Hacker News Reader

Comment by huitzitziltzin | Hacker News Reader

Actual economics professor here.

This is a partial list. Any specialist in any subfield of the discipline could list a large set of open problems. Nearly all have broader significance.

In my own specialty (industrial organization) unsolved problems which would be of interest to readers here include optimal regulation of platforms and two-sided markets. Theory and empirics are difficult, so regulators are to some extent fumbling in the dark a bit. It is not at all clear how to regulate (for example) Amazon’s offering its own versions of products sold on its marketplace OR what the right tariff Apple should charge for access to the App Store is.

Extremely confident claims that such behavior is permissible or should be forbidden are likely unsupported by any careful analysis one way or the other.

View on HN

I posted the Wikipedia article hoping you and others like you would fill out what is an obviously incomplete list.

If you mean filling it out on wikipedia they will have little to no hope unless they are accepted by the wikipedia editorial politiburo.

This is weird. I have made edits with a new account on articles and they were accepted. Not very significant articles, maybe.

It's the same thing with stack overflow/exchange, in which good questions and answers not uncommonly get downvoted and or removed. I think there is an epidemic of petty academic tyrants. These are people who are objectively quite smart but maybe not as accomplished as they wished or hoped so they take out their frustrations on others in the only way they can have control in a world in which they are otherwise irrelevant and powerless. Universities , graduate departments churn out so many students relative to the supply of good paying jobs annually that not everyone will be able to achieve their idealization of career success. What I do know: the TED talk stuff will not work on these people.

I have tried adding edits, many that I thought would have been beneficial as being part of that particular industry / having intimate knowledge of a topic; only to be rejected by some moderator that could not bother to understand the topic.

It's infuriating. I've stopped trying to contribute.

yup this is a common problem with Wikipedia.

"It's infuriating. I've stopped trying to contribute."

You aren't the only one (albeit in my case, for more mundane/esoteric topics)

> only to be rejected by some moderator that could not bother to understand the topic

This is a tired trope, and it shows you're the one who can't be bothered to understand Wikipedia. Wikipedia is basically an anarchy (most of the time...), meaning that it's unlikely that the person who "rejected" your changes had elevated privileges, it was probably just another editor like you. Incompetent and/or malicious admins do exist on Wikipedia, but its much less likely you encountered one of those as opposed to just a regular user who disagreed with your change.

I'm neither a fan nor a Wikipedia apologist (WP has loads of issues), just trying to explain some basics.

BTW, I assumed you meant "Wikipedia admin" when you said "moderator"; real moderators (Wikimedia employees) do exist, but they're even less likely to encounter.

On the other hand, you say:

> having intimate knowledge of a topic

These kinds of edits sometimes get reverted on Wikipedia, and rightly so. I'm sure you understand that anyone can pretend to have "intimate knowledge of a topic", ideally edits should be backed with good references/sources instead of the editor's (unverifiable) knowledge.

The de facto reality of how wikipedia operates is that certain people own certain topics, usually with an agenda in mind. If you try to push back they call in their friends and shout you down. The actual merits of the edits are irrelevant. The anarchy stuff is just marketing nonsense.

Is that not anarchy though?

I basically agree with you, except for your last sentence:

> The anarchy stuff is just marketing nonsense.

With this sentence you're implying that Wikipedians want the public to believe that Wikipedia is anarchic, but this is incorrect, see: https://en.wikipedia.org/wiki/Wikipedia:What_Wikipedia_is_no...

The link is a Wikipedia policy, so pretty much as official as it gets. The sad truth is that Wikipedia policies, guidelines, etc are mostly irrelevant to the functioning of Wikipedia.

The real de facto reality of how Wikipedia operates is that it's a complicated system with most decisions being made being local to only a single or a few articles. Furthermore, nothing is set in stone at Wikipedia, be it content or policy, they all change with the whims of the editors currently active at the relevant place and time. Which is also why policy is mostly powerless.

All of my contribution are always referenced and well sourced.

My references themselves (from impact factor 10+ journals) were attacked and the cause of rejection. I'm not pretending that I know more than someone else; I'm leading the audience toward where they can assess the data for themselves and make their own conclusion.

I do this by referencing high quality, peer-reviewed, reproducible research. I don't post opinion (on Wikipedia). I state facts corroborated by multiple sources.

Out of curiosity, how does one even begin to address such a question? Because it seems to me that the answer turns entirely on your choice of a quality metric, which is entirely subjective. If you are a consumer, you would come up with one answer, and if you are a shareholder you would come up with a different one. On what principled basis could one possibly resolve this?

