PDA

View Full Version : No correlation between deficit and dollar = weasel words



bart
08-11-06, 02:24 AM
Through a silly oversight, I neglected to post an article I wrote about a month ago here - silly me.

---------------------------------------------

No correlation between deficit and dollar = weasel words

Not rocket science...

Of course, even the quickest glance at the charts of the US current account deficit and the dollar shows that they're not correlated well.

http://www.nowandfutures.com/download/cad_usdx.png


... until other deficits and a flow are added in

Let's change it around some and look at a combination of the US trade deficit, the US budget deficit and something called Treasury International Capital (TIC) flows:

http://www.nowandfutures.com/images/tic_trade_budget_usdx1992-current_old.png


And now we see a very good visual correlation and an "un-weasel-ing" of the relationship between deficits and the dollar. Yes, there really is a relationship between deficits and the dollar.

So what's this Treasury International Capital flow thing? Here (http://www.ustreas.gov/tic/ticsec.html) is the home page of it at the US Treasury. Basically, its just a measure of the net investment flows into and out of the US and mostly contains stocks & bonds.

What's with the chart title - "The international dollar - income & expense"?

In order to value a business or company and see what it's worth and how it will do in the future, three of the most important factors are sales, expenses and profit. Profit is basically sales minus expenses. If we go way out there and assume the entire US is a company, and pretend that the US dollar is its stock, then we have another way to look at the international value of the dollar.

The black line on the chart is the monthly TIC flow from other countries (income), with both the trade and budget deficits (expenses) subtracted. So you say - "So what?"... well, by doing that we show an income and expense statement for the US dollar itself. Any numbers above zero on the left hand scale mean a profit and if the number comes in below zero then there has been a loss.

In other words, if we back way off from the dollar and look at it from a 30,000 foot level as the stock of the USA itself, we need to figure out what would represent sales and what would represent expenses. We pretend that TIC flows are income and that the combination of the trade and budget deficit are the expenses.

Then, TIC minus (trade + budget deficit) represents net profit or loss of the dollar itself. Well, what happens when a company has losses - their stock price goes down... and the same thing has happened with the international value of the dollar since early 2002. When there was a consistent net profit between 1997 and 2001, the dollar value rose.

Some may say that what we're doing is way too simple and there's some truth there... but the bottom line is that it does work and does track and has tracked the value of the dollar for almost 15 years.

M3b update

It has been almost three months since the original "M3 is back" article was published - here's the current chart.

http://www.nowandfutures.com/images/m3b.png





A favorite quote

"The last duty of a central banker is to tell the public the truth."
-- Alan Blinder, Vice Chairman of the Federal Reserve, on PBS’s Nightly Business Report in 1994

jk
08-11-06, 10:27 AM
bart, interesting charts but there's one piece that doesn't make sense to me. in your model of the usa as a corporation, why subtract the budget deficit as an expense? to the degree that the deficit is funded INTERNALLY [i.e. by purchases of newly issued tbonds and tbills by domestic individuals and institutions] it doesn't represent the kind of flow-across-the-corporate-boundary that is inherent in "expenses" in your model. am i missing something?

more specifically, the tic number presumably captures foreign purchases of newly issued tdebt. then you subtract the budget deficit as well as the trade deficit. since the budget deficit = the sum of new tdebt purchases,
your graph shows tic-trade deficit-budget deficit = tic-trade deficit-domestic purchases of new tdebt-foreign purchases of new tdebt

while the trade deficit and foreign purchases of new debt fits your "expense" model, i don't see how domestic purchases of new tdebt fits.

bart
08-11-06, 11:12 AM
bart, interesting charts but there's one piece that doesn't make sense to me. in your model of the usa as a corporation, why subtract the budget deficit as an expense? to the degree that the deficit is funded INTERNALLY [i.e. by purchases of newly issued tbonds and tbills by domestic individuals and institutions] it doesn't represent the kind of flow-across-the-corporate-boundary that is inherent in "expenses" in your model. am i missing something?

more specifically, the tic number presumably captures foreign purchases of newly issued tdebt. then you subtract the budget deficit as well as the trade deficit. since the budget deficit = the sum of new tdebt purchases,
your graph shows tic-trade deficit-budget deficit = tic-trade deficit-domestic purchases of new tdebt-foreign purchases of new tdebt

while the trade deficit and foreign purchases of new debt fits your "expense" model, i don't see how domestic purchases of new tdebt fits.

