Macroeconomics

The Economist Intelligence Unit produces numerical forecasts for 150 countries, 117 of which are to a very high level of detail (up to 300 series per country), and headline numbers for the remaining countries. The forecasts are produced on spreadsheet-based models, developed by our in-house economics team. Each country model uses the same basic framework, but is customised to suit features specific to the individual economy: for example, the importance of the energy sector; an unusual exchange-rate regime; or the non-availability of particular data series. All the equations within the model have the same basic form across countries but are, of course, customised to reflect each individual country's characteristics. The models are used for forecasting inflation, interest rates, prices, gross domestic product from both the expenditure and supply side, trade flows, the current account, capital account, international debt and government finances. All these sectors are interlinked and consistency is imposed within the model.

Each country model is closely linked with the forecast for the rest of the world. Each forecast is based on our global assumptions (a snapshot of our global forecast including forecasts for 40 commodity prices, estimates for manufactured good prices, developed world exchange and interest rates). In addition, exports for each country are related to our import demand forecasts for that country's top 20 markets; export and import competitiveness is assessed with reference to competitor countries' trade prices; the value of foreign-currency debt is revalued every year with reference to international exchange-rate movements; debt interest payments are pegged to movements in developed-country interest rates; and global capital inflows and outflows are reconciled. The aim is to ensure that all Economist Intelligence Unit forecasts are fully consistent-that the forecast for Thailand is based on exactly the same global backdrop and is fully consistent with the forecast for Greece or Saudi Arabia.

However, it is important to realise that using even a very structured economic model still requires considerable expert judgement. Models can be used as "black boxes", with raw data being turned into forecasts with little input from the forecaster. But the results will be poor, as models are only ever an approximation of the real world. Therefore, our country experts spend a considerable time working with the model: for example, checking the output, and making judgements about policy, the likely path of consumer and industrial sentiment, and the path of the exchange rate, among others. To produce a high-quality forecast requires not just the best models, but the best country experts to run these models.

Back to forecasting and methodology