The OECD’s forecasts combine expert judgement with a variety of existing and new information relevant to current and prospective developments. These include revised policy settings, recent statistical outturns and conjunctural indicators, combined with analyses based on specific economic and statistical models and analytical techniques, as outlined below.
Assessing the current situation
An important starting point in the forecasting process is the re-assessment of the economic climate in individual countries and the world economy as a whole. Here, a combination of model-based analyses and statistical indicator models play an important role in "setting the scene" at the start of each projection round.
A first step is to look at the range of relevant new information since the last projections were produced - such as changes in commodity prices (in particular the oil price), exchange rates and interest rates, fiscal trends, the path of economic activity and other key variables – to see how the recent past has developed differently from what was previously expected. With this new information, and using the previous set of projections as a starting point, the effects of the new elements and revised judgments are typically assessed on the basis of model simulations using the NIGEM global model and short-term indicator models. Thus the likely impact of combined and individual changes in assumptions and new information on key aggregates can be assessed in consistent fashion for each of the major economies and economic groupings. These results are mechanical and therefore intended to be no more than a guide to the informed judgments of country and topic experts on the underlying “forces acting”.
The use of indicator models
For the euro area and individual G7 economies, the near-term assessment also takes particular account of projections from a suite of statistical models using high frequency indicators to provide estimates of near-term quarterly GDP growth, typically for the current and next quarter or so. This analysis builds on the work of Sédillot and Pain (2003) and Mourougane (2006) in using short term economic indicators to predict quarterly movements in GDP by efficiently exploiting all available monthly and quarterly information. These models typically combine information from both "soft" indicators, such as business sentiment and consumer surveys, and "hard" indicators, such as industrial production, retail sales, house prices etc. and use is made of different frequencies of data and a variety of estimation techniques. The procedures are relatively automated and can be run whenever major monthly data are released, allowing up dating and choice of model according to the information set available.
The most important gains from using the indicator approach are found to be for current-quarter forecasts made at or immediately after the start of the quarter in question, where estimated indicator models appear to outperform autoregressive time series models, both in terms of size of error and directional accuracy. The main gains from using a monthly approach arise once one month of data is available for the quarter being forecast, typically two to three months before the publication of the first official outturn estimate for GDP. For one-quarter-ahead projections, the performance of the estimated indicator models are only noticeably better than simpler time series models once one or two months of information become available for the quarter preceding that being forecast. Modest gains are nonetheless to be made in terms of directional accuracy from using the indicator models.
Statistical indicator models are nonetheless limited in their ability to forecast quarterly GDP growth. Even with a complete set of monthly indicators for the quarter, the 70 per cent confidence bands around any point estimate for GDP growth in that quarter lie in the range from 0.4 to 0.8 percentage points, depending on the country or region and the degree of uncertainty is found to widen as the forecast horizon lengthens. Forecasting errors can also arise for a variety of reasons, including revisions to the initial published data and inaccuracies in the projections of the incoming monthly data.
Regular indicator model-based estimates of GDP now feed into both routine Economic Outlook assessment exercises and interim analyses and forecast updates released to the press on a routine basis.
While the OECD's world trade forecast is built as the aggregation of individual country import and export forecasts, additional tools are used to assess the short term evolution of world trade and its consistency with the GDP growth projection. Firstly, indicator models to forecast world trade in the short term have been developed from the techniques used for short term forecasting of GDP growth to allow the incorporation of the most recent information from key monthly trade indicators. This approach includes a bridge equation model based on a limited set of variables (world industrial production, export orders for the G6 economies, 2 technology indicators, oil prices and the Baltic dry index) and a dynamic factor model using an extended dataset (including a larger number of monthly series at world and country levels), see Guichard and Rusticelli (2011). These models are used routinely during forecasting rounds and also for interim analyses. Secondly, a global equation linking world trade growth to world GDP growth is used to assess the consistency of world trade and world GDP forecasts drawing on the work of Cheung and Guichard (2009). To the extent that possible inconsistencies might be identified, this information is used iteratively in guiding the more detailed forecast components at country and regional levels.
The above use of statistical regression techniques relating GDP or world trade growth over the economic cycle to short-term indicator series contrasts with the longstanding approach used to produce the OECD Composite Leading Indicator series (CLIs). The latter are typically constructed for each country using a set of 5-10 variables that have been observed to be closely related to past turning points in a cyclical reference series such as GDP or, more typically, industrial production. Both techniques have different roles to play in the OECD’s assessment methods.
Key variables and relationships
In making the overall assessment of current and future economic performance in individual countries, a number of key variables and relationships are examined, broadly along the following lines:
Investment income receipts and payments are set to reflect returns on stocks of external assets and liabilities, while international transfer debits and credit are exogenous, subject to consistency checks across countries. An important feature of the trade and balance of payments exercise is the need to ensure consistency across countries and regions and iterative procedures for maintaining balance at the world level. A global equation linking world trade growth to world GDP growth is also used in checking the consistency of the world trade trajectory for given world activity, see Cheung and Guichard (2009).