Not the OP. Designing a policy for minimization of dead weight loss in the economy (typically centered around consumer surplus), with preferential properties including time stability, enforceability, and fairness across implementees.

But "dead weight", "fairness", and the desirability of time stability (weighed against other considerations) are all judgement calls.

No, they really aren't.

- Lacking time stability means no one tomorrow has any incentive to stick to the policy, and means the policy is unlikely to succeed in its objectives.

- Deadweight is lost efficiency

- The only arguable one is "fairness", and to clarify I blunted the term to avoid the nuance of explaining Pareto optimality. But sure, we can also consider egalitarianism and other moral frameworks.

Many industries got rid of deadweight only to find during the pandemic they and their industry had lost resilience when circumstances changed (which they always do).

See: CPUs and GPUs in 2020-2021, baby formula in 2022, etc.

It can absolutely be a judgment call.

deadweight loss has a specific meaning in economics.

TIL

Deadweight loss refers to a net reduction in consumer+producer surplus within the economy.

It has nothing to do with the everyday meaning of the term “deadweight”.

TIL

> - Lacking time stability means no one tomorrow has any incentive to stick to the policy, and means the policy is unlikely to succeed in its objectives.

That's true, but time stability means that a bad policy is going to be maintained even in the face of evidence that it is bad. Deciding which is worse is, as I said, a judgement call.

> Deadweight is lost efficiency

The flip side of increased efficiency is decreased resilience. Without unused capacity (i.e. inefficiency) the slightest contingency will send the entire system into chaos. How much inefficiency you want to maintain as a hedge against contingencies is a judgement call. Different people have different risk postures.

> Pareto optimality

I think it's safe to assume that people on HN understand Pareto optimality (or are capable of looking it up). Do you really think that Pareto optimality is synonymous with "fairness"?

> we can also consider egalitarianism and other moral frameworks.

Of course we can consider these things. But that doesn't answer my question, which is how do we address them in a principled way that doesn't degenerate into a political dispute?

Deadweight loss is not a judgement call, it's an actual calculable amount, effectively the loss of productivity due to the tax. https://en.wikipedia.org/wiki/Deadweight_loss

Maybe its just the way things are described on Wikipedia, but isn't there an almost complete lack of citing any experiments or empirical evidence on that page?

I'm not sure what you mean. Deadweight loss is tautological. Maybe you're asking for evidence that taxes distort markets? If you accept that as true it's easy to see that deadweight loss is just the difference in surplus before and after the tax. Every econ paper on tax policy will try to measure the deadweight loss of the intervention.

I guess I just find it weird to have a page on Wikipedia about a quantity with no mention of what that quantity has ever been measured to be?

Wikipedia is a very difficult way to learn economics. In many ways it's like trying to learn mathematics or physics from an encyclopedia. The encyclopedic approach is - pedagogically speaking - sub-optimal.

But if I look at the page for "friction" it has tables of actual values and many references to practical experiments?

NB I was just comparing the writing style of that page (which seemed a bit odd to me) with other examples.

https://en.wikipedia.org/wiki/Friction

So in this context, how does one calculate what's "socially optimal?"

The problem I have with much of economic theory is that it tries to tie things back to what's beneficial or not to societies and tries to do so through a very thick utilitarian lens. There's often a heavy amount of fairly sophisticated math, statistics, and modeling parading as certainty when the fact is, there's a very high amount of uncertainty in many assumptions and in variety of economic theories that are just hand-waved away. It's not to say the problems economists face are simple, they're actually incredibly challenging and I think it's good economist continue to push for a better understanding of these systems.

My gripe arises when economists masquerade or imply current theory with certainty when in many cases, as if it were a mature science. It simply isn't and far too many use the appearance of certainty hidden in complexity of theory as certainty to push a biased personal agenda or perspective. When you overload terms like efficiency, efficiency in what context? Efficient for whom and through what perspective of the world? Very clear definitions need to be made so I don't question the motives of the attempt of quantification and can look at flaws and oversights that may exist in such metrics.

Most economic theory lacks real empirical evidence because it's just to difficult to play the experiments and interplay of micro, meso, and macroscale systems at a global scale without actually running the experiment (changing policies).

Eh, it's difficult to blame economists, or any other type of scientist here. They all deal with improbability and uncertainty.

It's the politicians and voting public that are the problem.

