Why Model-Based Forecasting?
Start with a Paradigm
Models start with a paradigm, are easily explained (without any equations), and provide an integrated and internally consistent view of how the economy works. Models are self-documenting in that the key relationships in the economy are made explicit. As such, models provide a road map for thinking through both simple and complex macro issues.
Tell a Story
A model-based forecast tells a story. The model allows us to identify the key forces that are driving the economy: why consumer spending and business investment are strong or weak; why inflation is rising or falling; why monetary policy is tightening or easing; what is determining the shape of the yield curve, and so on. We quickly found that most of our clients didn’t want to sort through computer output for the hundreds of variables in our model over the next twelve quarters (or more). They wanted to understand why; they wanted stories. And we responded by becoming the best story tellers around.
Consider Alternative Scenarios
Models allow us to quickly design and simulate alternative scenarios, based on differing assumptions. For example, these alternative simulations give us insight into how alternative oil price paths, significant changes in fiscal policy, or changes in other key assumptions produce importantly different outcomes. This allows clients using our model to produce their own forecasts and their own “what-if” scenarios. Recent areas of heightened concern for which the model has been useful include the effects of fiscal stimulus, spillover effects to the U.S. from a benign (or not so benign) outcome of the European sovereign debt crisis, and the long-run economic effects of the government failing to take steps to ensure a sustainable path for deficits and debt.
Ensure Internal Consistency
A model enforces internal consistency on forecasts and alternative scenarios. That is, it ensures that the paths of output, inflation, employment and interest rates are internally consistent, consistent with the paradigm underpinning the model, and therefore comprise a coherent story.
Understand Historical Regularities and Incorporate Judgment
A model grounds forecasts of the future on regularities of the past. But, for us, model-based forecasting is about both the model’s predictions and our own judgment. When the underlying structure of the economy is changing quickly we need to update the model or judgmentally manage the model to account for such change. We have always said that forecast excellence is one part science (the model), one part art (judgment) and one part luck.
Learn from Our Errors
Finally, a model-based approach allows us to learn from our errors. We make a lot of them, of course. The best model-based forecasters are able to exploit their past errors to make better forecasts in the future.