Lack-of-omniscience Problems


By Don Ross

Will President Biden’s thrilling fiscal plans cause serious price inflation? I’d like to know. I’m curious about and emotionally involved with the future, but there’s also the practical reason that I’d then know whether I should reduce the weight of equities in my investment portfolio. Alex Rosenberg would also like to know, perhaps for a similar mix of reasons. He is disappointed, however, that when he asks economists he finds disagreement. This leads him to doubt, as a philosopher of science, that economic theory is of any value at all.

I’ve worked in a large economics department at a world-leading university for over 20 years. I’m a microeconomist so inflation forecasting isn’t my job. In my models, which concern people choosing health-related behaviors and local governments investing in roads, I care a lot about prices but not at all about money in the sense that bothers Rosenberg. Nothing he says in his article gives me any reason to think I need to find some way to get money into my models.

But I understand the stories told by my macroeconomist colleagues who expect serious inflation from the proposed fiscal splurge, and I also understand the stories told by colleagues who don’t. I judge the latter to be more persuasive, so I’m hanging on to my equities. If I had (a lot) more space here, I could tell you why. The explanation would use a lot of economic theory. I don’t know of better resources for such reasoning. Since my wealth in retirement depends on the decision, you should believe that I really believe in the value of the economics I use for thinking, and am not just saying it because I’m motivated to defend the value of economics.

Rosenberg thinks that macroeconomists should know whether there will be inflation because they should all agree on one model and that model should include the money supply and velocity as causal variables. In fact, macroeconomists use different models for different purposes. Some – Keynesian and monetarist models – include money measures as fundamental. Rosenberg might wrongly believe that these models are regarded by most macroeconomists as obsolete. I suspect he’s misled by the recent prominence of one form of macro modeling, called “DSGE”. The reason this model has dominated attention is not because economists think it says everything interesting there is to say about a whole economy. It’s because it focuses on what central banks had power over before 2008, the timing and scale of capital investments by companies. There are good reasons for thinking that money is ‘neutral’ where those decisions are concerned, and the non-monetary models’ pre-2008 forecasts of  that  performed pretty well until the COVID-19 shock. But in the response to the financial crisis central banks were allowed to take responsibility for a wider goal, preventing risk-averse investors from running for safety out of corporate finance. For that they used a policy, quantitative easing, that depended on forecasts involving money. So now macroeconomists who advise central banks have money in their models (again).

I think that the division of macroeconomic labor between questions that mattered to pre-2008 central banks and other problems wasn’t efficient, so DSGE models received too much relative limelight among theorists. But that’s a kind of problem every economist understands well, a problem of institutional incentives. It says nothing whatsoever about the value of economic theory in general.

The reason economists find money “puzzling” in ways that anthropologists don’t is because the money that matters in economics is itself much more complicated than what the anthropologists describe. It’s not just Dollar and Euro and Yuan balances in current accounts. It’s also <em>expected </em>money that institutions create by lending and borrowing. The unlimited scope for such activity is a source of complexity in economic systems that goes far beyond what happens when you add a controlled medium of exchange to a barter system. 

And then the reason there isn’t, contrary to Rosenberg, one monolithic macro model is because a whole economy is a complex system. Complex systems can’t be modeled ‘all at once’. They can be simulated, but simulations are terrible for producing reliable generalisations, which is what forecasts require. The earth’s crust is another complex system, which is why seismologists can’t predict earthquakes. Seismologists nevertheless know much more about earthquakes than other people, and if I wanted to make my house as likely as possible to survive an earthquake, they’re the scientists whose work I’d consult. Exactly the same points apply to macroeconomists. I can’t be  sure that Biden’s boldness won’t make my retirement poorer by inflating away a bunch of my wealth. That isn’t economists’ fault; it’s just the way the world is. But thanks to the knowledge produced by economists, I can make a scientifically informed bet.

About the Author

Don Ross is interested in economics; economic methodology; experimental economics of risk and time preferences, addiction and impulsive consumption; gambling behaviour; addiction policy; cognitive science; game theory philosophy of public policy; philosophy of science and scientific metaphysics. 

Alex Rosenberg's Money Problems is here

EJ Spode's response to Rosenberg is here

Diane Coyle's response is here