Economics Discussion Thread

[Great Minarchistan]

Your notion of it is false since the trend curves itself on a "U". :-)

I never argued for "the smaller the better". I said that US performed better under small(er) government, and I was right as you can see. :-)

"the trend line indicates nothing changes in a change from x=0 to x=1" is a ridiculous statement, precisely because the trend-line has such a low degree of correlation.

As if the polynomial curve was flat.

The crowding out effect is a myth.

Oops:

Contrary to Keynesian expectations, not only is fiscal stimulus through government budget deficits ineffective, as reported by Barro and Redlick (2009), it was found to be counter-productive for the US economy (Sy, 2016b). Fiscal stimulus turned out to be an economic depressant. The explanation lies in what the spending was actually stimulating.

The issuance of US government bonds from budget deficits led to more financial investment and less real investment in economic production, largely as a substitution effect, independent of interest rate policy. The borrowing by the US government was used mainly to finance welfare consumption. Decades of US government budget deficit spending led to ever rising personal consumption, but falling net investment, depressing economic growth (Sy, 2016b).

The type of statistical relationship between government budget deficits and economic growth rates found for the US economy over decades is also found across the 35 countries over a ten-year period, as the following chart shows.

(Image)

A linear regression analysis of the data shows a statistically significant anti-correlation of -44 percent, with an R-square of 0.19, suggesting that one percent increase in government budget deficit will likely cause a decline of -0.36 percent in annual economic growth. (Including China and India would reduce the anti-correlation to -35 percent and the R-square to 0.13 for the regression.) Even though there could be different reasons for this causality for each country, the empirical relationship is unlikely to be a statistical artefact because of the underlying causes identified for the US economy. We conjecture that, through various means (not specified), various government budget deficits lead ultimately to increased consumption, reducing net investment and hence economic growth, in accordance with the axiom of macroeconomics (Sy, 2016b).

Budget deficits are automatically destabilizing, in the sense that any unexpected economic contraction would automatically increase the budget deficits relative to GDP, which then exert a greater depressive effect on economic growth. Indeed, we have identified government budget deficits as the gravitational force of Keynesian black holes (Sy, 2016b), because budget deficits tend to increase, just like the increasing force of gravitation near a black hole. Persistent government budget deficits have created, around the world, mountains of government debt, which could be the fuel for a global financial supernova.

Banks can create unlimited amounts of money out of thin air

At the cost of solvency rate. Demanding such an egregious amount of extra spending believing banks will cover it through fractional reserves would imply such a low solvency maybe ten depositors withdrawing their money would cause a massive panic.

As an identity, public excess expenditure is private excess savings

It's not like deficits exist, right?

a government deficit, for example, increases corporate profits by mathematical necessity

Interesting. Do you have some empirical data to back your proof?

[Arkolon]

Your notion of it is false since the trend curves itself on a "U". :-)

I never argued for "the smaller the better". I said that US performed better under small(er) government, and I was right as you can see. :-)

How is this statement not 'the smaller the better [GDP growth]'?

"the trend line indicates nothing changes in a change from x=0 to x=1" is a ridiculous statement, precisely because the trend-line has such a low degree of correlation.

As if the polynomial curve was flat.

You have totally missed the point, I gave an example to show how a trend-line means nothing if the degree of correlation is very weak. Your polynomial line of best fit is too poor to treat the U-shaped curve as even remotely meaningful, for the reasons I have already outlined.

The crowding out effect is a myth.

Oops:

Contrary to Keynesian expectations, not only is fiscal stimulus through government budget deficits ineffective, as reported by Barro and Redlick (2009), it was found to be counter-productive for the US economy (Sy, 2016b). Fiscal stimulus turned out to be an economic depressant. The explanation lies in what the spending was actually stimulating.

The issuance of US government bonds from budget deficits led to more financial investment and less real investment in economic production, largely as a substitution effect, independent of interest rate policy. The borrowing by the US government was used mainly to finance welfare consumption. Decades of US government budget deficit spending led to ever rising personal consumption, but falling net investment, depressing economic growth (Sy, 2016b).

The type of statistical relationship between government budget deficits and economic growth rates found for the US economy over decades is also found across the 35 countries over a ten-year period, as the following chart shows.

