huitzitziltzin

Also: from your comment I’m pretty sure you don’t have the background (correct me if that’s wrong) and so don’t know what it really means in practice to make an equilibrium assumption in a macro model: Markets clear, on average people have reasonable beliefs about the evolution of aggregate variables, firms maximize expected profits. That’s all pretty harmless.

I definitely don’t agree that we are “using mathematical tools which aren’t suited for that.” We aren’t treating the economy “like a steam engine.” The entire revolution in macro from the 1970’s on involves optimizing agents. There is no useful analogy to a steam engine.

I suspect you have been reading criticism by people who are misinformed about what macro actually is and how it is practiced.

Ljungqvist and Sargent is a standard grad level text.

Acemoglu has one with a growth focus.

Miao is another one.

Stachurski and Sargent is one focused on computational issues.

Stokey Lucas and Prescott is the math but I would skip that.

Dejong and Dave covers macroeconometrics.

Generally agent based models have numerous parameters which can take many values (endowments, preferences) and the models don’t themselves give any guidance about how to set the parameters. Theory can give limited guidance (eg., that function is concave, this parameter is negative). Sometimes we have experimental data though its generalizability beyond the lab is uncertain.

What you want to do to create a scientific macroeconomics is to work backwards from the data you see in the economy (aggregate consumption, investment, etc.) and how you know the aggregates were generated (via the behavior of a lot of individual agents), and an equilibrium assumption to recover the parameters.

If you know the parameters of the model you assume, you can then simulate interesting counterfactuals. (And yes you assume the model - a “full” model including “all” of the individual endowments and parameters you can think of is completely intractable. You have to simplify.)

You’ll never get that out of the author’s computer game.

If you want references to the macro literature it’s enormous and I can provide them.

Agent based models have been around since the 1980’s at least. No one uses them in central banks, no one uses them in industry, and you can be very confident that they’ve tried.

Then there are people who clearly don’t understand (but should know better) like Navarro.

The beliefs of the man at the top, which are not particularly fixed and vary from hour to hour according to what has most recently been tweeted at him, are the only beliefs that matter.

He is not known to be a deep thinker about matters of international economics so I wouldn’t expect much!

There is no suggestion in the article that she is being detained for her opposition to the Ukraine war.

She probably doesn’t have a great case. The culpable party here is 1000% her adviser who asked her to bring the samples back given “the complications [her adviser] Peshkin has experienced when French colleagues try to mail him research samples.” (Ok so you aren’t supposed to bring such material here…)

It’s unconscionable and incredibly stupid to put a foreign student at risk by asking for such a “favor”, which she surely didn’t feel in a position to refuse. He shouldn’t even have asked an American (who can’t be deported) to do such a thing!

Also you seem to arrive at your first ML application after 1500 pages or so.

I’m reminded of a class at my university on “early Christian literature” where the professor announced on the first day that despite the title they would actually start in 3,000 BC.

(Slightly longer answer: I don’t have the funds to match what someone is likely to pay for this data, but if I bought it and gave it to you for free, and also gave you all the inpatient hospital discharge abstracts in the US in a matchable form, plus whatever health data you want, plus family relationships, plus a budget for 100 PhDs, product managers, and marketing people for 5 years you couldn’t turn it into a viable product. (Indeed that is literally what 23andme tried to do for years!))

And no I don’t think all of that DNA data would be valuable to the likes of a large health insurer like Humana or Aetna either.

The medical records you are imagining an insurer can link to genetic data are worth even less than these DNA sequences turned out to be worth.

Sincerely,

A former health economist who has worked both with tens of millions of inpatient discharge records, and (separately) a detailed survey which is complemented by genetic data.

These claims are quite hard to square with the long waits for H1B visas, extremely high salaries in the technology sector and net immigration to the US from China.

I’m not aware of any Americans or Europeans in my network who have gone the other direction to China.

Perhaps you have different data about the demand for tech worker visas in China.

https://en.m.wikipedia.org/wiki/The_Market_for_Lemons

The case you are confused by is a version of the model there. The relevant material is covered in any intermediate-to-advanced microeconomics textbook.

I am impressed by the progress on LLMs but I remain skeptical that they can replace humans.

Perhaps some (distant!) future model but I don’t fear mass unemployment (for example) or even moderate LLM-driven unemployment in the near-to-medium term.

They can clearly complement human labor but there are vanishingly few domains where they can be substitutes.

https://fred.stlouisfed.org/series/SFXRSA

Nothing like that claim appears to be true.

Especially if you ever have lived or will live in Texas the first volume of the LBJ series is essential.

“ A Study of Cartel Stability: The Joint Executive Committee, 1880-1886” By Rob Porter (northwestern)

That paper is a classic and I hope porter gets the Nobel prize for the work he has done.

