Hacker News Reader

Conversely, I think you're overestimating the technical capailities of Canadian federal or provincial regulatory organizations.

Government departments notoriously have serious technical debt, so while they might entitled to demand this kind of data, I sincerely doubt they are doing it with anything resembling modern data analysis. These orgs aren't going to have working data lakes, spark clusters or data warehouses on cloud infrastructure. They maybe have legacy SQL databases being pulled into spreadsheets.

So, while their data access might be better theoretically, I doubt they truly are able to even remotely analyze the data effectively. Data science in govt is notoriously poor outside of Statscan or other tech heavy departments.

never underestimate a nerd with moral high ground

Canadian economics departments are full of really smart economists. They are happy to help.

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

> Canadian economics departments are full of really smart economists. They are happy to help.

The appropriate incentive is lacking.

it should be easy to make good incentives - the analysts that produce the evidence can be awarded a (large?) portion of the fines levied to those companies that are colluding.

Just like how the SEC whistleblower laws will awarded part of the fines to those who they reported.

Famously , academics analysis found (virtually proved) share options manipulation by statistical analysis alone, which was a bombshell in the tech industry a decade ago, threatening many executives with imprisonment. However , it's not obvious how you could employ the same process here. with share options there was a liquid market as a source of external information.

1. If anything, this public dataset will increase 'collusion' by giving the parties in the market access to the integral set of market prices.

2. Collusion is a strong word. It hints at breaking the law, willfull action, etc. Looking at the price of a competitor and acting upon that, and internal information, is not collusion. It is completely legal and economically rational behaviour. There are hints and information everywhere on which to act. Groceries are an oligopoly, and they are ogipolistic buyers as well. Lots of information for grocers in the price bids of their suppliers, that all groceries get in tandem.

3. I've worked as an economist at one of those government agencies. They are filled with keen PhD / MSc economists looking to bust kartels and find collusion. Geographical price dispersion models are so '00s. There is a lot of positive energy and interaction between academia and government. Actually _proving_ collusion is terribly hard. Prices that move in parallel is not collusion. Proving collusion is finding documents, emails, phone calls, tracking cars with staff members of various firms to motels , paper trails of inside information in another firm. That's collusion. And I fear that's not what happens in the grocery sector.

The motel one comes to mind as real life example of how DoJ in the US prosecuted a proper cartel. The talk was by more a broad-shouldered officer type, than an economist. Certain set of transport firms always had a interesting way of responding to auctions. Always the second best offer was quite far behind and there was a pattern to which firms did or did not reply to a bid request. End of the investigation, they could trace employees of the firms to meeting in a motel before bidding. And somebody left a paper trail.

> I've worked as an economist at one of those government agencies. They are filled with keen PhD / MSc economists looking to bust kartels and find collusion.

There is the skillset of the employees, and then there is what management allows employees to work on. A decade or so ago, a Statistics Canada senior manager was interviewed and claimed as fact that they were unable to reproduce an Excel model that analyzed the % of real estate sales to recent immigrants to Canada, one that an individual CRA team (one person I suspect, it's not complicated) in Richmond BC had done a few years before.

At the same time, they were given additional budget of many millions of dollars to accomplish this task and other things. They did not accomplish the task.

> Proving collusion...

The government has tipped their hand already as far as I'm concerned. I don't expect anything would ever be proven, I just want to see as much evidence as possible to help me decide what I am going to believe - I'm certainly not going to take epistemic advice from anyone in the Canadian government.

Totally get your point between individual skill / trust and the agency level. In my little country we had a recent event where a governement agency, pretty much failing for the last 20 years, but especially pronounced the last 5 years suddenly came clear in a PR-move. What? Turns out a newspaper had succeeded in a neatly tuned FOIA-request (local equivalent) and forced their hand. At the same moment a vote was underway for extra budget, looking to fail. So the sudden transparency was really pretty ugly.

Surrounding collusion. I can’t tell you what to believe, but keep in mind that there are a lot of reasons why correlated moves in prices can occur without communication. Lots literature in industrial organization (the field of why markets have particular ordening and how actors behave on markets) on signaling. A though experiment might be the games of poker versus bridge on explicit vs implicit signaling / coordination.

If we were actually serious here, we could introduce a variety of ideas into this conversation that would provide very interesting insight into what is going on in this thread, and in the realms surrounding these allegations...just one of them is this:

https://en.m.wikipedia.org/wiki/Closed-world_assumption

It would be fun if this fellow's initiative actually got some traction, and attention, and then out of the blue they launched a new educational section of their website to assist the public (and journalists...or not) in their thinking on such matters.

I suspect these giant retailers have a risk department, and at least some people in that department would be able to recognize that which the general public and participants in this thread lack the ability to recognize.

Can you put a link to the story about the Stats Canada manager? There was a story a few years ago about a Stats Canada employee who could not reproduce a model because of confidentiality reasons within CRA, but certainly you wouldn’t be getting confidentiality and privacy mixed up with competence. Can you put up a link?

https://www.theglobeandmail.com/canada/british-columbia/arti...

That article covers the initial report, not the subsequent claim that they do not have the ability to reproduce it.

> But the report was not made public until a five-year access-to-information battle concluded recently.

This demonstrates that Canadian democracy does not rule according to the will of the people - there was never a referendum on whether the government should hide the truth from citizens.

> Of the 46 houses bought in Burnaby, staff found 72 per cent were purchased by new arrivals to Vancouver who reported an average total family income of just $16,000.

Immigrate to Canada, enjoy all of its world class (at least formerly) social services, but pay no income taxes. Also not the will of the people.

