Friday, June 17, 2016

Jason Smith — What does it mean when we say money flows?

Even though the accounting is exact in the model above, I could make the wave travel faster or slower (and therefore the decay happen faster or slower) by changing the size of the debits and credits or changing the number of transactions per time step. The velocity of the wave is a free parameter not established by pure accounting. In the linked post, I called that free parameter Γ and was promptly attacked by the stock-flow consistent community for heresy.…
Information Transfer Economics
What does it mean when we say money flows?
Jason Smith

15 comments:

Matt Franko said...

Yeah you could see this coming....

Smith as a scientist is going to butt heads with the SFC "models" (so called) crowd....

Brian Romanchuk said...

He already butted heads with SFC models.

Working ftom memory, his big insight was that discrete time models have an embbeded time constant within them, related to the sample frequency. As someone who actually taught the mathematics regarding sampling theory, I pointed out that his observation was trivially obvious, and that everyone who studied SFC models would regard the point as a non-issue.

The discussion did not go well.

Peter Pan said...

Shouldn't he be comparing stocks with flows? Bill M. takes the time to explain the distinction between the two.

John said...

What Brian says is true and worrying. Far too many scientists think "trivially obvious" points are important insights, or more worrying think some of the methods that work well enough in science can be transferred to the social science domain. As the truly great Philip Anderson put it, More Is Different. But there is no accounting for stupidity, even amongst scientists, who in their infinite unwisdom believe concepts that are narrowly applicable to one manifestation of nature is applicable to all. Einstein used to point out that the problem with science is that their curricula no longer teach the history and philosophy of their respective subjects. Economics has the same problem.

Anybody who has ever sat in a high level science seminar will attest to witnessing arguments over utterly trivial things, or worse over what is obvious nonsense proclaimed as absolute truth. It becomes like a session of the South Korean or Turkish parliaments!

Tom Hickey said...

I think Jason Smith's flow analysis is interesting. It is stock flow consistent since it just reflects accounting entires in real time as flows that change stocks. This could be traced out unreal time if journals entries were both known and immediate. Now with digital records real time money flows in terms of transactions are traceable in principle, but much of the data is either propriety or behind privacy walls, so aggregate flows would remain unknown for practical purposes in real time.

His point about stock flow consistency as reflected in economic reports is a summary of flow collapsed in time, e.g., a reporting period, is correct. It does not capture the real-time flow as transactions are recorded in journals, transferred to general ledger and then summarized in accounting reports. Generally, economists work with accounting reports and the "flows" are changes over a period, usually monthly, quarterly or yearly.

The question is what difference does it make. What would change is this information were available.

Brian Romanchuk said...

(In my earlier comment, I was referring to an earlier debate.)

His latest article has a similar level of unreality. I have a hard time wading through the physics analogies, but it has something to do with money velocity. If all transactions consisted of people handing eachother dollar bills, this physics analogy might be useful. But that's not how large transactions work.

All business transactions are done via accounts payable, and the money only flows at a time point determined by the negotiations between the parties - 30 days, 90 days, etc.

Even in a highly computerised area like online book sales, payments follow the calendar. I see my ebook sales in something resembling real time, but I get paid quarterly, with a pretty serious lag. The person buying gets an instaneous hit on their credit card balance, but they only pay at the end of the month, and Amazon only gets paid periodically as well.

This will be all captured as credit entries on balance sheets, but money flows are out of sync from those balance sheet flows. Economists work with aggregated accounting data and not continuous time not because they do not understand Smith's theories about fundamental time constants, rather because accounting time aggregates best matches how to look at the data. We are not discussing subatomic particles, wherre the only sensible time frame is continuous.

Tom Hickey said...

It's hypothetical at this point, since the information is not publicly available or even centrally. But let's assume that all transactions are digital and recorded in real time, so the time lags would also be known. Let's further assume that this information is centralized and make public without identifying the specific transactions and transactors to protect privacy. What difference would that make. How could the information be used.

Alternatively, assume that his information is only made available centrally, say to the monetary authority (cb) or fiscal authority (or both). What difference would that make for policy.

Right now we stand in the middle of a spectrum that began with markets incorporating stocks and flows of both goods and money, with no standard accounting practice and no reporting, so the stocks and flows were largely unknown and based on very imprecise estimates if at all. We can envision a time when everything is know about everything through real time surveillance and big data collection with AI processing and evaluating all this instantaneously and reporting on it, to those have access anyway.

What does all this suggest.