(Image)

A linear regression analysis of the data shows a statistically significant anti-correlation of -44 percent, with an R-square of 0.19, suggesting that one percent increase in government budget deficit will likely cause a decline of -0.36 percent in annual economic growth. (Including China and India would reduce the anti-correlation to -35 percent and the R-square to 0.13 for the regression.) Even though there could be different reasons for this causality for each country, the empirical relationship is unlikely to be a statistical artefact because of the underlying causes identified for the US economy. We conjecture that, through various means (not specified), various government budget deficits lead ultimately to increased consumption, reducing net investment and hence economic growth, in accordance with the axiom of macroeconomics (Sy, 2016b).

Budget deficits are automatically destabilizing, in the sense that any unexpected economic contraction would automatically increase the budget deficits relative to GDP, which then exert a greater depressive effect on economic growth. Indeed, we have identified government budget deficits as the gravitational force of Keynesian black holes (Sy, 2016b), because budget deficits tend to increase, just like the increasing force of gravitation near a black hole. Persistent government budget deficits have created, around the world, mountains of government debt, which could be the fuel for a global financial supernova.

This (unsourced, might I add) glorification of Robert Barro's neoclassical economics only tangentially refers to the crowding out effect (it is mentioned in passing in one single sentence which isn't even the argument's crux, it totally assumes the existence of the effect), and more principally rests on the same poorly-correlated data you have presented me to make its case. An r-squared of 13% is not very convincing.

Banks can create unlimited amounts of money out of thin air

At the cost of solvency rate. Demanding such an egregious amount of extra spending believing banks will cover it through fractional reserves would imply such a low solvency maybe ten depositors withdrawing their money would cause a massive panic.

Fractional reserve banking is also a semi-myth, in that it is misleading to focus on reserves and thinking banks create money leading to some kind of money multiplier. You can have zero-reserve banking for example. But even in FRB, imagine a commercial bank that loans out (to the economy, or to other banks) $100bn and receives $90bn in deposits (to be in accordance with a 10% reserve rate); if it wants to increase lending to $120bn without new deposits, it can take a loan from the interbank market or the central bank of $2bn, with this money being created out of thin air. Panics are a problem if a bank's access to money from other banks runs dry, especially if their lending practices are heavily based on it (eg Northern Rock).

As an identity, public excess expenditure is private excess savings

It's not like deficits exist, right?

I have no clue what you're trying to say here. I literally just said one person's deficit is another person's surplus, and you're telling me ... what, exactly, that deficits exist?

a government deficit, for example, increases corporate profits by mathematical necessity

Interesting. Do you have some empirical data to back your proof?

Do you know what this being a mathematical identity means? It's true by definition. I can show it to you now.

GDP = Consumption (C) + Household Investment (H.I) + Corporate Investment (C.I) + Government expenditure (G) + Net exports (X - M)
GDP = Consumption (C) + Household Savings (H.S) + Corporate Savings (C.S) + Tax Income (T)
... therefore
C + H.I + C.I + G + (X - M) = C + H.S + C.S + T
--> (C.S - C.I) = (H.I - H.S) + (G - T) + (X - M)
--> Corporate Savings (profit) = C.I + (H.I - H.S) + (G - T) + (X - M)

It is true by definition. Although this is unnecessary, because it is based on identities, there is this link which shows the relationship, comparing what government accountants estimated to what must be true by definition. Unsurprisingly, accounting for statistical error, the numbers match exactly.

[Great Minarchistan]

How is this statement not 'the smaller the better [GDP growth]'?

Noting that the current US govt spending is at 40% of the GDP, anything below that is considered smaller government, basing the current year as a reference.

You have totally missed the point, I gave an example to show how a trend-line means nothing if the degree of correlation is very weak. Your polynomial line of best fit is too poor to treat the U-shaped curve as even remotely meaningful, for the reasons I have already outlined.

Now let's be honest, what were you expecting? An increase in government leading to -20% growth perspectives?

