1987

Globalmodels 425

between the assumptions of short- and long-term exercises. For example, in looking at the short-run impact of protection, the LINK model concludes that while some countries gain, most lose, especially the developing countries. Given its short-term focus, the model cannot say whether all countries lose in the long run because of the knock-on effects throughout the world economy. This would support the notion of freer trade relations. Conversely, in the longer run other policy measures (such as income redistribution and better choice of technology) might be combined with selective protection in an inward-looking strategy so as to make more effective use of the internal multipliers than through exportation alone (as in the UNITAD and UNITAR models). The LINK and INTERFORUM systems transfer data directly (ie electronically) between the various models. Consequently they require common relationships and structure between the sectors, subsystems and regions of individual global models, at their interface. This linking is somewhat tighter than that between the components of the FUG1 and CDPPP models, and considerably more than in the consortia typified by the Global 2000, G-MAPP and UNU projects. The Global 2000 study, like the LINK project, was based on existing sectoral forecasting exercises (in this case those of various US government departments). Only a single long-term ‘trend projection’ was made, and in contrast to most other contemporary exercises reflected a Limits philosophy, projecting a longterm slow-down in growth and in attempting to account for economic, demographic and ecological factors. However, the project treated these issues independently. Even for the trend projection, absence of links between sectoral studies poses problems; for example, an independent study with the WIM model (as background to Global 2000) showed that linking together the main sectors treated in Global 2000 would significantly worsen the Limits scenario, demonstrating the risks involved in neglecting feedbacks. The project closed with the last change in the US administration.

G-MAPP and the UNU The formal comparison of the results of several global models generally proves to be difficult, because structures, definitions and relationships vary. In some respects this was less difficult with the earliest models whose key underlying assumptions were very different and also clear-cut. It may also seem reasonably straightforward with models of fairly similar structure and assumptions. In the G-MAPP project, for example, a small number of models (FUGI, AREAM-a version of SARUM-and the WDR) embodying more or less the same (neoclassical) assumptions and each containingsubmodels of key Asian countries are brought together. From this basis of more or less common assumptions, a ‘stocktaking’ exercise based on considerable prior experience could be made in order to improve the models separately and also to define the next steps in the modelling agenda. An early conclusion was that even though the models have similar theoretical underpinnings, they are ‘methodologically incompatible’ to the point that FUG1 and AREAM may be used only to explore mutually exclusive problems. For example, because of differences in model structure, the GDP growth rates and other indicators forecast for 1970-90 differ, even though they start from the same database for 1970. With its enormous detail and fixed FUTURES August 1987

426 Globalmodels

econometric relations, FUG1 is suitable for exploring the fine detail of smooth trajectories which postulate no major change in the status quo. AREAM was considered more useful for investigating possible ‘major’ changes, such as shifts in trade alliances or major new industrial breakthroughs. These comparisons suggest that simple models perform as well as more eiaborate models (or at least give equivalent results) but with much less elaboration. For example, it was noted earlier that the FUG1 model provided inputs to the DPRA. In so far as direct comparisons are made between the very complex FUG1 model and the much more simple GEM model of the DPRA, the models give roughly similar results (ie a given set of assumptions about investment lead to roughly similar results for growth and trade), despite the fact that the FUG1 model is estimated on data from a shorter time span. The difficulty of making detailed cross-comparisons is illustrated also by the UNU Food-Energy project in which three models (SARUM, UNITAD and UNITAR) were selected for a comparative investigation became of their supposed differences and complementarities. Because the various models used two standard projections, free trade u selfdifferent methods of projection, reliance, were selected. Trajectories were defined by different assumptions about regional economic growth, trade arrangements, technology and so on, although in contrast to G-MAPP, more extreme variants were investigated. Even so, the very different treatment of key variables meant that only approximations to the agreed scerlarios could be calculated. For example, economic growth for the industrial economies was an input to the UNITAD model, but calculated in SARUM from assumed investment rates and so on. In some cases a more elaborate approach to the modelling of a particular subsystem was seen to be a constraint on the exploration of those same issues. It is evident for these joint exercises that, in bringing models to the point at which they can be compared, their distinguishing structures and modelling philosophies are muted under peer group pressure to achieve a convergent explanation of events. Given the various compromises this entails, it is not surprising that, like many other disciplines, the global modelling fraternity has become a network of somewhat incestuous interest groups.

