Using Foursquare to Detect Tax Evasion / Money Laundering | Hacker News Reader

Using Foursquare to Detect Tax Evasion / Money Laundering

diegobasch.com

27 pointsdbuthay51 comments

Open articleView on HN

The example to date is the Eiffel Tower, with 36429 checkins over the 3 years of Foursquare's existence.

Approximately 6.8 million people visit the Eiffel Tower each year.

It doesn't take a math major to realize that this is a laughable idea.

Probably the majority of those check-ins happened in the past year. Also, look at a Starbucks near where I live:

https://developer.foursquare.com/docs/explore#req=venues/49f...

Over 10k check-ins, and I doubt it has 2 million visitors per year.

Please elaborate on why you believe this is a laughable idea just from a numerical standpoint.

Have you ever checked into a business because you were there even if you did not buy anything? Have you ever gone to a business and not checked in even though you were buying something? I have done both. The number of people that check-in at a business has absolutely nothing to do with the amount of sales that business is on the hook to pay taxes for. One person might check in to Starbucks every morning as they purchase $50 worth of coffee and pasteries for their office. Another person might check in every morning when they buy their $2 double shot. Another person may go every morning yet check in only from time to time. There is no way this could be calculated to the accuracy that (dear god I hope) the IRS is held to.

OMFG Starbucks reported $3B in revenue this past quarter, but this store only has 10k checkins! Send in the G-Men, they must be laundering money!

Come on, seriously.

That's nonsense. There are 20k Starbucks in the world, this is one.

http://en.wikipedia.org/wiki/Starbucks

Congratulations on completely missing my point.

You say that this could be yet another datapoint for the IRS to collect taxes with.

The example you use has the number of Foursquare (and I'm sure Facebook) checkins within the statistical amount of noise for the actual number of people who visit the venue.

The example you give in this thread is a corporation who undoubtably has a team of dozens if not hundreds of tax attorneys on staff to ensure that the company pays exactly the amount of tax they owe, at the federal state and local levels, and not one cent more or less.

But let's take it a step further, and say that Joe's Ajax Factory in East Blogentry, California is skimming off the register. Okay-- so now you need some city clerk in East Blogentry to run this magical data analysis software which says Joe isn't putting enough into the city coffers?

Again-- this is a laughable idea. Even if the software were completely accurate (which is impossible) and admissible for a local tax audit (which it wouldn't be), and the amount of local taxes that you recouped were more than the TCO of this software (which it wouldn't be), you're assuming that the unscrupulous store owner (undoubtably laundering money a la Office Space) would just sit back and take it?

I highly doubt the government would trust a service like that for such problems or even waste their valuable time. I'm sure they have much more efficient ways of detecting this then trying to sift through that junk. It's also very easy to manipulate check-in's (Foursquare specifically), in respect to checking into locations when you don't actually have to be there, physically. And as far as Facebook goes, you can check others into a location, which could easily spike false information.

I've heard, no first-person knowledge here, that upon suspecting a business, taxation agencies are not unknown to send out a person with a clicker to tally patronage during spcific time periods and see if that lines up with their expectations and what the business reports. As others have said, this could be one of many signals to use in determining who to follow up on.

I've heard of tax inspectors counting the cars parked outside small businesses to get an idea of number of customers.

I guess the more general point I was thinking about is that there are tons of new signals that governments or private groups could use to assess all sorts of activities. For example the government has easy access to cellphone density data from carriers, which would be harder to game. Foursquare check-ins could be a cheap way to decide which places to look into first.

The effect is probably real, but the problem is the correlation coefficient probably varies over space and time.

E.g. you probably have a lot more Foursquare users in Silicon Valley than in Smalltown, USA.

Also, Foursquare probably has its popularity change over time. Right now it's still a little new and shiny, but its popularity might change over time.

If it was the only information available, it would be better than nothing. But your proposed method would be extremely noisy and unreliable.

The good thing is that it's not the only information available, just one more signal. And what's wrong with trying it out only in the Bay Area and New York, for example?

