Recovery from Economic Collapse: Insight from Input-Output Models and the Special Case of a Collapsed Oil Producer

Working and discussion papers
March 2004
John Roberts

This paper presents evidence that recovery from economic collapse in poor countries has been on occasion rapid once the causes of collapse are removed. It argues that demand-side stimulus plays a strategic role in ensuring swift recovery, alongside the removal of supply-side constraints. In this perspective, the paper considers the case of post-war recovery in Iraq.

Economic collapse is very often associated with the collapse of political and social institutions, and quite commonly also with the breakdown of civil peace and order, in other words with state collapse. This, however, is not always the case. Economies also collapse because harmful policies are pursued over long periods of time, and/or because rigorous economic sanctions are applied by trading partners and capital markets. Economic collapse is here defined as a prolonged decline in per capita income (or GDP).

Prevailing view

The prevailing belief is that recovery from economic collapse, following the removal of its primary cause(s) is likely to be delayed because of the inheritance of the dysfunctionalities, disorder and administrative decay dating from the previous period. This belief has been confirmed and reinforced by econometric evidence, using the Collier-Dollar model of aid effectiveness, that aid is ineffective in the first three years of recovery, but becomes strongly effective for a few years thereafter when absorptive capacity rises above normally expected levels. In this second phase the most serious administrative, security and physical bottlenecks have been removed, and underutilised productive capacity is put back in service. Rates of return on capital may be particularly high in this second phase in countries which have experienced major destruction of their physical capital stock and loss of human capital.

The agenda of recommended actions to hasten economic recovery is supply-side focused, with an emphasis on the restoration of civil institutions, the reconstruction of infrastructure and the provision of supplies with which to revive production.

However, there are stages in the process of recovery when demand-side stimulus may also be valuable to hasten the re-employment of available, but underemployed, resources. This paper argues that cases can readily be identified in which an early application of demand side stimulus can hasten economic recovery and that the constraints on revival arising from supply-side bottlenecks, though real, can be exaggerated.

Experiences of eight low-income countries

The post-collapse experiences of Cameroon, Cambodia, Ethiopia, Mozambique, Nicaragua, Rwanda, Uganda and Zambia are reviewed. These countries suffered prolonged periods of at least a decade of falling per capita income for reasons to do with conflict, economic policy, falling terms of trade, natural resource depletion and economic sanctions, and have a subsequent track record - dating from 1985 to 1995 - of post-collapse performance.

In all but two cases the first three years of recovery saw positive per capita income growth. The pace of recovery is not obviously related to the prior extent of collapse or to its duration. Of much greater apparent significance in explaining the post-collapse rate of recovery is the growth of real final demand - government consumption expenditure, investment expenditure and exports. Nicaragua and Zambia were unable to increase their per capita incomes in the immediate post-collapse period because of their inherited macroeconomic and external financial situation, and the need to rein-in government expenditure. The recovery paths of seven countries have been inconsistent with the thesis that economic revival accelerates after the initial three years of the recovery phase.

Case of Iraq

Iraq's GDP declined steeply in the first half of the 1980s because of way with Iran, revived briefly in the second half, and then plunged to new depths in the early 1990s with the imposition of rigorous UN sanctions following Iraq's invasion of Kuwait. Per capita GDP (in 2000 US $) fell from a peak of nearly $3000 in 1980 to a trough of below $ 500 in 1991. It then revived to close to $1000 in 2000 and 2001, with the benefit of imports of food and other supplies to run infrastructural and other public services permitted under the UN Oil-for-Food programme. It has subsequently declined again, with lower oil exports in 2002, and sharply lower oil exports in 2003 in the aftermath of the war of March-April 2003.

Iraq has a mainly urbanised society, and a domestic economy geared towards the production of goods and services needed to meet the effective demand created by oil revenue-financed expenditure - by government, by the wider public sector and by public sector employees. Sanctions, first prohibiting the export of oil altogether, then placing the financial proceeds in an offshore escrow account available only for the purchase of imports under strictly supervised conditions, deprived the domestic economy of its financial lifeblood. Though public service employment was largely maintained wages and salaries lost most of their purchasing power, and effective demand for producer and consumer goods evaporated, leading to under-and unemployment in state and private enterprises. The economy experienced not only supply bottlenecks - which were alleviated by the Oil-for-Food programme - but also, and crucially, severely deficient demand.

With the abrogation of sanctions in June 2003 revenues from oil will flow once again through the public accounts, and effective demand for domestically supplied goods and services should revive.