Hanlon's Razor seems to be the guiding principle of journalists, and the general public in general. Gosh, I wonder if our educational curriculum (or lack of specific content) has anything to do with this state of affairs. Note also who it is that determines curriculum.

https://en.m.wikipedia.org/wiki/Hanlon%27s_razor

> The market is an oligopoly so you have 10-15 store chains.

Not that prices aren't consistent across a chain, individual stores in one chain will have different prices, calibrated to local supply and demand. In many cases they'll also have local products that aren't in most of the stores in that chain. I think you have to track prices in each and every store, not just per chain.

Still, I think you're right about this fitting in memory. This isn't "big data".

Hey man, I'm just creating the hammer - you can throw it in the lake, make a treehouse for your kid, smash a window, up to you :-D

Assuming you’re the author (and again this is a nice idea!) there is a paper you should check out.

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.

Actually I’m going to add to your reading list:

“ 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.

huitzitziltzin, I really appreciate that you've replied with so much detail, named the field of study, and gave references to research for me to follow up. I am going to be reading up on these topics.

In retail (or any market with elastic pricing, not sure if I'm using that term correctly), collusion is really hard to gain hard evidence (evidence that will hold up in court) using data alone. You need to use the data as a tool for on-the-ground and in-depth qualitative analysis. I would not be surprised if this holds "even more true" for consumer goods like this than it does for other, highly regulated fields like finance and healthcare. I personally like the idea of focusing on an individual good or category of item, and seeing if you can find similar patterns, in other goods.

isnt the difference that when you have an open dataset, you get many many more eyeballs, and some very passionate ones. and they can make it their hobby, unconstrained by day to day needs of a job. eh?

No. They have to know what they are doing.

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.

ok. i believe you, but hope that your maybe a bit overly pessimistic

You get the top 100 most plausible hunches and then investigate further from there… why focus on proving something in the data itself?

Investigators work on hunches just fine if you pay them enough.

> 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.

Mocking the epistemic skills of others while engaging in literal soothsaying (stating facts about future events) shows how culturally conditioned North Americans have become regarding government wrong doing.

I don't expect these people to be successful in their primary goal, but an enticing second order effect is it could accidentally catalyze public interest in epistemology, logic, rhetoric, the nature of Human culture and cognition, and a variety of the other things that keep commoners locked in a sophisticated virtual reality without their awareness (in any sophisticated manner at least). Now that would make this whole game a lot more interesting, because it applies to everything.

Is it necessary to prove it? I think it's sufficient to come up with an unbiased metric for which players are the most likely to be colluding at any given time. That would be a starting point for encouraging them to behave such that they're obviously not colluding: compete louder please, or you'll be at the top of this list.

Well… what do you want to happen ?

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?

I think there are a lot of people in desperate situations who feel like the deck is stacked against them. They end up lashing out in completely irrational, often violent ways. You see them on the news all the time.

What I want to happen is for those people to realize that they're not alone, and that the whole world is not working to put them at a disadvantage. Really, it's a rather small group of people who are setting the rest of us up to be rats in a maze with no exit. Then I want us to find those people, and go after their bottom line. Nothing violent, just like... welding the doors shut on opening day, that sort of thing.

It might be less rational than something more restrained like getting a law degree and working in the system to bring them to actual justice. But it has... a glimmer of rationality. You're no longer an insane lone wolf lashing out against an impossible situation... now you're a shitty activist, and even if you're not acting with a cohort, you know that cohort is out there. You're not stuck, you're not alone, there's a path forward--you can aspire towards being a less shitty activist.

But it requires a justifiable target, a representative of the untenable situation. A list of potential enemies that's far less than everybody. Once that list exists, I want companies to compete to not appear on that list.

Markets work without being rational, as does evolution (some predation required). This would be the same.

Yes, I’m sure that welding the doors shut on opening day will really stick it to the man. It certainly won’t horrify the minimum wage employees who go in there to work.

What an unbelievably stupid thing to advocate for, much less under the guise of non violence.

Of course there are a million better ways. You could gather some money an pay the employees to strike, for instance. It should depend very heavily on your target. My point is just that if companies are organizing against you, you should organize against the companies. Symmetry. Deterrence. Doing it well... well that's a life's work. The goal is to get the ball rolling and incrementally improve, not expect people to become Ghandi overnight. And at the outset, the credibility of the threat you can present is more important than the direct applicability of your action. We're not even at the point where "how strong of an action is warranted?" is a conversation. So step one is starting that conversation.

The problem is that you can only analyze pricing at the retail level. At the retail level, prices can legally be set in any number of ways. For example, it is totally legal to walk into a competitor’s store and set your prices accordingly.

That’s all this metric will show.

It also won’t show supply chain issues that make the Canadian grocery industry look the way it does. As an example, buy 100 head of cattle in western Canada and get the beef into a store. Are you having any luck finding a place to process it?

They're still trying to resolve the bread price fixing from years before the pandemic.

The agency is likely underfunded, understaffed, and disincentivized from actually fixing the issues which are so pervasive across the country that it's just business as usual for most people..

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

Would this be the sort of analysis that measured transfer entropy or granger causality?

Nice try, Loblaws.

Yeah we have a competition bureau that’s known for never doing its actual job.

The fact that they keep approving mergers and acquisitions in telecom, banking, and groceries despite those being some of the most comically oligopoly-esque industries within a country on earth shows just how tragically wrong you are.

If there was 2 telecom companies left standing in Canada, our competition bureau would pretend to study their desire to merge for a year before approving it, under the condition that their collection of subsidiary brands pretend to be different companies while advertising.

> ~ an actual antitrust guy

So, unfirable? Essentially useless and beholden to little if any performance metrics?