"Hey everyone! The IRS is using your check-in data to police the places that you frequent!"

I'm sure nothing bad at all would happen upon people reading that headline.

Well, perhaps it's already happening. After all, you're voluntarily publishing your location in real-time to the world when you check into Foursquare.

I find it interesting that we do this naturally and without thinking twice about it. Ten years ago the idea might have seemed preposterous.

tldr: just because lots of people use foursquare there should be a direct correlation between check-in's and revenue; this method should be used by the IRS when collecting taxes.

Are you serious? This is no way front page HN worthy. This is the worst thing I've read since that "wake up at 5am and eat a slice of ham to start my metabolism" trash.

A good exercise, but nothing beyond that. Too many factors based on assumption. Also repeat visitors tend to skip continued the check-ins after the novelty of it wears down (unless there are extremely good reward offers associated with number of check-ins).

As long as the "repeat visitors" pattern is the same across business categories, it wouldn't add noise. There are always first-time visitors.

The difficulty is in the details. Normalizing would have to account for a variety of different variables such as location, average age/customer base, average revenue per customer, etc. Furthermore, you would need all of that data many different businesses to even be able to build your normalization.

Sure, you could do a subpar job with the normalization but then that will increase your false positives and decrease confidence in the method.

Perhaps a slightly more realistic usage would be for financial analysts to use the data as a (weak) signal as to whether or not companies are on track to meet their earnings expectations. I believe that they have gone as far as to buy satellite pictures of parking lots (empty/full) to try to gauge how well big retailers are doing...

If you've ever seen the code behind what the IRS uses today, you'd realize that it couldn't possible integrate with anything, let alone it's own systems.

Plus this doesn't work for so many reasons, many of which people have outlined here.

Related question. Am I wrong in assuming virtually every business that has a "cash only" policy does so to facilitate tax fraud?

I know there are legit exceptions, but isn't this the main reason they do it?

You would be wrong in that assumption. Some businesses don't like paying the additional fees charged by banks when you use a credit card, they can be up to 8.5% according to a friend of mine that runs a locksmith. He would rather give 8.5% more service to the folks who come in the door. He has an ATM on site that will let you get a cash advance on your credit card if all you brought was plastic. It has a $3 transaction fee (plus what ever your bank charges). When people complain he says, "you can pay it or you can pay it." Then explains that he (like everyone else) would simply raise his price making that fee 'invisible' to you and let the banks get away with it. This way you have a choice (of sorts).

Your friend's rates seem a bit crazy. Square charges under 3%, and a dedicated merchant account can be less still.

Fees are nowhere 8.5% I've never heard that and have had a merchant account since the 80's at several businesses.

Not taking credit cards will cost you business in a retail location.

He's showed me bank statements, things like 'points' and 'cashback' and other 'features' of the card all seem to come back to the merchant as deductions against the amount billed vs the amount payed. He summed up the total billed (what he charged at the register) and the total deposited, and computed the difference, that difference was 8.5% rounded to the first decimal place.

Sure if he's doing small transactions but in general you shouldn't be allowing people to use a credit card to make a $2 purchase, right?

What is defined (in this thread) as "most cash businesses" would be to me businesses that would typically allow someone to pay with a credit card.

That wouldn't be a local bodega allowing some to pay for a pack of gum with a credit card. It would be a business allowing cc use in a typical case where the per transaction charge is not significant and you don't stand to loose business because you don't accept credit cards.

Examples:

You run a restaurant and your average ticket is $45. If you don't accept cc's you will generally loose business.

You run an ice cream stand and your average ticket is $5. Less expectation of being able to (or needing) to use a cc for a small transaction like that. Chance of loosing business: significantly less.

Entirely possible. for a long time in California it was illegal (state law) to require a minimum purchase amount in order to use credit cards I believe (but can certainly be mis-remembering) that this was only recently changed. And yes, his business does include a lot of 'small' transactions like someone coming in to get a key made etc.

On a related note, Fumiko Hayashi [1] has done a lot of research on the fee structures (much under the question of how one would regulate them and what constitutes an out of balance fee structure) but her papers are pretty accessible.

