Who Forecasts Best? An Empirical Analysis of Canadian Private-sector Forecasters

Governments in Canada, whether at the national or provincial level, believe that forecasting key macroeconomic variables is important enough to justify building their own macroeconometric models rather than relying on an independent service to provide the forecast. Of course, forecasting is but one aspect of model-building activity. Governments can also use models to measure the impact of policy changes on the economy. Private sector companies, by contrast, have diverse information needs. Macroeconomic variables which are very important to some firms may be less important to others. Many firms want to have an idea of where the economy is headed, but do not feel that they need to invest in a forecasting capability of their own. As well, many decision makers are interested in economic variables over the very short term rather than the quarterly or annual forecasts generated by forecasters. In this broader context, forecasting in itself can involve rather disparate activities. The systematic perusal of newspapers or newsmagazines may make some company analysts sufficiently confident that they have a handle on the future direction of macroeconomic variables of interest to their company. Other companies make somewhat more rigorous forecasts by using the increased power of computer spreadsheets to exploit perceived relationships among variables. A small number of consulting firms provide for their clients forecasts of a large number of macroeconomic variables based on mediumor large-scale macroeconometric models. Even with this approach, as in the less rigorous models, the forecaster can exercise a fair amount of judgement in framing the forecast for individual variables. For example, a variable may depend on an equation which has been estimated over a period in the past. The equation is then fitted to the past and an “add factor” between actual and estimated values is created. In order to forecast the variable, the forecaster has to make some assumption about how this add factor will behave in the future. The article considers the following issues. Are some firms’ forecasts more accurate than others for certain variables? Can some firms be said to be optimistic in so far as their forecasts are consistently higher than their competitors? As well, we investigate whether the forecasts themselves are biased or exploit available information. It should be mentioned that whereas there is a tradition of private-sector forecast evaluation in the United States, particularly of the Blue Chip Indicators1 — to our knowledge there has been little investigation of individual Canadian private-sector forecasters.2 We also assess the performance of the private-sector forecasters relative to the consensus forecast, defined as the average forecast. In turn, the consensus forecasts are compared with Department of Finance forecasts to ascertain whether government forecasts make good use of, or encompass, the consensus forecasts.