Mostly, the parallel with a profit or corporate model is borrowing - with a company its borrowing from a bank, issuing bonds, etc. and with the government its borrowing measured by the budget deficit or surplus. Plus, funding is very much not all from internal sources. The last time I checked, it was around 45% external.

Additionally, it just plain works way better and has a much better correlation than using the trade deficit alone... as you can see with this chart:

http://www.nowandfutures.com/images/tic_trade_usdx.png

jk
08-11-06, 12:55 PM
i see the graph as working empirically. [as the joke goes: i know it works in practice, but will it work in theory?] i'm trying to figure out why it works. the corporate analogy doesn't fly. if borrowing is funded 55% internally, that would correspond to a corporation funding investment in one division with earnings from another division. of course for a corporation investment is handled differently than expenses, but that won't apply to our governments funding of current expenses. [of course our government also doesn't have a separate capital budget, but now this gets very far afield.]

on second thought, i guess i'm also not convinced the graph works even empirically [let alone in theory]. it is clear from your other graph that current account doesn't explain the dollar's movement, but the constructed variable of tlc-[budget deficit+trade deficit] may be just as uncorrelated, but in smaller chunks. i.e. it disagrees with the dollar's movement from 72-74, looks correlated from 74-77, may or may not correlate 77-00, doesn't move together 00-02, and is likely uncorrelated from 02-present. there's so much noise in your variable you'd need to do some smoothing, pulling out what looks like a periodic element that flows through even to the ma. i don't know any statistics, but some kind of fourier analysis pulling out the fast cycle would likely work better than an ma.

bart
08-11-06, 01:33 PM
i see the graph as working empirically. [as the joke goes: i know it works in practice, but will it work in theory?] i'm trying to figure out why it works. the corporate analogy doesn't fly. if borrowing is funded 55% internally, that would correspond to a corporation funding investment in one division with earnings from another division. of course for a corporation investment is handled differently than expenses, but that won't apply to our governments funding of current expenses. [of course our government also doesn't have a separate capital budget, but now this gets very far afield.]

on second thought, i guess i'm also not convinced the graph works even empirically [let alone in theory]. it is clear from your other graph that current account doesn't explain the dollar's movement, but the constructed variable of tlc-[budget deficit+trade deficit] may be just as uncorrelated, but in smaller chunks. i.e. it disagrees with the dollar's movement from 72-74, looks correlated from 74-77, may or may not correlate 77-00, doesn't move together 00-02, and is likely uncorrelated from 02-present. there's so much noise in your variable you'd need to do some smoothing, pulling out what looks like a periodic element that flows through even to the ma. i don't know any statistics, but some kind of fourier analysis pulling out the fast cycle would likely work better than an ma.


Maybe I'm wrong here, but are you a believer in the general "deficits don't matter" view?

I think where you're missing it on the corporate analogy is that its not the 55% being borrowed from other divisions, but rather from the employees.

I have no clue where you're getting the "72-74, looks correlated from 74-77, may or may not correlate 77-00" - all the graphs only go back to 1992?

I have done different things like EMAs, longer MAs, lags and various regression tricks that do align the data much better, but decided to make public just the raw data and a simple MA so that I wouldn't be accused of excess statistical "special moments".

You're obviously welcome to believe or disbelieve the correlation, but my internal/unpublished work show it to be quite useful and of predictive value... and it does work for other currencies for me too.

jk
08-11-06, 08:14 PM
Maybe I'm wrong here, but are you a believer in the general "deficits don't matter" view?
not at all. i think they matter a lot. i'm just trying to understand how you're using them.


I think where you're missing it on the corporate analogy is that its not the 55% being borrowed from other divisions, but rather from the employees.
i still think you're better off without bothering with the corporate analogy. because if the 55% is being borrowed from the employees, who in this analogy is liable for the trade deficit?