Economic structures and policies for supporting recovery

Most of the eight low-income countries are predominantly rural, with large, informal, rural sectors, relatively small informal urban sectors, and a formal non-rural economy in which the government plays a large role, alongside relatively small industrial and mining activities. Only Zambia, which is relatively urbanised and where industry was economically twice as large as agriculture during the years of decline, does not conform to this archetype. Economic collapse in the mainly rural countries is neither sudden nor total because of the resilience of the informal sector. The formal sector usually suffers worst, reducing government revenues, and causing decay in government services.

'Urbanised', extractive industry-dependent, economies like Iraq and Zambia, with formerly higher per capita incomes and high levels of tax-financed public expenditure are less well hedged against adverse shocks to their formal sector earnings. Their informal agricultural sectors are relatively smaller and a higher proportion of the employed population is directly or indirectly dependent on public expenditure for its livelihood. Dutch Disease, and the domestic market focus of much of their non-mineral tradable goods output, inhibits the early and easy diversification of their export earnings.

Conventional policy prescriptions for economic recovery and for recovery-oriented assistance emphasise, correctly, physical security, rehabilitating infrastructure, resettlement, reviving agriculture and formal sector production through input provision, restoring state institutions and restoring/preserving macroeconomic stability. However, this strategy is wrong to imply that supply-side actions are not only necessary but also sufficient to promote recovery because it omits complementary demand-side action.

Modelling economic recovery

Simple linear models are constructed to represent, illustratively, the typical low-income economy in collapse and the case of Iraq. Such models, though once popular, are now correctly deemed inappropriate to represent the processes of economic growth. However, they can yield valuable insights into the processes of recovery if, and so long as, there are underutilised productive resources, and if supply elasticities are consequently high, and provided that major supply-side bottlenecks have been removed.

In constructing linear models appropriate to countries starting recovery the question arises whether to use pre-collapse or post-collapse intermediate consumption, final demand and value added coefficients. Pre-collapse coefficients are likely to be the more relevant for charting the effects of rising demand on economic recovery because agents are striving, in the phase of recovery, to restore transactional patterns with which they are familiar from the past. In a first approximation, the characteristics of the pre-collapse economy can be assumed to typify the economy in recovery, certainly as the recovery reaches a mature phase.

The construction of statistically sound input-output tables is data intensive and laborious, and most economies in collapse are unlikely to have such tables. However, for broad-bush policy analysis purposes, it is possible to construct simple ad hoc linear models from readily available sources, augmented if need be by coefficients borrowed from other countries.

The illustrative numerical model constructed for the typical low-income country case has a relatively modest multiplier effect from rise in government expenditure, and a rather stronger effect from an equi-proportionate rise in export earnings. If increases in actual government consumption, investment and exports experienced in the six 'general case' countries are applied as shocks to the illustrative model, the model reacts with GDP increases which, in five of the six countries, approximate the actual rates of growth experienced by them in their initial phase of recovery. The model is less predictive of actual growth in the mature recovery phase. This result suggests that demand-side stimuli are particularly important in promoting growth in the initial phase.

If this exercise is performed in the planning phase, before major interventions to remove supply-side bottlenecks, it can highlight transactional linkages which must function properly if recovery is to proceed, and thus help to identify priority projects.

An illustrative model for Iraq is also constructed, loosely based on what is known about the structure of the economy before its precipitate collapse in the 1990s. A high arc elasticity of supply is assumed on account of the high level of current unemployment affecting inter alia skilled and educated personnel, and pressure within the international community to make rapid restoration of productive and infrastructural capacity. This model, by construction, shows that the indirect and induced effects of government expenditure, with its strong backward and forward linkages, are significantly stronger than those of exports and investment spending where linkages are weaker. A 10% rise in government consumption yields a 3.5% increase in GDP, and a possible increase in employment of 117 000.

Using this tool a partial economic recovery scenario is constructed, making realistic assumptions about the revival of oil production, the value of oil exports, and assuming that Exchequer receipts of oil revenue are used to finance government expenditure. It holds out the prospect that per capita GDP (in 2002$) may rise by 50% from $830 in 2002 to $ 1225 in 2005 or 2006.


The main conclusions drawn are that:

# economic recovery after collapse - or prolonged decline - is not necessarily delayed by as long as inferences from cross-country analyses suggest,

# demand-side stimulus plays an important supporting role in promoting recovery, alongside action to remote supply-side restraints, because of the presence of underutilised but productive resources; increased public expenditure can therefore be helpful in accelerating recovery - if it can be financed without violating macroeconomic constraints and applied without undermining structural reforms,

# simple linear models are usable as a tool for simulating the likely effects of demand-side stimuli, and of identifying the economic linkages that need to be strengthened, and the supply-side bottlenecks that need to be removed, if the recovery process is to proceed smoothly and rapidly.

Read the research