Bibliography

by project

CDPPP DPRA Forum Humanum FUGI GEM Global 2000 GLOBUS G-MAPP GPE GPID Heritage Foundation IF IFDA

(cf main Bibliography) See DPRA. Costa (1984), DPRA (1981 a,b,c), DpR.A (1983), Granberg Rubinschtein (19841: see also GEM. Abbruzzese (1‘983).” Kaya (1977), Onishi (1984). DPRA (1981); see also DPRA. Barney (1980, 1984). Bremer (1982, 1983), Cusak and Eberwein (1985). MacRae and Mula (1981), Chadwick (1984, 1985). Jackson and Allen (1985). Galtung (1984), GPID Working Papers. Moore ( 1985). Hughes (1985a, 1985b); see also Strategyfor Sumiual/WIM. IFDA Bulletin (most issues).

and

FUTURES August 1987

Global models

IMF INFORUM Interagency Interfutures Latin American Leontief Limits LINK Megatrends OECD POLIS RIO SARUM Secretariat for Futures Studies Strategyfor SurvivaNWIM UN Secretariat UNITAD UNITAR WOMPlWSI World Bank

World Futures Worlds Apart Worldwatch WTI Reviews and critiques Economic

Political Technology Trade SAM/IO Other abbreviations CDP CDPPP DIESA IIASA IMF OECD

427

IMF (1982-85). Nyhus and AImon (1977), AIman (1984). DPRA (1985). OECD Interfutures (1979), Norse (1984). Herrera (1976). Leontief and Carter (1976). Meadows et al (1973), Forrester (1971). KIien and Su (1979). Naisbitt (1984). See Interfutures. Wilkenfield (1983). Tinbergen (1976), Leuddijk (1979). SARU (1978), Roberts (1977), Parker (1984). Inglestam (1973), Huldt et al (1980). Mesarovic and Pestel (1974) Hughes and Mesarovic (1978). See DPRA. Royer (1984, 1985). Chichilnisky and Cole (1978), Cole and Miles (1984); see also Worlds Apart. Faik (1973), Galtung (1984), Lazlo (1974). Tims and Waelbrook (1982), World Bank (1979-85), Round (1982), King (1981), Gupta and Waelbrook (1984), Hayter and Watson (1985). Freeman and Jahoda (1978), Miles (1978), Cole et al (1978). Cole and Miles (1984), UNESCO (1981), Miles (1981). Brown (1983-85). DPRA (1983). Churchman and Mason (1976), Clark and Cole (1975), Cole (1977), Hickman (1983), Leuddijk (1979), Lesourne (1984), McHaIe (1980), Meadows et al (1983), Mula and MacRae (1981), Munton (1978), Sanderson (1981), Siegmann and Deutsch (1986), Barney (1980), Lutz (1983). Alker (1981), Deutsch (1979), Guertzkow and Valadez (1981). Cole (1974), (1986), Hughes (1985 b), OECD (1979). Cole (1974), Khen and Su (1979), Pollins (1980). Blitzer ei al (1975), Cole and Miles (1984) King (1981), Round (1982), UNIDO (1984). Committee for Development Planning. Centre for Development Projections Policy and Planning. Department of International Economic and Social Analysis. International Institute for Applied Systems Analysis. International Monetary Fund. Organization for Economic Cooperation and Development.

Bibliography S. Abbruzzese et al, The Factors of Peace in the World Community, Interim Report (Rome, Forum Humanum Project, 1983). ACC Task Force, The Respects for World Development: Present Trends and Possibilities for Greater Cooperation between Developing Countries (Geneva, Interagency Technical Working Group, 1985). H. AIker, “From political cybernetics to global modelling”, in Merritt and Russett (1981), op cit. I. Alman, “The Inforum-IIASA International System of Input-Output Models”, in UNIDO (1984), op cit.