It's easy to get massive numbers of people to vote for you by giving out is completely wrong as long as you're extremely confident in your presentation of it. It's easy for these political winners to drive the economic systems to collapse while increasing authoritarianism by said false confidence. People want to vote uncertainty away, authoritarianism is one way of doing that.

The first sentence in the article is: "Deadweight loss ... is a measure of lost economic efficiency when the socially optimal quantity of a good or a service is not produced."

But there is no link for "socially optimal", and the phrase appears nowhere else in the article, so this definition is incomplete at best and vacuous at worst because its meaning turns entirely on the meaning of "socially optimal" but I can't think of any reasonable way to define it that is not a judgement call.

You are right, wikipedia is not a great source for economics. I just reached for it first because it's generally decent for most things. You can find a great definition in https://bankicollege.ac.in/wp-content/uploads/2020/12/Princi... but simply it's difference between the surplus before and after an intervention. Obviously minimizing deadweight loss isn't our end goal, it's just one measure we can use to compare interventions.

While not an economist, I've read enough economic papers that I actually agree with Wikipedia's definition. You can "plug-in" values to your dead weight loss calculation depending on judgement calls. For example, do you consider the fact that the carbon footprint of an object is ignored a subsidy? If so, your dead weight loss calculation changes. Suddenly you're trying to discount cash flow against the future cost of carbon removal and things are very complicated.

> when the socially optimal

Technically Pareto optimal. Pareto optimal points are not necessarily socially optimal.

https://en.wikipedia.org/wiki/Pareto_efficiency

If you go look at the theories, you will find that economics uses a lot of those overloaded terms, but they have perfectly reasonable objective meanings for them that are different from the popular subjective meaning.

OK, but it doesn't help to answer a question posed by a layman using overloaded terms of art without defining them, or at least providing references, or at least pointing out that the words you are using do not mean what they mean in common usage.

So where do I find these definitions? Because I am still highly skeptical that they can be defined in ways that do not conceal hidden judgement calls.

If we applied that same standard when talking about programming here, we would never get past talking eli5 semantics and definitions.

You can be skeptical, but the cure for that skepticism is something more akin to taking an undergrad course, or two.

The "fair" way to do it is allocation of costs per value of sales AND zero access to data by market owner.

So everyone pays the same charge for IT costs AND market does not have an unfair advantage.

Theory and data.

Both are hard, especially in the cases I listed.

With a good theory of (e.g.) platforms, you could likely identify a few different fundamental forces which may push in different directions. Hypothetical example: force X would benefit the platform at the expense of sellers, force Y would benefit the sellers at the expense of the platform. A priori it may not be possible in theory to put a magnitude on those effects: there may be nothing that would say "we can show that X > Y in every case."

This is where you need data. (This part is especially hard for platform markets!) You would need data on the platform side AND on the consumer side AND on the supplier side. You write down a model of (generally) consumer demand, supplier response and platform behavior (at least - probably taking a somewhat simplified approach to at least one of those sets of agents). With your estimates, you would try to answer the "Is X > Y?" question posed above.

You can also simulate the impacts of counterfactual policies, among other things.

Thanks!

> This is where you need data.

More specifically, this is where you need data which does not and cannot exist. Specifically, an objective scale which would allow you to compare the cost or benefit of policy to the platform and the sellers. There is no such objective scale, because value is subjective and not comparable between different economic actors, or even for the same actor at different times. Any model which assigns numeric "estimates" for these costs and benefits so they can be compared is substituting the authors' own subjective valuations in place of those of those who will be affected by the policy.

> More specifically, this is where you need data which does not and cannot exist.

This is wrong. All important work in my field in the past 25 years has been empirical.

> Any model which assigns numeric "estimates" for these costs and benefits so they can be compared is substituting the authors' own subjective valuations in place of those of those who will be affected by the policy.

Very, very wrong. You’ll never get away with picking your own parameter values in modern empirical IO.

If you don’t know the field that’s fine. I probably don’t know your field either. But don’t make these overconfident and totally wrong claims about what’s done in a field you don’t know anything about.

Now I'm inordinately curious how many other economics PhDs and professors visit here, between your comment and Amazon's hiring under Pat Bajari.

One additional data point: I'm approximately an associate professor of economics in North American terms. Also curious how many of us there are here. As it happens, one of my colleagues was recently hired by Amazon (quite nearly appending an extra zero to their salary in the process).

Amazon is the second largest employer of economists, behind only the Fed: https://www.cnn.com/2019/03/13/tech/amazon-economists/index....