This (unsourced, might I add)

http://www.asepp.com/much-capitalism-socialism/

glorification of Robert Barro's neoclassical economics only tangentially refers to the crowding out effect (it is mentioned in passing in one single sentence which isn't even the argument's crux, it totally assumes the existence of the effect)

And his hypothesis adds up to mine, that comproved with empirical data show a meaningful evidence of crowding out on the economy.

and more principally rests on the same poorly-correlated data you have presented me to make its case. An r-squared of 13% is not very convincing.

Again, what were you expecting?

Fractional reserve banking is also a semi-myth, in that it is misleading to focus on reserves and thinking banks create money leading to some kind of money multiplier.

Actually, nope. It's creating money at the cost of solvency, you can't simply increase your supply ceteris paribus without expecting a fall in the solvency rate.

You can have zero-reserve banking for example.

That is extremely risky, as I noted.

But even in FRB, imagine a commercial bank that loans out (to the economy, or to other banks) $100bn and receives $90bn in deposits (to be in accordance with a 10% reserve rate); if it wants to increase lending to $120bn without new deposits, it can take a loan from the interbank market or the central bank of $2bn, with this money being created out of thin air.

Only created out of thin air if the interbank loan was conceded by governmental institutions. Without a central bank that acts through discount lending your logic doesn't apply.

Panics are a problem if a bank's access to money from other banks runs dry, especially if their lending practices are heavily based on it (eg Northern Rock).

As I noted. FRB doesn't come without consequences.

I have no clue what you're trying to say here. I literally just said one person's deficit is another person's surplus, and you're telling me ... what, exactly, that deficits exist?

Wew, sorry on there.

Do you know what this being a mathematical identity means? It's true by definition. I can show it to you now.

GDP = Consumption (C) + Household Investment (H.I) + Corporate Investment (C.I) + Government expenditure (G) + Net exports (X - M)
GDP = Consumption (C) + Household Savings (H.S) + Corporate Savings (C.S) + Tax Income (T)
... therefore
C + H.I + C.I + G + (X - M) = C + H.S + C.S + T
--> (C.S - C.I) = (H.I - H.S) + (G - T) + (X - M)
--> Corporate Savings (profit) = C.I + (H.I - H.S) + (G - T) + (X - M)

It is true by definition. Although this is unnecessary, because it is based on identities, there is this link which shows the relationship, comparing what government accountants estimated to what must be true by definition. Unsurprisingly, accounting for statistical error, the numbers match exactly.

Interesting. Thanks.

[Taihei Tengoku]

(Image)
I don't think so.

for the hundredth time, productivity-wage scissors are entirely explained by nonwage benefits and differences in deflators

[Orostan]

(Image)
I don't think so.

for the hundredth time, productivity-wage scissors are entirely explained by nonwage benefits and differences in deflators

Actually I did some reasearch, and you're incorrect.

http://www.epi.org/blog/american-pay-an ... -together/

The Heritage "Foundation" measured average pay, not median pay. They messed up by doing this because of the sheer ammount of capital taken up by the rich offsets the average.

And the graph in this study counts benefits too.

http://www.epi.org/productivity-pay-gap/

[Orostan]

Where is this data coming from, and did you make this graph yourself?

https://www.measuringworth.com/
And indeed.

>what is time

A constant. :^)

It doesn't last forever. The emergence of Neoliberalism in the 70s-80s made capital accumulate faster for a while, until the late 90s.

Interesting. Some club sauce, if you may?

I don't think so.

Hourly wages kept flat until 2001 and then sprang, kept stable/growing slowly until 2013 when they sprang once again and picked up speed.

1- good job with graph, but Arkolon seems to have thourouhgly debunked it.

2- club sauce? You mean sources? I'm basing my claims on Michael Robert's graphs, such as these:
https://thenextrecession.wordpress.com/ ... he-latest/

3- Wages and hourly compensation for the normal worker still remain stagnant. This is because during the 70s outsourcing became really possible on a global scale. Now instead of paying an American 15 dollars an hour, capitalists can pay a pakistani .37 cents an hour.

[Taihei Tengoku]

for the hundredth time, productivity-wage scissors are entirely explained by nonwage benefits and differences in deflators

Actually I did some reasearch, and you're incorrect.

http://www.epi.org/blog/american-pay-an ... -together/

The Heritage "Foundation" measured average pay, not median pay. They messed up by doing this because of the sheer ammount of capital taken up by the rich offsets the average.