The OECD’s Forecast Entry system
The OECD’s forecasting process is greatly assisted by a purpose-built Forecast Entry system which both centralises the forecast data management process and allows individual country experts to view most recent data outcomes, new information and assumptions and revise their projections in a consistent manner, also taking account of in-built policy rules and equation-based estimates for key variables, such as inflation, trade volumes and prices, etc. At the same time, the system maintains the consistency and coherence of the data set by incorporating all the relevant National Accounts, trade and other accounting identities linking the various concepts. Thus as individual forecast components are updated and submitted, all identities are automatically re-evaluated to provide a fully consistent data set. The underlying data base is maintained and updated continuously through the forecasting round by the centralised Analytical Data Base team, which also prepares associated data sets for publication.
The Forecast Entry system also provides an efficient means of managing and monitoring the overall shape of the forecast, by country and economic region, through a series of purpose-built tabular and graphic outputs. These are used intensively in the production process and also form the basis of corresponding documents prepared for internal, committee and final publication uses, including the various Economic Outlook country specific and cross-country Annex tables and charts.
For macro-economic assessment in the context of the Economic Outlook, the OECD uses the NiGEM model of the British National Institute of Economic and Social Research is an estimated model, which uses a ‘New-Keynesian’ framework in that agents are presumed to be forward-looking but nominal rigidities slow the process of adjustment to external events.
A policy-advice model, NIGEM is also designed to be flexible where assumption on behaviour and policy can be changed. Agents can be assumed to look forward in some scenarios, but not in others. Financial markets are normally assumed to look forward and consumers are normally assumed to be myopic but react to changes in their (forward looking) financial wealth. Monetary policy is set according to rules, with defaults designed for speed. However, interest rate feedback rules can be changed, and their parameters adjusted.
The structure of the NIGEM is designed to correspond to macroeconomic policy needs. NiGEM is a structured around the national income identity, can accommodate forward looking consumer behaviour and has many of the characteristics of a Dynamic Stochastic General Equilibrium (DSGE) model. Unlike a pure DSGE model, NiGEM is based on estimation using historical data. It thus strikes a balance between theory and data and enables using the NIGEM both for policy analysis and forecasting.
Most countries in the OECD are modeled separately. The rest of the world is modeled through regional blocks: Latin America, Africa, East Asia, Developing Europe, OPEC and a Miscellaneous group mainly in West Asia. All models contain the determinants of domestic demand, export and import volumes, prices, current accounts and net assets, and the OECD countries are more complex than those of the non-OECD countries.
The core of each of these country models consists of a production function determining output in the long term; a wage-price block; a description of the government sector; consumption, personal income and wealth; international trade; and financial markets. We use a dynamic error-correction structure on the estimated equations, which allows the model to adjust gradually towards equilibrium in response to a shock. In some cases the speed of adjustment will depend on expectations as well as distance from equilibrium.
Linkages in NiGEM take place through trade and competitiveness, interacting financial markets and international stocks of assets. The model is homogeneous in exchange rates, and exports demand equals imports across the world. Competitiveness acts as an important stabilising feedback on the model, as shifts in the domestic price level or the exchange rate feed into relative trade prices, allowing net trade to offset shifts in domestic demand.
In assessing the fiscal situation of Member countries, the OECD uses a wide range of indicators over a period of several years, since looking at one concept for a single year could give a distorted picture, given changes in economic conditions and special one-off factors.
More specifically, the cyclically-adjusted budget balance represents what government revenues and expenditure would be if output were at its potential level. In evaluating the stance of fiscal policy it is also useful to correct the cyclically adjusted balance for interest payments on government debt since these payments do not represent discretionary spending items. Thus, the primary cyclically-adjusted budget balance is derived by adding back net interest payments to the cyclically-adjusted balance. Changes in the primary cyclically-adjusted balance can then be used as a rough indicator for changes in discretionary fiscal policies.
Special attention is also paid to the general government's consolidated gross financial liabilities which measure the total debt held outside the government's accounts and provides an indicator of the likely future debt servicing burden of the economy. It should be noted that measured debt does not give a complete picture of debt servicing burdens, as it generally excludes contingent liabilities and financial assets (i.e. pensions, health care, deferred taxes) and the value of the government's real assets. The "true" value of the government's financial assets is also often difficult to gauge (e.g. government loan programmes and holding of shares in state owned enterprises). Nevertheless, the size of government debt – both gross and with financial liabilities netted out - is a key variable for estimating and evaluating issues related to fiscal sustainability and the room of manoeuvre for fiscal policy.
These concepts are explained in detail in the Notes to the Annex Tables of the Sources and Methods document.
Last updated: December 2011