I've talked to several big names that have told me about megaoffers from Amazon. (They probably don't care if people know, since they told me, but I'm not going to name them.)

What does an Amazon economist do?

They hire in three buckets:

- Time series forecasting.

- Reduced-form causal inference

- Structural, especially what is called "structural IO".

Among the questions people in the last category will answer are: how should we price this product? How should we decide which new markets to enter?

Quite a few, but the benefit of participating in discussions about economics on HN is less than the opportunity cost. As economics is heavily computational, HN is as natural a fit for an economics PhD as a CS PhD.

Checking in. I did my PhD in polisci, but invaded.

Econ PhD here! Also IO.

Please chime in if you think anything that I said is wrong.

Finished my masters in economics. Went on to do my PhD in economics but dropped by the end of the first semester. Realized it was going to be too much with a young family at the time.

PhD economics/econometrics, ex-research economist, turned machine learning engineer. In my personal experience the mathiness in both fields are comparable, but pay-to-effort ratio was much better in machine learning.

> In my own specialty (industrial organization) unsolved problems which would be of interest to readers here include optimal regulation of platforms and two-sided markets. Theory and empirics are difficult, so regulators are to some extent fumbling in the dark a bit. It is not at all clear how to regulate (for example) Amazon’s offering its own versions of products sold on its marketplace OR what the right tariff Apple should charge for access to the App Store is.

The subject was "unsolved problems in economics", not "unsolved problems in politics". Regulation is a political topic, not an economic one.

That last point might be economic in nature, but only if Apple is the one making the decision, and is free to choose their own price without fear of regulation. Otherwise it's still politics.

> the subject was "unsolved problems in economics", not "unsolved problems in politics". Regulation is a political topic, not an economic one.

If you don’t care about getting it right, fine. Ask politicians to do whatever they want on the basis of whatever biases they have.

If you want to get it right, then you need to weigh the costs and benefits of different regulatory policies. That requires theory and data.

I suspect that "Solved Problems in Economics" would be a shorter list.

I have a couple questions, from the point of view as an outsider, I see there being three big problems with economics at least the way it is popularly understood:

1) many of the assumptions in economics are culturally specific

2) many of the models in economics, while self consistent and plausible, have not been adequately tested with real world data

3) economic models don’t seem to take constraints into account leading to absurd results like infinite growth

Within the field, are any of those three seen as important problems? Thank you

>1) many of the assumptions in economics are culturally specific

Which ones and maybe I can comment. I don't think this is true.

> 2) many of the models in economics, while self consistent and plausible, have not been adequately tested with real world data

This is everything that we do. The entire field is all about data and has been for probably 30 years (more actually, but we can point to an improvement in our own standards in the mid-1980's).

> 3) economic models don’t seem to take constraints into account leading to absurd results like infinite growth

A very basic definition of economics in undergrad classes is "the theory of optimizing behavior subject to constraints" so constraints are very much our bread and butter.

Separately I don't think most (or any) macro economists (or growth guys) would say "yeah my model gives great predictions for what the situation will be 1,000 years in the future". Even 50 years in the future is pushing it.

>> at least the way it is popularly understood

This may be your problem. The understanding of the field you find from commenters on this website is a million miles from what it is really about.

> three big problems with economics at least the way it is popularly understood

The popular understanding of economics has very little to do with economics.

You say “optimal” and “right” nonchalantly. Under what ethical framework are you trying to solve these “problems”?

> Theory and empirics are difficult, so regulators are to some extent fumbling in the dark a bit.

So, just like economists!

> It is not at all clear how to regulate (for example) Amazon’s offering its own versions of products sold on its marketplace

I’m glad people like you are at least thinking about it. I have no insights (different degree), but I was wondering about that exact thing a few months ago.

What would be a good place for a beginner to start working on a problem related to IO economics of platforms?

I've studied MITx Microeconomics course. It appears undergrad level Micro course, and a background in machine learning and software engineering.

It's a hard field b/c there is not that much introductory / accessible material.

There is a recent book by Belleflamme and Peitz, which should some day be followed up by a second volume about regulation.

There is also a recent survey of the empirics by Marc Rysman and two coauthors.

A lot of these seem like systems thinking problems, rather than pure economics. Tackling it across multiple perspectives might not even result in a satisfactory answer for any party, but it can improve the situation.

>This is a list of some of the major unsolved problems, puzzles, or questions in economics.

The opening line.

Reply on news.ycombinator.com