I have no clue where you're getting the "72-74, looks correlated from 74-77, may or may not correlate 77-00" - all the graphs only go back to 1992?
sorry, i glanced down and misread the numbers. it's '92-'94, etc.


I have done different things like EMAs, longer MAs, lags and various regression tricks that do align the data much better, but decided to make public just the raw data and a simple MA so that I wouldn't be accused of excess statistical "special moments".

You're obviously welcome to believe or disbelieve the correlation, but my internal/unpublished work show it to be quite useful and of predictive value... and it does work for other currencies for me too.
it's not clear to me, at least, looking at the graph, that there really is a correlation. and i am not knowledgeable in statistics to perform any kind of analysis even if i had the raw data.

if you're saying that you've used it in a forward-looking, predictive way [and it certainly sounds like you're saying that] then it is clearly useful. if it works, it works. i'm just struggling with why it works. if in fact your more complicated manipulations of the data yielded a better fit, perhaps the specifics of those manipulations would shed some light on the underlying relationships.

bart
08-11-06, 09:14 PM
not at all. i think they matter a lot. i'm just trying to understand how you're using them.

Cool, I honestly wasn't sure.



i still think you're better off without bothering with the corporate analogy. because if the 55% is being borrowed from the employees, who in this analogy is liable for the trade deficit?

Fair enough - is there another analogy that you think could be better?

The employee part of the analogy definitely does break down when you start talking about liabilities and such... but the major point of the chart is to show a correlation with deficits and currency values. Keep in mind too that the article and chart are intended for a very broad audience. many of whom I think would be blown off by any significant complexity, or a real economic analysis/paper.




it's not clear to me, at least, looking at the graph, that there really is a correlation. and i am not knowledgeable in statistics to perform any kind of analysis even if i had the raw data.

if you're saying that you've used it in a forward-looking, predictive way [and it certainly sounds like you're saying that] then it is clearly useful. if it works, it works. i'm just struggling with why it works. if in fact your more complicated manipulations of the data yielded a better fit, perhaps the specifics of those manipulations would shed some light on the underlying relationships.

The full set of private charts I have do indeed show a much better correlation (.87) than the published one (.59), but they include a number of "manipulation" factors which wouldn't go over well with folk who aren't fond of tinfoil hat type factors.
But when I just do a weighted and exponential moving average on the same data and apply about a 10 month lag, the correlation goes from .59 to .73... and again, I didn't publish that one due to the broad target audience... and perhaps I should have.
Right now, the full chart is pointing at a higher dollar - for what its worth.

jk
08-12-06, 01:44 PM
The full set of private charts I have do indeed show a much better correlation (.87) than the published one (.59), but they include a number of "manipulation" factors which wouldn't go over well with folk who aren't fond of tinfoil hat type factors.
But when I just do a weighted and exponential moving average on the same data and apply about a 10 month lag, the correlation goes from .59 to .73... and again, I didn't publish that one due to the broad target audience... and perhaps I should have.
Right now, the full chart is pointing at a higher dollar - for what its worth.

as i said, it certainly looks like there's some kind of periodic movement embedded in the chart. if may be that the weighted 10mo lag captures roughly a half cycle, and so cancels out the periodic fast cycle.

it's interesting that it's pointing to a higher dollar. of course the whole world is predicting dollar weakness, which would support dollar strength.

i'm going to read your post overagain. is there more discussion of this on your website? [i think i'm going to go over there and look around but if you have a link, thanks in advance.]

bart
08-12-06, 05:09 PM
as i said, it certainly looks like there's some kind of periodic movement embedded in the chart. if may be that the weighted 10mo lag captures roughly a half cycle, and so cancels out the periodic fast cycle.

it's interesting that it's pointing to a higher dollar. of course the whole world is predicting dollar weakness, which would support dollar strength.

i'm going to read your post overagain. is there more discussion of this on your website? [i think i'm going to go over there and look around but if you have a link, thanks in advance.]