In that paper, we actually know for sure that there was collusion. Porter is trying to assess how hard it was to sustain collusion using the data. (One thing you are not crediting in your website is that collusion is hard to sustain bc all of the colluders have the incentive to cheat!) It takes a lot of assumptions and serious econometric chops to make the analysis credible.

That paper is 40 years old but what you’re really trying to do with your project is called “structural industrial organization.” It’s been around in the economics literature for a long time. It’s my field. It’s hard.

If you want a recent overview which still won’t answer the question the way you wanted to, get “the handbook of industrial organization” edited by hortacsu et al, or the book draft “empirical methods in industrial organization” by aguirregabiria (who is at U Toronto!) or “quantitative techniques for competition and antitrust” by Davis and Garcés.

Good luck.

You can publish some correlations between prices. (Remember there is no known method where you can look at a time series of prices across stores and say “that’s collusion!”). And hey if you have 10,000 - 100,000 products per store you probably have correlations as high as you want. None of that will actually show collusion or more importantly suggest a policy response.

Even very high correlation _within a category_ across stores isn’t going to do that. Say you find price movements in tomato paste across all N chains are 0.95… ok now what? There are 10,000 or 2,000 product categories so you found a high correlation in one of them what do you want to do about it?

This paper:

https://www.aeaweb.org/articles?id=10.1257/aer.20170116

Is the best case scenario (=easiest) for the exercise you want to do and will illustrate what is hard about the grocery case. In the paper there is:

- One product

- one clearly isolated market.

- four or five firms.

- A clear mechanism for collusion (a government price sharing website).

Even then there is no statistical or econometric test out of the box which “shows collusion”. It’s a great paper (both in my opinion and that’s why it got into AER) but there is nothing they do that says “hey here’s collusion based on this t-stat.”

The grocery case (overlapping markets, thousands of products, many firms) is vastly, vastly harder.

Also imagine the false discovery rate of <whatever analysis you think you can do> with the thousands of products which exist in a grocery store!

“Ah look we found collusion in salty snacks between weeks 38 and 45 of 2022 in stores 1, 5 and 13!”

Even if you knew what analysis to do (and you don’t) you are going to end up crying wolf a million times and solving no problem at all.

Again - a worthy effort but the methods just don’t exist.

The issue here has nothing to do with the size of the data. A grocery store will have 100,000 products at most, probably sets prices about weekly and maybe you have ten years of data. The market is an oligopoly so you have 10-15 store chains. That fits in memory.

The issue is not at all the mechanics of running the analysis but what analysis you run. Or don’t run bc as I said there is no easy way to just “identify collusion” in data like this or every agency would already be doing it

You can do economic analysis of pricing data, and especially the interaction/correlation of multiple streams of prices across time ‘’’

The item in that quote is significantly, significantly harder than the author is giving it credit for. Canada has a competition policy agency. They are (almost surely) entitled to demand data from firms as part of an investigation. Their data will be better than the data here.

You can rarely if ever prove these cases on the basis of data analysis alone. (Indeed if you could the world’s antitrust agencies would just monitor such data! And anyone engaging in collusion would attack the monitoring adversarially - not that hard to predict!)

In grocery data you will be looking at thousands and thousands of prices of different goods from different suppliers with different costs who are exposed to different shocks due to variation in the costs of their inputs or god knows what else.

Not to be negative bc the idea is nice, but this is a complete waste of time.

~ an actual antitrust guy.

Here’s a chart prepared by a real estate research firm, linked by a housing journalist:

https://x.com/kristoncapps/status/1826701823522341065

If you live in one of the very very few markets where this behavior is slightly more frequent and hate these firms… why not mess up the value of their investment and harm them financially by permitting more housing to be constructed?

Sure it could die of a weird disease bc we didn’t have the full genetic code. And yeah I’m equally unconvinced from the article that we actually can do this, but it’s unclear to me what the risks are if we can and do?

Scenario: Bring back one to one hundred wooly mammoths and… X worst case scenario happens. What’s X in your view for the wooly mammoth? I can’t think of any that seem actually bad but maybe my creativity is lacking?

“Quantitative analysis of social data has an alluring exactness to it. It allows us to estimate the average number of minutes of YouTube videos watched to the millisecond, and in doing so it gives us the aura of true scientists. But the advantage of quantitative analyses lies not in the ability to derive precise three-decimal point estimates; rather, quantitative methods shine because they allow us to communicate methodological goals, assumptions, and results in a (hopefully) common, compact, and precise mathematical language. It is this language that helps clarify exactly what researchers are doing with their data and why.”

A lot of industry “data science” is moronic, which is to say not clear about assumptions or about what is and isn’t possible with the estimates. Leading case: Models are treated as causal when they are not. Antidotes to that kind of thinking are great. Hopefully this can be one of them. I’ll keep looking at it.