FUTURES August 1997

428

Globalmo&ls

G. Barney et al, The Global 2000 Report to the President-Entering the Twenty-First Century (Washington, US Department of State, 1980). “The Global 2000 Report and its implications”, in Levy and Robinson (1984), G. Barney, op cit. C. Blitzer et al, Economy W& Modclrfor Development Planning (Washington, World Bank, 1975). Brandt Commission, NORTH-SOUTH: A Programmefor Survival (London, Pan Books, 1980). S. Bremer, The GLOBUS Moa!el-A Guide to its Theoretical Structure (Berlin, Wissenschaftzentrum, 1982), pages 82-105. S. Bremer, The GLOBUSMO&~ (Berlin, Wissenschaftzentrum, 1983). L. Brown, State of the World (Washington, Worldwatch Institute, 1983-85). R. Chadwick, GMAPP Summary of Objectives, Findings and Concrete Results, mimeo (Hawaii, East-West Center, 1984). R. Chadwick, Modelling Political-Military Policy Dynamics in a Global MO&~, mimeo (Hawaii, East-West Center, 1985). H. Chenery et aI, RedistributMn with Growth (Washington, World Bank/Oxford University Press, 1974). G. Chichilnisky and S. Cole, “A model of technology, distribution and North-South relations”, Technological Forecasting and Social Change, 13, pages 297-320. C. Churchman and R. Mason, World Modclling: A Dialogue (Amsterdam, North-Holland, 1976). J. Clark and S. Cole, Global Simulation Models-A Comparative Study (Chichester, UK, Wiley, 1975). Futures, 6, (3), 1974, pages 201-218. S. Cole, “World models, their progress and applicability”, S. Cole, Global Models and the International Economic Order (Oxford, Pergamon, 1977). S. Cole et ul, “Scenarios of world development”, Futures, IO, (l), 1978, pages 3-20. S. Cole, “Methods of analysis for long term development issues”, in UNESCO (1981), op cit. S. Cole and M. Miles, Worlds Apart-Technology and North-South Relations in the Global Economy (Brighton, UK, Harvester/Rowman and Allenheld, 1984). World Development, Nov-D~c 1986. S. Cole, “The global impact of information technology”, agencies”, International Studies Quarter&, S. Cole, “World economy forecasts and the international December 1987 (forthcoming). projections on the basis of alternative A. Costa, “United Nations global modelling: experimental procedures”, in UNIDO (1984), op cit. Council of Europe, The Use of Long Range Forecasting Techniques and ProJection for the Definition of a European Regional Planning Policy (Paris, Futuribles, 1978). Council of Europe, Global Prospects: Human Neea!s and the Earth ‘s Resources (Strasbourg, Council of Europe, 1981). T. Cusak and W. Erberwein, The GLOBUS World MO&~-Some Preliminary Results (Berlin, Wissenschaftzentrum, 1985). K. Deutsch, From the National [email protected] State to the International Wevan System (Berlin, Wissenschaftzentrum, 1979). DPRA, Compendium of World Development Indicators (New York, DIESA, 1980). DPRA, 7%~ Macro-Economic Data System of the DPRA: Purpose, Description and Data Base Coverage (New York, DIESA, 1981a). DPRA, Handbook of World Development Statistics (New York, DIESA, 1981b). DPRA, The Global Econometric MO&~ of the United Nations Secretariat (New York, DIESA, 1981c). DPRA, United Nations Trade Matrix Studies, Report 2 (New York, DIESA, 1983). DPRA, The Prospectsfor World Development: Present Trena!s and Possibilitiesfor Greater Cooperation between Developing Countries, mimeo (Geneva, ACC Interagency Technical Working Group, 1985). DIESA, An Overall Socio-Economic Perspective to the Year 2000 (New York, Projections and Perspectives Study Branch, DIESA, 1984). EEC , Eurofutures: The Challenges of Innovation (London, Butterworths, 1984). R. Falk, A Study of Future Worlds (New York, The Free Press, Macmillan, 1973). R. Falk and S. Kim, The War System: An Interdisciplinary Approach (Boulder, CO, USA, Westview, 1980). J. Forrester, World Dynamics (Cambridge, MA, USA, Wright Alenn, 1971). C. Freeman, Long Waves in the Global Economy (London, Francis Pinter, 1984). C. Freeman and G. Jahoda, World Futures- The Great Debate (London, Martin Robertson, 1978).

