Are economic forecasts biased downward because of economists’ antipathy against President Trump? I don’t know the answer, but I’m pondering the question.
We economists tend to be well educated, often from prestigious universities. We range from liberal to conservative or libertarian in our politics, but we are part of what many call the elite. We have chatted with senators and governors and Federal Reserve board members. We’re quoted in newspapers and interviewed on television. We are part of the group that Donald Trump sneered at in his electoral campaign: “I want you to imagine how much better our future can be if we declare independence from the elites who’ve led us to one financial and foreign policy disaster after another.”
Opposition to the president is illustrated by Bill Dudley, former president of the Federal Reserve Bank of New York, who recently wrote “There’s even an argument that the election itself falls within the Fed’s purview. After all, Trump’s reelection arguably presents a threat to the U.S. and global economy….”
We economists generally dislike President Trump. Some of us agree with his deregulatory push or his tax policy, but hardly any of us agree with the trade war or his immigration rhetoric. As those issues have risen in prominence, economists’ opinions about the president have fallen. In addition, most of my colleagues dislike his un-presidential style.
Economists on The IGM Economic Experts Panel, for example, mostly believe that the incidence of U.S. tariffs will fall primarily on American households, and that the impact will be felt most heavily by lower income groups and regions. Textbook economics has favored free trade for centuries. This issue has strongly colored economists’ perceptions of the president’s understanding of economics.
Most presidents have had their supporters and their opponents among professional economists. As good policies get twisted by political considerations, we all groan. As bad policies are pursued for political purposes, we groan some more. We (mostly) focus on our economic analysis rather than party politics. For example, most Republican economists opposed President George W. Bush’s steel tariffs, for the same reason they now oppose President Trump’s China tariffs.
In an ideal world, economists would push their personal attitudes aside when they develop forecasts. They would reflect the impact of policies in their forecasts, but not shade down predictions just because they find the occupant of the White House disgusting. However, economists are human.
Bias in forecasts is possible because most of us tweak our forecasts judgmentally, rather than taking computer output as perfect. In this age of big data and sophisticated data analysis, judgmental adjustment of forecasts may seem anachronistic. We do that for three primary reasons.
All computer models of the economy omit critical variables. The economy is extremely complex, and our data are limited. Surveys of attitudes are only a rough guide to optimism and pessimism, and some sectors of the economy are very hard to assess. New goods and services are not easily incorporated into a model whose relationships were estimated before the new products were developed.
The second reason we adjust forecasts is that we know economic data have errors. Many critical variables are collected through surveys. For example, the unemployment rate calculation derives from a telephone survey. Sampling error is one concern, especially when we look at state or county level data, where the sample size is low. We also worry that some people may give erroneous answers, either intentionally to hide personal failings, or simply having forgotten critical information. Sometimes we economists use our judgment to make up for what we believe are errors in the data.
The third and most important reason to use judgment to adjust forecasts is that economic relationships change over time, and our statistical models don’t handle changing relationships very well. With globalization, new technology, evolving social attitudes and changing government policies, hardly any past relationship is a reliable indicator of future patterns of behavior. Some correlations do work fairly well for a while, but all should be suspect.
For these reasons, economic forecasting is a mixture of scientific analysis and gut feel. And once we start using our judgement, bias can creep in.
There may be times when bias is intentional. Arthur Okun founded the largest economic forecasting consulting firm of the 1970s and ‘80s. His staff once told me, “Arthur has a social conscience,” meaning he thought the country would be better off hearing an optimistic forecast, hopefully as a self-fulfilling prophecy. And a bank economist friend of mine told me that he was not allowed to communicate pessimistic forecasts to the bank’s customers.
Most bias, I believe, is unintentional. Despite the anecdotes above, most of my colleagues in the forecasting profession try to provide the most accurate projections possible. However, human nature colors our view of the future. Disgust over political leadership can certainly spill over into forecasts. Unfortunately, I cannot determine how pervasive this is. But it is something I worry about, especially when I see my colleagues sharing negative feelings about the president and then making pessimistic forecasts. And perhaps I do that myself.