The article was pretty much cut & pasted from the original but I left out a few lines related to where other related and important charts exist on my site.
The full article (http://www.NowAndFutures.com/articles/20060630No_correlation_between_deficit_and_dollar_ =_weasel_words.html)
Key stats (http://www.NowAndFutures.com/key_stats.html)


hmmm... you definitely have me going on this. I think I need to somehow find some time to publish a version 2.0 that both tightens up the correlation and also leaves out some of the tinfoil hat elements.

My site is quite light on discussion or conclusions, partly since I'm a proponent of "educating a man to fish" and also since I haven't finished a number of other projects (like having more charts of the rest of the key central banks).

bart
08-16-06, 10:43 PM
as i said, it certainly looks like there's some kind of periodic movement embedded in the chart. if may be that the weighted 10mo lag captures roughly a half cycle, and so cancels out the periodic fast cycle.

it's interesting that it's pointing to a higher dollar. of course the whole world is predicting dollar weakness, which would support dollar strength.

i'm going to read your post overagain. is there more discussion of this on your website? [i think i'm going to go over there and look around but if you have a link, thanks in advance.]


I got a small shipment of round tuits in and the chart has had some small modifications made in order to show a better correlation. The 3 month moving average was changed to a 12 month moving average, and a 6 month lag was added. The correlation is now a lot clearer and, as a bonus, we now have a mildly predictive chart.

I'm linking it below again, for convenience and I'll also edit the initial post so that it continues to link to the original chart.

http://www.nowandfutures.com/images/tic_trade_budget_usdx1992-current.png


Do continue to keep in mind that the basic purpose of the original article and the analogy was to show that deficits do matter, and that there is a correlation.
Perhaps now that some time has elapsed, you might have an alternate analogy in mind?

Finster
08-19-06, 12:49 PM
I got a small shipment of round tuits in and the chart has had some small modifications made in order to show a better correlation. The 3 month moving average was changed to a 12 month moving average, and a 6 month lag was added. The correlation is now a lot clearer and, as a bonus, we now have a mildly predictive chart.

I'm linking it below again, for convenience and I'll also edit the initial post so that it continues to link to the original chart.

...

Do continue to keep in mind that the basic purpose of the original article and the analogy was to show that deficits do matter, and that there is a correlation.
Perhaps now that some time has elapsed, you might have an alternate analogy in mind?

Deficits DO matter! Either one of two things - or some combination of them - has to happen. Government borrowing drains the available pool of credit and exerts upward pressure on interest rates, or the pool of available credit is artifically expanded and exerts upward pressure on inflation.

In the past few years we've seen mostly the latter. So we wind up paying for those deficits at the pump and in our monthly bills. Anybody who thinks "deficits don't matter" just isn't looking in the right place.

bart
08-19-06, 02:14 PM
Deficits DO matter! Either one of two things - or some combination of them - has to happen. Government borrowing drains the available pool of credit and exerts upward pressure on interest rates, or the pool of available credit is artifically expanded and exerts upward pressure on inflation.

In the past few years we've seen mostly the latter. So we wind up paying for those deficits at the pump and in our monthly bills. Anybody who thinks "deficits don't matter" just isn't looking in the right place.


There is at least one other short/medium term alternative, and the new chart is slightly pointing at it. There are indications that the budget deficit is dropping due to unusually high tax receipts. There are also indications that the current liquidity crunch is helping the TIC flows, via the old datum about the dollar and Treasuries being a safe refuge or haven.

Do note that I'm not saying that the dollar really is a safe refuge or haven based on the true long term fundamentals, just that sentiment is a major factor and that manipulation of inflationary expectations is part of the game of central banksters.


You'll get a kick out of the original impetus of the article - I actually have two friends who stated that "deficits don't matter" at a lunch a few months ago... and I couldn't let it pass. They're still not 100% convinced but the correlation blew them away and has them thinking at least.

Finster
08-19-06, 02:25 PM
There is at least one other short/medium term alternative, and the new chart is slightly pointing at it. There are indications that the budget deficit is dropping due to unusually high tax receipts. There are also indications that the current liquidity crunch is helping the TIC flows, via the old datum about the dollar and Treasuries being a safe refuge or haven.

But if tax receipts increase, then the deficit is reduced, so to that extent there are no longer deficits. A bit circular, no? And regardless of where the source of the credit is - domestic or international - it's still a pool of credit that is being drawn from. So we still have only two alternatives.

