Here’s another article I want to recommend strongly. It’s an op-ed in the New York Times by Raj Chetty, John N. Friedman and Michael Stepner on the stark differences in the present recession between “high-income workers” and “low-income workers”. Here’s a snippet:
Policymakers are debating new stimulus payments. As they do so, we are able to track the effects of the $600 checks paid in the last stimulus bill. Within just three weeks after the checks were deposited, low- and middle-income households immediately spent a significant portion of the money they received. But our early estimates show that households making more than about $75,000 per year have spent less than $45 out of the $600.
The staggered nature of the current stimulus payments — $600 in early January, $1,400 now under consideration — illustrates how this real-time data can aid policymaking. Based on our analysis of the impacts of the $600 checks, we predict that sending higher-income households $1,400 would cost the government $200 billion — but generate only $15 billion in immediate economic activity. Hence, targeting stimulus checks to lower- and middle-income households and using the money that is saved for programs to support those who need the most help is likely to yield greater economic benefits.
That passage and, indeed, the entire piece just filled me with ideas. For example, helping the needy isn’t as simple as voting Democratic rather than voting Republican. It doesn’t just require data but being open to the direction in which the data lead.
In the piece the authors call for more real-time data-based policy-making:
When politicians lack clear answers to such questions, policy choices often rely on intuition and political ideology rather than scientific evidence. The indicators that we currently use to guide policy decisions — statistics like gross domestic product and unemployment rates — are collected by surveying businesses and households. They provide vital information about the economy as a whole but are often insufficient to guide policy in the moment. The surveys take weeks or months to gather and release. Moreover, they are limited in how finely you can zoom in on the experiences of specific cities or subgroups of Americans.
But we live in the age of information, where virtually all economic transactions leave a digital trail — from credit card receipts to paychecks to loans. These data are routinely used by companies and financial analysts to make better business decisions. And when the same data are put in the hands of the public, they can be used to guide our most important policy decisions, too.
without recognizing that the actual conclusion to which that drives one is that what is needed is more pragmatic rules-based policy-making that won’t be skewed by politics.
There’s another problem. The data sources used by the authors capture at most 85% of the economy. It doesn’t include the cash economy at all and that’s a sizeable chunk of total economic activity.
What that tells us is that any policy conclusions one might reach based on their data will be wrong because the data they’re using just aren’t as accurate as they might wish and that they are missing a lot.
Real-time, data-based policymaking just isn’t in the cards. For example:
Households making more than $75k as of when? At best, such income information is from 2019. The federal government has no ability to collect income data in anything close to real-time. And there are huge artifacts at the individual level when it comes to income. There are the cost-of-living comparisons I make all the time, but also income according to the IRS isn’t the same thing as actual household income.
Our close friends here, for example, rolled over their traditional IRA’s to Roths in 2019. Doing that spiked their IRS income for that year which made them ineligible for the two rounds of Covid payments because eligibility was based on 2019 tax returns. However, when they file for 2020 taxes this year, they can take those as a credit because their 2020 income will be back to “normal” make them eligible.
Finally, real-time, data-driven policymaking by the government is going to require that the government collect a lot more information from individuals and organizations which will raise a lot of privacy objections. Governments have a tendency to engage in mission creep and it’s inevitable that such data would be used for many other government functions besides economic policymaking.
There are many amusing things about the op-ed that I didn’t comment on. For example, their use of the term “high income”. Using their definition most Chicago public employees including the median school teacher, the median firefighter, and the median police officer are “high income”.
i also didn’t mention that the obvious conclusion from their data is that there are too many low-income workers. Taking a “don’t raise the bridge lower the river” viewpoint, the easiest strategy for accomplishing that would be fewer low-skill immigrants with inadequate fluency and literacy in English and fewer young people.
As I noted in the post I think their approach strongly suggests a rules-based approach to policy rather than the present approach. One of the many risks of rules-based approaches is that not everyone has the same relative priorities.