And the graph in this study counts benefits too.

http://www.epi.org/productivity-pay-gap/

Actually you Googled for about fifteen seconds, which is how long it takes to get to that EPI page. It's not just Heritage Foundation that says it's all NWB+deflators, it's also the London School of Economics.

[Arkolon]

You have totally missed the point, I gave an example to show how a trend-line means nothing if the degree of correlation is very weak. Your polynomial line of best fit is too poor to treat the U-shaped curve as even remotely meaningful, for the reasons I have already outlined.

Now let's be honest, what were you expecting? An increase in government leading to -20% growth perspectives?

This is literally what you've been trying to convince me is true, and this is literally what I've been telling you. Your data set is so poorly correlated, the trend-line so poorly fitting, that you can't even say with any respectable degree of confidence that a relatively bigger government affects private sector GDP growth in any meaningful way. You're the one who's been trying to tell me that the "U-shaped curve" you keep referring to matters at all, that a slight change in x reliably leads to a proportional change in y, but this isn't true at all, and you've now admitted it. There is no reliable effect you can point to.

glorification of Robert Barro's neoclassical economics only tangentially refers to the crowding out effect (it is mentioned in passing in one single sentence which isn't even the argument's crux, it totally assumes the existence of the effect)

And his hypothesis adds up to mine, that comproved with empirical data show a meaningful evidence of crowding out on the economy.

This is not what crowding out is. Crowding out is a purported banking sector phenomenon that has to do with 'loanable funds' being sacrificed to buy government debt rather than make private sector loans. Of course, because we know that there is no limit to loanable funds - banks create money out of thin air for whatever purpose, borrowing from the central bank if they need to – we know that 'loanable funds' aren't limited at all. What you gave me as 'evidence' just says 'we believe that government surpluses and GDP growth are (very very weakly) positively correlated', which is no different to the same conversation we've been having all along.

Fractional reserve banking is also a semi-myth, in that it is misleading to focus on reserves and thinking banks create money leading to some kind of money multiplier.

Actually, nope. It's creating money at the cost of solvency, you can't simply increase your supply ceteris paribus without expecting a fall in the solvency rate.

How does increasing lending – that is to say, assets – does a bank become less solvent? Sometimes creating money out of thin air is better than taking deposits: imagine banks offering a deposit interest rate to customers of 3% APR while funds are offered for 1% APR on the interbank lending market. It would be wiser to base your investments on cheap money rather than less cheap money, and so long as your investments are sound (you expect >1% return), the bank will make a profit. Ceteris paribus, increasing the amount of assets you own with a corresponding liability (either to depositors or other banks) doesn't make a bank less solvent.

You can have zero-reserve banking for example.

That is extremely risky, as I noted.

No it's not, it's no more risky than 1% FRB, or 2% FRB, because reserve ratios are not an important part of the monetary system. If a bank is lacking reserves, it can borrow money from the central bank to make up the shortfall. The traditional view is that banks are 'limited' to being able to create money depending on their reserves, but this clearly isn't the case.

But even in FRB, imagine a commercial bank that loans out (to the economy, or to other banks) $100bn and receives $90bn in deposits (to be in accordance with a 10% reserve rate); if it wants to increase lending to $120bn without new deposits, it can take a loan from the interbank market or the central bank of $2bn, with this money being created out of thin air.

Only created out of thin air if the interbank loan was conceded by governmental institutions. Without a central bank that acts through discount lending your logic doesn't apply.

Without a central bank the entire banking system would be different, so I'm not sure what your point is. Secondly, loans are always created out of thin air: there is no such thing as 'lending out deposits'. Go to a bank and ask for a loan, they'll give you money that didn't exist before; other customers' deposits are untouched. At the end of the day (every day) the bank loans out or borrows on the interbank market the difference between money created out of thin air and money received.

Do you know what this being a mathematical identity means? It's true by definition. I can show it to you now.

