The largest association of professional economists, the American Economics Association, recently announced a new standard for publications: all data must be legally obtained.
The AEA is implementing a new Data Legality Policy, effective for submissions after July 1, stating all data used in papers published in the @AEAjournals should be legally acquired. Learn more at https://t.co/PtSNOisZHx.— AEA Information (@AEAInformation) January 19, 2023
This is a major change for professional economists, as the AEA houses some of the most influential academic journals in the field including the American Economic Review, the Journal of Economic Literature, and the Journal of Economic Perspectives, among others.
Defenders of the policy say the purpose is to discourage people from engaging in illegal hacking to obtain data, but it’s important to note the policy bans the use of any illegally obtained data— even if the data was not originally obtained by the authors.
This policy is, on a basic level, unethical. And it may sound counter-intuitive, but the use of illegally obtained data serves the public good.
Selective Data as a Public Bad
To understand what’s so bad about banning illegally obtained data we need to consider the incentives that face politicians. This policy applies to both private sector and public sector data, but since the public sector determines legality itself, I'll focus on the issues there. Let’s consider two different qualities that data can have.
The first quality is that data can be beneficial for politicians and bureaucrats or harmful to them. Think of jobs numbers, for instance. Data which shows massive increases in unemployment tend to be harmful for politicians and, to a lesser degree, their appointed bureaucrats.
On the other hand, low unemployment is good for politicians. You don’t have to look far in history to see politicians leverage employment data to benefit themselves or attack their opponents.
The second quality data can have is it can either be obvious to see in our daily life or it can be difficult to see. Inflation, for example, is pretty easy to see in your daily life. Even if the government never released CPI data, you’d be able to tell if your trips to the grocery store are becoming more expensive.
On the other hand, it would be very difficult to see, for example, how many of your calls and texts were being monitored by government agencies like the NSA. A normal person going about their everyday life would probably never uncover someone monitoring their communications.
There are a lot of other qualities we can think of data having, but these two margins are the margins that matter for us now. Armed with this information we can think of four types of data summarized on the following table.
Beneficial to Politicians
Harmful to Politicians
Easy to Observe Normally
Hard to Observe Normally
Table 1: Types of Data
So, both Type I and Type II data are easy to observe in our everyday lives. Politicians may or may not like our access to that data, but they can’t help it. Anyone who buys things knows when inflation is up.
Type III and Type IV data is, in some sense, more valuable for us to be given because it is expensive to find it ourselves. What will happen in these cases?
Let’s start with type III data: data that is hard to see, but beneficial to politicians if citizens see it. What will happen with type III data? Politicians will be willing to spend a lot of resources making sure data of this type gets out. For example, you and I may not be able to feel that “the world is safer for democracy” in our everyday lives, but if politicians can find data which indicate this is the case, they’ll be sure to let us know.
Type IV data is where we have a problem. It’s both difficult for someone to discover in their everyday life, and politicians have no incentive to release it because it would be harmful to their careers.
If we assume politicians are purely benevolent actors who work on behalf of the public good, we could assume they would release type IV data even if it hurt them. The problem is, economists never make this assumption about private firms.
For example, if firms have the ability to pollute without bearing the cost of pollution, economists assume they will do so and frequently use this as a justification to tax polluting firms. This is sometimes known as a Pigouvian Tax.
If you presented a paper at an AEA conference saying that the solution to over-pollution by firms was to trust firm owners to be benevolent, you’d be laughed out of the room.
So why would we assume politicians are benevolent actors? We shouldn’t. The logically consistent thing to do would be to make symmetric assumptions. If private firms seek their own narrow self-interest, it’s fair to assume politicians will as well.
So what would self-interested politicians do? They’d make the release of Type IV data illegal (by classifying it, for example) and they’d focus only on type III data. Whistleblowers like Edward Snowden would be punished rather than protected. In other words, politicians have an incentive to “pollute” the information space.
Notice the effect of the AEA policy now. The only way to get type IV data is illegally, but the AEA has made that off-limits. By making use of illegal data unacceptable in AEA journals, the association has essentially added another tax onto an already under-released type of data.
The amount of exposure and analysis this data gets will be lower, other variables held constant. So the benefit of releasing it is also lower. And when the benefit of doing something goes down, that means people will do that thing less.
So the AEA is now in the business of subsiding information polluters and combating anti-pollution agents.
But how much actual data will be suppressed? It’s tough to say, but, anecdotally, I was quickly able to think of examples in Drs. Christopher Coyne and Abigail Hall who research, among other things, US militarism. In their book Tyranny Comes Home, the authors comment on how Edward Snowden’s NSA leaks revealed several facts including that the NSA spied on 120 world leaders.
Would such a fact be allowed to be stated in an AEA publication? The problem gets even worse if we consider the fact that researchers frequently talk about data as being either numerical (quantitative) or non-numerical (qualitative). Would discussing the Snowden leaks qualitatively be allowed, or would researchers need to shut up about Snowden’s revelations entirely?
Being Bound to the Mast
The AEA recognizes the awkward position they’ve put themselves in. Academics pride themselves (whether correctly or not) in a commitment to the pursuit of truth above fear of punishment. In this light, the AEA policy seems cowardly.
To remedy this, the association has put a clause intended to protect the policy from criticism. The policy states that an exception can be made in cases “in which, by the handling (Co)Editor’s judgment, the societal benefit of the publication outweighs the use of illegally acquired data.”
The problems with this are obvious. First, what makes the editor of an academic journal qualified to run cost-benefit analyses for the entirety of human society? Second, and more importantly, with discretion comes an increased ability to apply external pressure to the person who has discretion.
In Homer’s classic epic, The Odyssey, the hero Odysseus (sometimes called Ulysses) sails by the mythical creature known as a “siren”. Sirens are essentially mermaids but not friendly like Ariel. In The Odyssey, siren songs enchant anyone who hears them to swim out into the ocean toward the siren. When the person gets close, the sirens kill them.
Odysseus wanted to hear the song of the siren, but he didn’t want to die. His solution? He tied himself to the mast of the ship, and instructed the crew not to untie him until they were out of range of the sirens.
One wisdom to draw from the story is that you can overcome temptations by tying yourself to the mast. If you don’t buy a quart of ice cream, you can’t cheat on your diet by eating a quart of ice cream.
This speaks to the problem with giving discretion to an editor. If the government puts tremendous pressure on an academic journal editor to shut down a publication that utilizes illegally obtained data, it’s not hard to imagine the individual caving to pressure regardless of the societal good. If the editor could respond by saying, “sorry AEA policy allows this— it’s out of my hands,”it would be easier to deflect pressure.
This logic is similar to how Apple has, in the past, refused to create a backdoor way of allowing law enforcement to unlock iPhones. Could Apple create the backdoor and allow it to be done based on discretion? Sure. But doing so would make every request harder to turn down.
The AEA’s decision to ban the use of illegally-obtained data is unethical because it only serves to interfere with an idea-space already polluted by political incentives and coercion. I hope economists will join in rejecting such a ban and embrace the position of Ludwig von Mises (a distinguished fellow of the American Economics Association), who noted:
Economics as such is a challenge to the conceit of those in power. An economist can never be a favorite of autocrats and demagogues. With them he is always the mischief-maker, and the more they are inwardly convinced that his objections are well founded, the more they hate him.