[1] http://www.kansascityfed.org/speechbio/hayashi.cfm

The actual fee charged could be close to 8.5% if you are charging mostly small transactions b/c they usually charge a percent plus a flat $0.30 fee.

Say you charge the customer $5: 2% + $0.30 is $0.40--8% of the total.

It depends on your transaction amount, the fees can easily exceed 8.5%. When I had no change the other day and paid 45 cents at a parking station with a credit card I can guarantee the fees were more than 8.5%.

We're talking about merchant fees here. Not what a credit card user pays etc.

That's exactly what I meant, I paid $.45 but the city likely paid $.31 in fees. Braintree for example would run 2.89% + $.30. That's a very high percentage (69%).

I always assumed it was to avoid card swipe fees. Margins at small restaurants can be incredibly thin, so a percent or two off each transaction could make a big difference.

Credit card fees cost money and credit card companies don't let retailers charge a higher price for credit card transactions, so the retailer has to get the hit.

However if your plumber or tradesperson gives a lower price for cash, then that's probably tax evasion.

Cash can be spent immediately, CC receipts not so much.

In the case of CC you'll get a payment 30 days later, so in VISA vs. cash, you're not getting the interest on the money, if you need to use the money you'll have to pay your bank interest, and you'll pay a processing fee on top of it.

About the only advantage CC has is that it's difficult for employees to steal from the register, so instead they steal the magstrips.

"In the case of VISA you'll get a payment 30 days later, so in VISA vs. cash, you're not getting the interest on the money,"

Wha? I get my Visa money within a day or two.

Amex and Diner's Club were the bad ones in my experience--regularly taking 30+ days to pay.

Avoid CC fees.

This seems to me like it would be an interesting hypothesis for a an experiment, but a bad idea to suggest/encourage in any "real" (IRS) capacity. Academically the idea is interesting, but the thought of this being actually implemented is a little scary.

... but you could say the same thing about a lot of other areas of research, too. Face detection at Facebook's scale, for example: really cool academically and potentially really scary in application.

You can slice this the other way also: using check-ins for a particular person to track where they visit. If their reported income is low, yet they check-in regularly at expensive restaurants and travel the globe, you've probably got a tax cheat in there somewhere or another.

Seriously, I have no idea why we discuss some of this stuff. It only serves to give ammo for other people to monitor and control us in ways they never have before.

I can see IRS building a fun little crawler. They already know everyone's name, address, place of employment, age, marital status, salary, the church they donate to. It would be interesting (and scary) to try to identify potential auditing target based on these check-ins and other stuff (from say, Facebook posts, twitter).

The tricky thing here, IMO, is that spending is based on assets + income, not just income. If I inherited $3m, make $50k/year, and spend $100k/year, it doesn't mean I'm lying about the $50k, it might just mean I have enough savings to live beyond what my income suggests.

"Seriously, I have no idea why we discuss some of this stuff."

There are times on HN when I laugh because people discuss things that could happen that in a practical sense stand little chance of happening. But they hang on the fact that it is possible the same way they might know someone who got ticketed for going 2 miles over the posted speed limit.

The IRS has (or had) a program called TCMP where they dissect everything on a tax return to see if it's kosher from many different angles. They will also see the "lifestyle" that you are leading and dig depending on that.

The IRS doesn't need foursquare to gather information to determine tax cheats they have other ways to do that if they wanted to. They could rely more on information on snitches or they could simply do a better job of correlating where someone lives with what they pay in taxes as a starting point (I believe they do some of this already but it's probably only at the zip code level). A good test for how they don't do this is to try and tip them off to someone who you think is cheating. Without hard evidence they won't do anything.

"check-in regularly at expensive restaurants and travel the globe"

You can buy some of this data in various forms the same way any marketer does. Or you can access property records. If someone has two houses and is reporting $20k in income that's a good place to start asking questions. You could then correlate it with what their reported income is as a starting point. From everything I've seen though the IRS doesn't operate that way.

I had the same reaction, this stuff makes me queasy.