FUTURES August 1997

Global mo&ls

J.

429

Galtung, The True Worlds: A Transnational Perspective (New York, The Free Press, Macmillan, 1984). A. Granberg and A. Rubinstein, “Development in the United Nations Input-Output Model”, in UNIDO (1984), op cif. H. Guertzkow and J. Valadez, Sin&fed Zntemationuf Processes (Beverly Hills, Sage, 1981). S. Gupta and J. Waelbrook, “World Bank global modelling research”, in UNIDO (1984) op cit. T. Hayter and C. Watson, Aid-Rhetoric and Reality (Brighton, UK, Pluto Press, 1985). A. Herrera, Cufustmphe or a New So&~ (Ottawa, IDRC, 1976). B. Hickman, Global Zntemational Economic MO&, IIASA Papers (Amsterdam, North-Holland, 1983). M. Hopkins et al, “Evaluating the basis needs strategy and population policies: the BACHUE approach”, Zn&eafional Labour Review, Z14, (3), 1976, pages B. Hughes, World Futures: A Critical Analysis of Alternative (Baltimore, Johns Hopkins, 1985a). B. Hughes, “World models: the basis of difference”, International Studies Q~rte$y, 29, (l), 1985b, pages 77-102. B. Hughes and M. Mesarovic, “Population, wealth and resources up to the year 2000”, Futures, IO, (4), 1978, pages 267-282. B. Huldt et al, Swede in a World Society, English translation (Stockholm, Secretariat for Futures Studies, 1980). ILO, Emplomt Effects of Multi-National Enterpriies in L&loping Countries(Geneva, ILO, 1981). IMF, World Economic Outlook (Washington, International Monetary Fund, 1982-85). L. Inglestam, To Choose (I Future (Stockholm, Secretariat for Futures Studies, 1973). R. Jackson and W. Allen, “The Global Political Exercise: a computer based simulation for decentralised teams”, Zntcrnufional Studies Notes, 11, (2), 1985. H. de Jouvenel, Social Change: The Next Twenty Years (Paris, Futuribles, 1985). H. Kahn and A. Weiner, The Yeur 2000 (London, Macmillan, 1967). Y. Kaya, Future of Global Znderdependence (Laxenburg, Austria, IIASA, 1977). Y. Kaya et al, “Long term projection of economic growth in ESCAP”, in UNIDO (1984), op cit. B. King, R%at is a SAM? A layman’s guide lo Social Accounting Matrices (Washington, World Bank, 1981). L. Klien and V. Su, “Protectionism: an analysis from Project LINK”,Joumul ofPolicy Mo&lling, I, (l), 1979, pages 5-36. E. Lazlo, A Strateu+r the Future: The Systems Approach to World Or& (New York, Braziller, 1974). W. Leontief and A. Carter, The Future of the World Economy (New York, United Nations, 1976). J. Lesoume, “The use of global models-synthesis”, in Levy and Robinson (1984), op cit. D. Leuddijk, World Or& Studies (Amsterdam, RIO Foundation, 1979). M. Levy and J. Robinson, Eneru and AgricultureTheir Interacting Futures: Policy Implications of Global Models (London, HarwoodZUNU, 1984). C. Lutz, Was sagen uns die Weltmocielle und Szenarien der letzten 15 Jahre? (Zurich, West Germany, Gottlieb Duttweiler Institute, 1983). R. Lynch, “An assessment of the RAS approach for updating input-output tables”, in UNIDO (1984), op cit. M. McHale, Ominous Trends and Valid Hopes: A Comparison of Five World Rgorts (SUNY Buffalo, Center for Integrative Studies, 1980). D. MacRae and J. Mula, Global Mo&ls Acccptancc in the Policy Rocws (G-MAPP) (Hawaii, EastWest Center, 1981). D. Meadows d al, The Limits to Growth (New York, Universe, 1973). D. Meadows, J. Richardson and G. Bruckmann, Groping in the Dark: T&e First Decal? of Global Modelling (Chichester, UK, Wiley, 1983). R. Merritt and B. Russett, From National Development to the Global Community (London, Allen and Unwin, 1981). M. Mesarovic and E. Pestel, Mankind at fhc Twning Point (New York, Dutton, 1974). I. Miles, “ Worldviews and scenarios’ ’ , in Freeman and Jahoda (1978), op cif. I. Miles, “Scenario analysis: identifying ideologies and issues”, in UNESCO (1981), op cit. S. Moore, Half Truihs and Conscqzznces: l7ie Legq of Global 2000 (Washington, Heritage Foundation, 1985). J. Mula and D. MacRae, GUAPP: A Review of Global Modelling Research ROJ~C~J (Hawaii, East-

