Around this time in an election year many people anxiously await economic data releases, hoping to uncover some fact that can be twisted to advantage. So, it is helpful to explain how economists try to understand these data.

Economic forecasting almost exclusively uses large macroeconomic models. These are simply series of equations, often several hundred, which outline relationships in the economy. For example, we know that the demand for new cars will be dependent on several things; gas prices, interest rates, consumer expectations, household incomes, and historical purchasing patterns. These factors allow us to hypothesize an equation for cars sales.

The large model attempts to capture these types of relationships across the economy in such things as diverse as individual household decisions to retire to the changing price of gas. Many of the equations will depend on other equations. For example, interest rates in the car equation will be influenced by actions of the Federal Reserve, which may be influenced by gas prices and consumer expectations. So, these equations work together to offer a model of the real economy.

Economists use actual data to calibrate the model. This is applied statistical work, which often incorporates the type of non-linearities associated with chaos theory. In this way we can nail down the relationship between say, gas prices and new car sales to an actual number.

Once that is done, the forecast involves solving a few hundred simultaneous equations, which is nothing more than the Algebra I taught in every middle school in Indiana. This is what we do each year at Ball State, producing national, state and regional forecasts.

Forecasting is a costly enterprise so much of the valuable analysis lies in interpreting the data releases between forecasts. The process of the forecast gives us actual relationships to give us the most likely values for things we care about, like unemployment, incomes, population growth and the like. The forecast also gives a likely range of outcomes. Once new data is released, an economist can compare those numbers with the forecast to see how well the forecast performs.

All forecasts are wrong, but understanding why they are wrong is useful. In fact, the most important part of evaluating the forecast lies in examining the errors. Some of this bleeds into media reports; for example, as the unemployment data are released each month, they are compared to forecasts.

Most of last year’s economic forecast called for faster GDP growth, less inflation and higher unemployment rates than we have seen. The poor GDP growth, modestly growing inflation and much lower unemployment rates send mixed signals to forecasters. Listening to mixed signals is also important.

When the economy is getting much better or much worse, then the trend is obvious even if there are occasional mixed signals. When the economy changes little then almost every piece of new data can be confusing. That is where we now find ourselves, in a very slow growing national economy with poor job growth, weak incomes and a hint of inflation. That won’t change in the coming months no matter what the pundits say.­