I don't like tax cheats, but a tax cheat isn't going to imprison me for having the wrong opinions.

We have to remember that our set of laws was designed with a huge amount of slack. You can have a speed limit that nobody follows because hardly anybody gets caught. Once you start catching everybody, you've changed the social dynamic just by following the rules more strictly.

This kind of thing is a huge deal. For decades politicians have spit-balled new laws that, even though passed, don't amount to squat because nobody can figure them out or enforce them.

The more we automate this stuff the more miserable all of our lives will be. This kind of thing is not good at all. Can emphasize enough how these technological developments are a terrible evil (and I love technology)

There is a differentiation here that needs to be made. There is a problem if laws are so complicated and so strict and everyone ends up breaking them, the problem becomes a serious problem once the government has the ability to collect and record all times someone broke the law and then decide, later, for some possibly unrelated reason, to selectively enforce it.

Then a history of law breakage can be use for blackmail, threats, for silencing for coercing.

You are caught at an Occupy event and booked in. You are taken to the station and they pull up the fact that a speed camera caught you going 20 mph over speed limit. "5 years ago, it looks like IRS detected a tax discrepancy in your reporting, it would be a shame if they started an audit...maybe you should stop being involved in this movement", stuff like that.

I would almost more surprised if it won't happen sometime in the near future than if it does happen.

This trend is extremely dangerous. I could not agree more.

There's the massive intelligence collection and physical detection effort against terrorists.

There's been a more constrained system of criminal and criminal activity detection and correlation using GPS tracking, phone surveillance, drones, license plate recognition, and data mining of government and commercial databases.

The two are merging. Soon, everyone will be terrified of free speech and free association. Speech may still be free, but the government can always find some law you're violating.

As Breyer noted in his dissent in favor of granting cert in Rubin v. United States [1], "the complexity of modern federal criminal law... make it difficult for anyone to know in advance just when a particular set of statements might later appear... to be relevant to some such investigation." Even if you're guilty of no crimes and you have good lawyers, is participating in some political activity worth getting charged and brought to court?

[1] http://www.law.cornell.edu/supct/html/98-93.ZD.html

Yes. What'll happen is that since there is a practical limit on the number of people the government can put in prison, they'll just "defer" huge amounts of crime for people as long as they stay on good behavior.

Then, when the government chooses, they'll let the entire bomb drop on the guy. Of course there's also the blackmail and coercion potential (Looking at these traffic and land-use violations, you really don't want to criticize the governor in that way, do you?) It used to be that whichever party won the White House, the next year all the NGOs that supported the other party would be due for a tax audit. This increases that kind of thing by several orders of magnitude.

We voted for politicians who made promises and changes based on an existing manual system. All of those hundreds of years of voting and changing were inside of a system that doesn't exist any more. The feedback loop of the democracy was based on real humans going out there and enforcing the laws against real other humans, not the idea that whatever was said would automatically magically happen. Tom Smith in 1880 could get elected mayor on a platform of prohibiting spitting on the sidewalk because we all knew Tom, and we all knew that the entire purpose of the law was Joe, proliferate sidewalk spitter. Nobody cared about that law because once Tom got elected, he made Joe stop spitting and that was that. And who cared if it stayed on the books for the next 130 years?

But the world where laws were made by people, for people, to be enforced against people doesn't exist any more. Now we have technology that can take pictures all day long of people spitting, do facial recognition, and add it to a little file somewhere.

It's almost like we need to completely start over. We've built a machine that never could work, so it has all these patches and over-engineered pieces. Now that we're actually getting it to work, it'll eat us alive.

I'm all for unique uses of programmatic data collection, but this is a huge, huge stretch, with no practical implementation details.

Conversely if you have a grudge against a neighbouring business, make lots of fake accounts, check in there on foursquare to make the tax collectors think something is up.

Statistically it's horrible, however, it's probably good enough that some politician will make the IRS do this and now businesses will concern themselves with ensuring they have the right number of checkins.

Luckily it should be a great business to ensure that local businesses get the right number of checkins.

Reply on news.ycombinator.com