FUTURES August 1987

430 Global models

West Center, 1981). D. Munton, Global Mo~ls Politics and the Future (Dalhousie University, Centre for Foreign Policy Studies, 1978). J. Naisbitt , MegatrendsTen New Directions Transforming Our Lives (New York, Warner, 1984). D. Norse, “The Interfutures Project of OECD”, in Levy and Robinson (1984), op cit. D. Nyhus and C. Almon, The INFORUM International System of Input-Output Mo&ls and Bilateral Trade Flows (Laxenburg, Austria, IIASA, 1977). OECD/Interfutures, Facing the Future: Mastering the Probable and Managing the Unpredictable (Paris, OECD, 1979). A. Onishi, “The Japanese experience with global modelling”, in Levy and Robinson (1984), op cit. K. Parker, Scenario Based Ewpoiments using SARUM (London, Technical Change Centre, 1984). B. Pollins, A Survq of Fourteen Formal Models of International Trade (Berlin, Wissenschaftzentrum, 1980), No 80-104. global modelling project”, Futures, 9, (l), 1977, pages 3-16. P. Roberts, “SARUM-a P. Roberts, “Two cheers for modelling”, Futures, IS, (3), 1984, pages 214-216. J. Round, A Review of Expe&nce in the Design and Use of Social Accounting Matrices (University of Warwick, Department of Economics, 1982), No 19. model”, in Levy and Robinson (1984), J. Royer, “Long term perspectives in the UNITAD op cit. J. Royer, Unemployment Strategies in World Regions (Geneva, UNITAD, 1985). W. Sanderson, Economic-Demographic Simulation Models: A Review of their Usefulnessfor Policy Analysis (Laxenburg, Austria, IIASA, 1981). SARU, SARUlW Han&ok (London, Systems Analysis Research Unit, 1985). B. Schwartz, Methods in Futures Studies-Problems andApplications (Boulder, CO, Westview Replica, 1982). M. Siegmarm and K. Deutsch, Modelling: An Ovtrview in Perception and Analysis of World Problems (Paris, UNESCO, 1986). J. Simon, “Political risk forecasting”, Futures, 17, (2), 1985, pages 132-148. M. Sommer, “On the applicability of econometric methods to system dynamics models”, ~~NAM~GA, IO, (ll), 1984. J. Sterman, “Integrated theory of the economic long wave”, Futures, 17, (2), 1985, pages 104131. W. Tims and J. Waelbrook, Global Modelling in the World Bank, 1973-76 (Washington, IBRD, 1982), No 544. J. Tinbergen, Reshaping the Znternationat O&r (New York, Elsevier/Amed Nations, 1976). UNESCO, Methods for ~velop~t Planning: Scenarios, Models and Micro-studio (Paris, UNESCO, 1981). UNIDO, Proceedings of theSeventh International Confnme on Input-Output Techniques (New York, UN, 1984). J. Wilkenfield, “POLIS-computer assisted international studies”, Teaching Political Science, X, (4), 1983. World Bank, World Tables--Data Files (Washington, IBRD, 1981). World Bank, World ~veLo~~t Report (Washin~on, World Bank/Oxford, 1979-85). Note The Bibliographies refer to studies described in this article and the forthcoming articles noted in the text. The author would appreciate information on these and other global forecasting studies now under way.