The Art of Economics

Here is a very good discussion of the limits of economics and the current fight in the US over the stimulus. This is by Sherry Cooper, the chief economist at Bank of Montreal:

Many fear that mounting deficits and debt will trigger inflation in the future and call for a Fed exit strategy; others are now clamouring for additional fiscal stimulus. While some are railing against the budget deficits in both the U.S. and Canada, pointing out that government forecasts of deficit reduction in coming years are far too optimistic, others suggest that more (or different) fiscal stimulus is needed. These bipolar judgments are correct and incorrect. The world is far more complex than bipolarity suggests.

Economic analysis and forecasting have been constrained by the ability of statistical analysis to aptly handle uncertainty. As the study of economics has become more and more mathematical, we fooled ourselves into believing that it was a physical science rather than a social science. In physics, you can model outcomes with a clear cut degree of precision. In economics, we deal with human behaviour in the face of uncertainty, where outcomes cannot be known with certainty because the system is far too complex.
Even the most sophisticated computer models failed to encompass the risk involved in structured instruments like credit default swaps and CDOs, let alone the systemic risk they could trigger.

At some point in the postwar period we forgot that our economic and financial models were just that—models and not the real thing. The real thing, the economy and financial system, is far too complex to analyze with predictive precision. So we made simplifying assumptions (about the uncertain distribution of risks) so that we could use statistical techniques to create models of the economy that are simple enough for a computer to analyze and forecast. These models are schematic descriptions of the economy that account for its known or inferred properties and may be used for further study of its characteristics. But they are much simpler than the true economy because they cannot account fully for uncertainty. These models have enormous theoretical and analytical value. They can test the impact of alternative phenomena assuming that all other things are held constant (ceteris paribus). They test alternative theories of economic behaviour and they can analyze the impact of alternative policy initiatives. But they are used, as well, to forecast the economy and that is where they fall short for reasons outlined earlier. That is why most economists rely on more than what their computers spit out. The dirty little secret of practicing economists (at least the good ones), as opposed to the ivory-tower types, is that we know the models are inadequate to forecast the real world.

The Fed’s economists work on econometric modeling and forecasting, but they always use what they call ‘judgmental adjustment.’ And while the human brain is often better at forecasting in the face of uncertainty, it, too, is inadequate to the task, by definition. There wouldn’t be uncertainty if we could predict the future with certainty.

Traditional economic forecasting models failed to predict the financial crisis. These statistical techniques assume that the relationships of the past will hold in similar ways in the future. Put simply, economists draw a line between two data points and use it to predict the future. (Some even draw lines in any direction that suits their theory.) This, by definition, does not account for unprecedented situations or the financial innovations that alter the system.

It is premature and erroneous to suggest that the recent stimulus programs ‘are not working’. Yes, the unemployment rate continues to rise, but how much higher would it be had no stimulus been introduced?

There is a significant recognition and implementation lag for any program, and apparently only a small portion of the $787 billion American stimulus package has been put to work. Emergency extension of unemployment insurance programs is in effect, no doubt
cushioning the blow to the rising number of long-term unemployed. Monetary stimulus has helped to ease the credit freeze and bring interest rates down to levels lower than where they would otherwise be. Treasury yields have risen in response to the sprinkling of positive economic data and the inflation fear generated by increasing bond supply, but they arguably would have risen even more had the Fed not been in the market buying Treasuries.

...

The Bottom Line: The math does not currently exist to model the full complexity of our uncertain world; we can only iterate towards a solution through trial and error. The Treasury has repeatedly modified its programs to support banks and high-risk families and will continue to do so. The greatest risk to this process is political. My greatest fear is we do not have the political will to accept the trial-and-error nature of this process, and to act exigently when adjustments are needed. There is no room for partisanship and there is no room for the blame game.

She hits the right notes.

• Models have their limits and they are particularly unuseful in unpredented eras.


• The biggest threat to recovery is politics. Managing political will through a risky, uncertain, learn-as-you-go recovery will be the most difficult task.


• Economic forecasting requires a judicious use of educated guessing since you cannot rely just on economic models.