Money, Media, Machines
It’s a New World in All of Them!
My Latest Podcast Episode
My most recent guest on Brave New World is Josh Tucker, who is professor of politics at NYU. Josh has been studying things like polarization and news consumption in the world of social media for a long time using large amounts of data from the major platforms such as Facebook and Twitter (now X).
Social media has transformed how news is created and consumed. It has had all kinds of unintended side effects that we’re trying to understand, such as its impacts on polarization and mental health. Should we worry about who controls what content is presented to us and how it is presented? Should the state be involved in putting guard rails in place? Or should we let private actors and the market to figure it out? There’s no easy answer.
Josh and I had a brisk discussion about what the data are telling us so far about the impact of algorithms on politics and polarization, and how to think about the implications of who designs and controls them.
So, check out my episode with Josh:
Money Makes The World Go Round
Last week, I invited macroeconomist Paul Sheard to talk to my class about Money. Paul was the chief economist at Standard and Poor’s for many years, and prior to that chief economist for Lehman Brothers in Japan in the early 2000s. That’s when the Central Bank of Japan introduced “Quantitative Easing” (QE) to the world of economic policy making. The US followed suit a few years later during the great financial crisis (GFC) of 2008-2009, and subsequently with COVID. QE has now become a common Central Bank tool across the world.
Paul has written a wonderful book called The Power of Money, which defines money, and how governments and central banks create and control it. Paul was a guest on Brave New World last year when we discussed his book. What I find unique to Paul is his balance sheet approach towards illustrating how the assets and liabilities of governments, central banks, and commercial banks can be altered to nudge the economy in desired ways. Traditionally, central banks lower short-term interest rates to stimulate the economy, but run out of this ammunition when rates go to zero, and in a few cases, negative! So, QE is a stimulative tool central banks use when they can’t lower rates any more, but what they can do is to adjust – in this case, expand – their balance sheets by buying assets in the open market. As Paul points out, QE involves the central bank forcing an “asset swap” on banks, but you’ll have to read Paul’s book for a full exposition of what this really means and how it works. It’s fascinating.
Central banks are a financial innovation of the 20th century. Over the centuries, a critical lesson we’ve learned is that politicians can’t be trusted with money. A central bank is the government’s bank, but it is staffed with expert professionals who are supposed to be independent of politics. A critical role of the central bank is to place shackles on the government, so that politicians won’t do dumb things like print money like they often did in the past, causing rampant inflation. In their book called The Narrow Corridor, political economists Daron Acemoglu and James Robinson describe how strong liberal democracies avoid despotic governments by shackling State power to ensure that the needs to the citizenry are met by credible rule-following institutions. The role of the central bank is to shackle the government from screwing up the economic system, and to step in when necessary to maintain economic stability.
AI as the Central Bank
During our podcast conversation, I asked Paul whether an AI might be able to do the work of a central bank. Such a question would have sounded preposterous a couple of years ago. But now that we’ve seen how a machine can ingest every conceivable piece of data out there and learn from it, why couldn’t AI figure out a model of the economy from the data, instead of requiring human experts to come up with the theory to model such a complex system? I’m not knocking the need for good theory, but questioning how such theory is created in the era of data and AI.
To my surprise, Paul didn’t think it was an outrageous question. After all, what does a central bank do? It has a model of the economy, which has been carefully crafted over decades. It waits for signals from the economy, which arrive after a lag. It analyses the signals, and takes an action with a future intended consequence. Actions take time to play out. These lags, coupled with the complexity of the economy, create a constant guessing game for the central bank. Scores of experts digest the latest research and data to model this complexity.
These models have deep human biases that have accumulated over time as they have been tweaked by the professionals. Their opacity makes them fragile. A small tweak to a model might produce a different reported number in, say, inflation, which is scrutinized very carefully by the world. Indeed, a recent New York Times article reported how an economist at the Bureau of Labor Statistics set off a firestorm by sharing how an obscure change in the method for how the government calculates inflation could have led to an unexpected jump in the Consumer Price Index in January of 2024. The more vexing part for market watchers was whether the changed numbers produced by the new method going forward would be more inflated, which could make the Fed more cautious about cutting interest rates in the future. All because of a small change in how inflation is calculated.
Might AI help us better understand the data and design better economic policy interventions?
Last week I had lunch with some central bankers who wanted to talk to me about AI. They were interested in discussing how they should think about the risks and opportunities associated with the increasing presence of AI in financial markets. I asked them the same question I had asked Paul: could AI do the work of the central bank? Over the course of the meal, they realized that an AI would indeed have a lot to offer in terms of connecting the dots better than humans can. The dots would include historic data on all past central bank monetary policy decisions, past communication and forecasts, all past and contemporary data available at the time, and all literature published about the subject over time. In principle, we felt that the AI might do a better job of modeling the economy. At the very least, its models should be compared to the ones currently in use.
But we’d still need humans to blame, they concluded. After all, you can’t get mad at an AI if it’s wrong.