Brian Romanchuk said...

Tom,

Most large transactions are credit transactions. For my consulting work, the only time that a transaction would show up is when the cheque goes through. But in reality, the transaction was negotiated ahead of the work, I do the work, I send an invoice, and only after a period of time the cheque goes through. Neither myself nor my client is going to run off to some central database to register the earlier steps. They would be captured on accounting statements, which would show up if we were publicly-traded corporations, or on our income tax statements. (You need to use accrual accounting, so if I did half of the work on a contract before year end, I would show half of the contract's value as income, even though I may not have sent an invoice.)

But even if we had the information, we would rapidly be swamped. Things like securities purchases do not affect income, but they dwarf income flows in size. Those sorts of transactions would have to be eliminated from the data set in order for it to be compared to things like GDP. Accounting principles do exactly that; we want to look at things like income statements, not every single monetary flow. Changes in how things are paid for could cause huge disruptions in the data set.

In reality, most economists do not even want to look at non-seasonally adjusted data, and the seasonal effects on data are even smaller than what would happen if we could see the raw transaction level data. Mike Norman (and friends) is one of the only persons I know of that looks at the daily Treasury data. Hellfire, most economists did not even look at the average maturity data of Treasury issuance when they were blathering on about QE.



.

Tom Hickey said...

Thanks, Brian.

My sense is that governments want to capture as much as they can digitally in real time as part of total awareness and they are thinking of more ways to make this happen, like eliminating cash. It's not a big step to requiring everything (both money exchange and barter) to be available to the authorities digitally in real time on settlement.

Matt Franko said...

What the SFC people are doing imo is that they are effectively saying:

"GDP has been increasing over the last few quarters so when we next measure GDP ex post GDP will show an increase...."



Brian Romanchuk said...

Tom -
As another example of the difficulties one would face, imagine how a house purchase dwarfs regular spending. If those house payments are mixed up with regular cash transfers, then house sales would pollute your data. And home sales are fairly irregular (at least in Canada). A late snowstorm can delay a significant number of home sales into the next month, blowing out the usual seasonal pattern. We probably can only work with certain data sets in real time, like retail sales.

Matthew,

SFC models are a framework for building economic models, but the behavioural relationships can take many forms. There is no magical behavioural laws that can perfectly match real world data, unfortunately. However, we at least develop a good idea of what the tradeoffs are. And you can also develop models that show why forecasting is inherently difficult; mainstream models, which rely on a representative household forecasting everything perfectly, cannot do that.

Tom Hickey said...

As another example of the difficulties one would face, imagine how a house purchase dwarfs regular spending. If those house payments are mixed up with regular cash transfers, then house sales would pollute your data. And home sales are fairly irregular (at least in Canada). A late snowstorm can delay a significant number of home sales into the next month, blowing out the usual seasonal pattern. We probably can only work with certain data sets in real time, like retail sales.

Actually, governments are increasingly interested in house purchases since that is where a lot of the laundered money ends up. I am pretty sure that they are figuring out how to include those types of flows, too.

Governments have figured out that the way to control crime and also terrorism, etc., is through total awareness in real time, including all financial flows. This is in the works right now. This is as important as military right now, along with cybersecurity. They may never achieve perfection in this regard but they will be working assiduously to close all the holes as soon as they are discovered.

If this info is available to security forces, it's a good bet it will be used for other purposes, too, proprietary info and privacy be damned.

Tom Hickey said...

governments are increasingly interested in house purchases since that is where a lot of the laundered money ends up.

BTW, I am not just talking about high-end purchases in cities like London and NY. A lot of West Coast housing has been bought up by Asians, often through agents, and the suspicion is that much of the $ used was laundered. Lots of drug money too.

Tom Hickey said...

Oh, and lots of Asian money has been flowing into Canada, too.

The US, UK, Canadian, Australian, and New Zealand intelligence community is essential the same community, It's called "Five Eyes." Apparently German intelligence was just rolled in, too, but that may be as a subsidiary rather than a partner.

Tom Hickey said...

There is of course the question as to why the intelligence/security/law enforcement community is interested in collecting all the data before processing, selection, and analysis.

That's just the way the process works. Gathering less then total information risks missing something. This creates a problem for filtering later but that's a major portion of what intelligence does.

Clearly, once the data is collected, it can be used in many ways as information for a variety of purposes. That's what what dogs are concerned with. The initial purpose may be quite reasonable but then there's mission creep and finally exploitation.