It's a matter of exhausting the logical possibilities. If I borrow a widget for my own use, somewhere, somebody, has to forego possession and use of that widget or a new widget must be created. Substituting "the government" for "I" and "dollar" for "widget" in the statement does not alter the logic. Either the pool of existing dollars available for lending elsewhere is depleted by the act of its being borrowed by the government or new dollars must be created.

bart
08-19-06, 03:41 PM
But if tax receipts increase, then the deficit is reduced, so to that extent there are no longer deficits. A bit circular, no? And regardless of where the source of the credit is - domestic or international - it's still a pool of credit that is being drawn from. So we still have only two alternatives.

It's a matter of exhausting the logical possibilities. If I borrow a widget for my own use, somewhere, somebody, has to forego possession and use of that widget or a new widget must be created. Substituting "the government" for "I" and "dollar" for "widget" in the statement does not alter the logic. Either the pool of existing dollars available for lending elsewhere is depleted by the act of its being borrowed by the government or new dollars must be created.

You lost me here... tax receipts don't seem to be analogous to borrowing... although those consumer or business tax expenses do decrease money available to loan or invest. Perhaps that is what you're driving at?

If the deficit goes down and TIC flows maintain or go up, then the chart and relatonship shows that the dollar will gain relative value - that's my point.

Finster
08-19-06, 03:48 PM
You lost me here... tax receipts don't seem to be analogous to borrowing... although those consumer or business tax expenses do decrease money available to loan or invest. Perhaps that is what you're driving at?

I'm just saying here that to the extent there are tax receipts to cover spending, there is no deficit to begin with. Deficits are the difference between what the government spends and what it takes in in taxes - i.e. what it borrows. So taxes not an alternative to borrowing or inflation as a means of funding deficits.


If the deficit goes down and TIC flows maintain or go up, then the chart and relatonship shows that the dollar will gain relative value - that's my point.

Not disputed; in fact that makes perfect sense. My point was an ancillary one - simply that deficits represent borrowed money, and that that money either has to be foregone by a real lender or created out of nothing (inflation). This is where are lot of that deficit money comes from - right from the thin air of the Fed and banking system. Consequently, we have had rampant inflation, as reflected in house prices, oil prices, etceteras.

bart
08-19-06, 04:10 PM
I'm just saying here that to the extent there are tax receipts to cover spending, there is no deficit to begin with. Deficits are the difference between what the government spends and what it takes in in taxes - i.e. what it borrows. So taxes not an alternative to borrowing or inflation as a means of funding deficits.



Not disputed; in fact that makes perfect sense. My point was an ancillary one - simply that deficits represent borrowed money, and that that money either has to be foregone by a real lender or created out of nothing (inflation). This is where are lot of that deficit money comes from - right from the thin air of the Fed and banking system. Consequently, we have had rampant inflation, as reflected in house prices, oil prices, etceteras.


Cool... all better now.

I was considering the possibility that I was being set up for a *finn*... ;)

Finster
08-19-06, 04:20 PM
Cool... all better now.

I was considering the possibility that I was being set up for a *finn*... ;)

Not this time at least ... you still have your get-out-of-finned-free card ... :cool:

jk
08-19-06, 07:28 PM
Bart, I finally got around to thinking about your graph again. i'm just thinking this out, thinking "out loud" as it were, so i'd be interested in and open to feedback here. i'm trying to construct some theory based way to understand what you've shown:

you show tic minus [trade deficit+ budget deficit] correlated with the dollar index. First, I’m curious if you tried current account deficit instead of trade deficit. I would imagine that current account wouldn’t be that different, but I’m wondering if it would be even better.

Anyway, here’s a go at my thinking about this:

Tic= net change in dollar based foreign portfolio investments. Important to note that this leaves out direct investment, e.g. daimler acquires chrysler.

Let’s regroup tic-[trade+budget deficits] = [tic-trade deficit]-budget deficit

Tic-trade deficit = net change in dollar based foreign portfolio investments – the number of dollars thrust into foreign hands via our trade [current account is more accurate] deficit. In order for foreign entities to acquire dollar portfolio investments greater than the trade deficit, the foreign entities must borrow dollars from u.s. domestic lenders. So tic-trade deficit represents excess foreign demand for u.s. dollar loans.