GDP = Consumption (C) + Household Investment (H.I) + Corporate Investment (C.I) + Government expenditure (G) + Net exports (X - M)
GDP = Consumption (C) + Household Savings (H.S) + Corporate Savings (C.S) + Tax Income (T)
... therefore
C + H.I + C.I + G + (X - M) = C + H.S + C.S + T
--> (C.S - C.I) = (H.I - H.S) + (G - T) + (X - M)
--> Corporate Savings (profit) = C.I + (H.I - H.S) + (G - T) + (X - M)

It is true by definition. Although this is unnecessary, because it is based on identities, there is this link which shows the relationship, comparing what government accountants estimated to what must be true by definition. Unsurprisingly, accounting for statistical error, the numbers match exactly.

Interesting. Thanks.

So now that you know that increasing the government deficit leads to greater private profits, and that trying to turn a deficit into a surplus will decrease private profits, and private profits are a principal source of capital investment which leads to greater employment and/or productivity, tell me again why you think that the government increasing its spending is bad for the economy?

[Great Minarchistan]

1- good job with graph, but Arkolon seems to have thourouhgly debunked it.

Hmmm, how come?

2- club sauce? You mean sources? I'm basing my claims on Michael Robert's graphs, such as these:
https://thenextrecession.wordpress.com/ ... he-latest/

My favourite measure is to take the annual net domestic product of any economy (that’s gross domestic product less depreciation) less employee compensation (wages and benefits paid by the employers) to get surplus value. Then I divide that by a measure of the cost of employing the labour force (employee compensation again) plus constant capital (which can be measured by the stock of fixed assets owned by the capitalist sector after allowing for depreciation). There are lots of other ways: just looking at the corporate sector, for example, before and after tax and so on. But my ‘whole economy’ measure is the simplest, takes into account all sectors in the economy, and is the easiest for comparisons between countries or in measuring a ‘world rate of profit’

How to calculate depreciation?

3- Wages and hourly compensation for the normal worker still remain stagnant. This is because during the 70s outsourcing became really possible on a global scale. Now instead of paying an American 15 dollars an hour, capitalists can pay a pakistani .37 cents an hour.

Cough.

[Orostan]

Actually I did some reasearch, and you're incorrect.

http://www.epi.org/blog/american-pay-an ... -together/

The Heritage "Foundation" measured average pay, not median pay. They messed up by doing this because of the sheer ammount of capital taken up by the rich offsets the average.

And the graph in this study counts benefits too.

http://www.epi.org/productivity-pay-gap/

Actually you Googled for about fifteen seconds, which is how long it takes to get to that EPI page. It's not just Heritage Foundation that says it's all NWB+deflators, it's also the London School of Economics.

>what is reading the article

1- good job with graph, but Arkolon seems to have thourouhgly debunked it.

Hmmm, how come?

You seem to be trying to convince him that a bigger government = less growth. He's been telling you that growth will seem to be less if government spending rises, and will seem to be more if it falls and that your 'correlation' on the graph is a consequence of these numbers and not any real kind of correlation.

2- club sauce? You mean sources? I'm basing my claims on Michael Robert's graphs, such as these:
https://thenextrecession.wordpress.com/ ... he-latest/

My favourite measure is to take the annual net domestic product of any economy (that’s gross domestic product less depreciation) less employee compensation (wages and benefits paid by the employers) to get surplus value. Then I divide that by a measure of the cost of employing the labour force (employee compensation again) plus constant capital (which can be measured by the stock of fixed assets owned by the capitalist sector after allowing for depreciation). There are lots of other ways: just looking at the corporate sector, for example, before and after tax and so on. But my ‘whole economy’ measure is the simplest, takes into account all sectors in the economy, and is the easiest for comparisons between countries or in measuring a ‘world rate of profit’

How to calculate depreciation?

You can get the rate of depreciation by looking at average resale price of machines, average original price, and average lifespan. Or you can use medians, but personally for this I think averages would be better for this situation.

3- Wages and hourly compensation for the normal worker still remain stagnant. This is because during the 70s outsourcing became really possible on a global scale. Now instead of paying an American 15 dollars an hour, capitalists can pay a pakistani .37 cents an hour.

Cough.

What does this graph mean? Who's it for? I can tell what the x axis is but what is the y axis? I need more information than just a graph with no notes attached.