The budget deficit represents excess domestic demand for u.s. dollar loans. This is a restatement of the “crowding out” description of the deficit.

If foreign demand for u.s. dollar loans exceeds domestic demand for u.s. dollar loans, the value of the dollar rises. If domestic demand exceeds foreign demand, the dollar index falls. ta-da!

bart
08-19-06, 08:36 PM
Bart, I finally got around to thinking about your graph again. i'm just thinking this out, thinking "out loud" as it were, so i'd be interested in and open to feedback here. i'm trying to construct some theory based way to understand what you've shown:

you show tic minus [trade deficit+ budget deficit] correlated with the dollar index. First, I’m curious if you tried current account deficit instead of trade deficit. I would imagine that current account wouldn’t be that different, but I’m wondering if it would be even better.

Anyway, here’s a go at my thinking about this:

Tic= net change in dollar based foreign portfolio investments. Important to note that this leaves out direct investment, e.g. daimler acquires chrysler.

Let’s regroup tic-[trade+budget deficits] = [tic-trade deficit]-budget deficit

Tic-trade deficit = net change in dollar based foreign portfolio investments – the number of dollars thrust into foreign hands via our trade [current account is more accurate] deficit. In order for foreign entities to acquire dollar portfolio investments greater than the trade deficit, the foreign entities must borrow dollars from u.s. domestic lenders. So tic-trade deficit represents excess foreign demand for u.s. dollar loans.

The budget deficit represents excess domestic demand for u.s. dollar loans. This is a restatement of the “crowding out” description of the deficit.

If foreign demand for u.s. dollar loans exceeds domestic demand for u.s. dollar loans, the value of the dollar rises. If domestic demand exceeds foreign demand, the dollar index falls. ta-da!

http://www.nowandfutures.com/grins/crowd_cheer.wav :-)

Nicely done, jk. Your way of looking at it is fine by me and aligns well with more conventional economic thinking too, and would also likely make more sense to those who prefer that view. Ye olde bottom line after all is having your own understanding and not blindly accepting some interpretation that doesn't work for you.

You're also quite correct that TIC is far from a complete stat and its very possible and likely that if stuff like direct investment was included, the correlation would be even higher. Quite frankly, I was surprised when I put the original chart together and saw how high the correlation was with just the raw numbers.

As far as using the current account deficit, it may very well be a touch more accurate but it has two severe problems that exclude it for my purposes - its only a quarterly stat and also has quite a large reporting lag after the end of a fiscal quarter.

But here's the two charts from the Fed - they may help answer your question:

http://research.stlouisfed.org/fred2/data/BOPGSTB_Max.png

http://research.stlouisfed.org/fred2/data/BOPBCA_Max.png




I'm tempted to add your view as a subnote to the article after some very minor cleanup - if that is ok with you?

jk
08-19-06, 11:14 PM
http://www.nowandfutures.com/grins/crowd_cheer.wav :-)

Nicely done, jk. Your way of looking at it is fine by me and aligns well with more conventional economic thinking too, and would also likely make more sense to those who prefer that view. Ye olde bottom line after all is having your own understanding and not blindly accepting some interpretation that doesn't work for you.

You're also quite correct that TIC is far from a complete stat and its very possible and likely that if stuff like direct investment was included, the correlation would be even higher. Quite frankly, I was surprised when I put the original chart together and saw how high the correlation was with just the raw numbers.

As far as using the current account deficit, it may very well be a touch more accurate but it has two severe problems that exclude it for my purposes - its only a quarterly stat and also has quite a large reporting lag after the end of a fiscal quarter.

But here's the two charts from the Fed - they may help answer your question:

http://research.stlouisfed.org/fred2/data/BOPGSTB_Max.png

http://research.stlouisfed.org/fred2/data/BOPBCA_Max.png




I'm tempted to add your view as a subnote to the article after some very minor cleanup - if that is ok with you?

sure, it's ok with me.