[Great Minarchistan]

You seem to be trying to convince him that a bigger government = less growth. He's been telling you that growth will seem to be less if government spending rises, and will seem to be more if it falls and that your 'correlation' on the graph is a consequence of these numbers and not any real kind of correlation.

He would be right if it was a linear correlation. It isn't though. For instance, growth grows along with government spending until 20%.

You can get the rate of depreciation by looking at average resale price of machines, average original price, and average lifespan. Or you can use medians, but personally for this I think averages would be better for this situation.

So basically accounting for CPI of machines?

What does this graph mean?

I means that wages haven't been stagnated. :-)

Who's it for?

You, your statement quoted above it to be more exact.

I can tell what the x axis is but what is the y axis?

It's indexed, meaning both started at 100 in order to see the evolution.

[Arkolon]

You seem to be trying to convince him that a bigger government = less growth. He's been telling you that growth will seem to be less if government spending rises, and will seem to be more if it falls and that your 'correlation' on the graph is a consequence of these numbers and not any real kind of correlation.

He would be right if it was a linear correlation. It isn't though. For instance, growth grows along with government spending until 20%.

It isn't a linear correlation and it isn't a polynomial correlation, it is very-weak-to-no correlation. The coefficient of determination is so low that the trend-line means absolutely nothing, since it links the data points with a pitiful degree of confidence. There are way, way too many factors at play in determining GDP growth for the size of government to lead to any correlation at all.

You picked a polynomial trend line, and haven't chosen to reveal to us its r-squared. Why didn't you pick a linear trend line? It would be flat, so wouldn't support your argument at all, but I'm confident it has a greater r-squared than the trend-line you put up.

[Taihei Tengoku]

Actually you Googled for about fifteen seconds, which is how long it takes to get to that EPI page. It's not just Heritage Foundation that says it's all NWB+deflators, it's also the London School of Economics.

>what is reading the article

EPI's thesis is that the scissor is partially caused by the decoupling of productivity from wage; this would be big if true. Turns out it is not, it's a statistical artifact from using different deflators. The wage inequality is fundamentally a productivity inequality: a lot of workers are stuck in more or less solved jobs where there is little expected improvement.

The LSE paper directly challenges the notion that there is, in fact, such a net decoupling. In the UK it does not exist and in the US it is overshadowed by explainable differences between compensations and deflators by a factor of five. You read the article.

[Great Minarchistan]

It isn't a linear correlation and it isn't a polynomial correlation

it is very-weak-to-no correlation.

For the 100th time, you can't simply expect a -20%, do you even understand how growth works?

The coefficient of determination is so low that the trend-line means absolutely nothing, since it links the data points with a pitiful degree of confidence.

Yeah, trend lines are fake news.

There are way, way too many factors at play in determining GDP growth for the size of government to lead to any correlation at all.

Interesting though, it's almost like OECD studies corroborate my thesis.

You picked a polynomial trend line, and haven't chosen to reveal to us its r-squared.

I don't think you can obtain a R² out of a polynomial trend.

Why didn't you pick a linear trend line? It would be flat, so wouldn't support your argument at all

It's actually slight descending. I put a polynomial line since I prefer to see peaks of performance rather than a linear trend.

[Arkolon]

First of all, you have this to reply to.

It isn't a linear correlation and it isn't a polynomial correlation

it is very-weak-to-no correlation.

For the 100th time, you can't simply expect a -20%, do you even understand how growth works?

The coefficient of determination is so low that the trend-line means absolutely nothing, since it links the data points with a pitiful degree of confidence.

Yeah, trend lines are fake news.

There are way, way too many factors at play in determining GDP growth for the size of government to lead to any correlation at all.

Interesting though, it's almost like OECD studies corroborate my thesis.

You picked a polynomial trend line, and haven't chosen to reveal to us its r-squared.

I don't think you can obtain a R² out of a polynomial trend.

Yes, you can obtain an r-squared from a polynomial trend, it is polynomial regression, and I know Google Drive calculates it automatically. If a trend line has such a dismal degree of correlation, yes it is fake news. Just because Excel drew a line on your graph doesn't mean that that trend is in any way useful. It just means Excel found the line of best fit at whatever degree you chose – it does NOT mean there is a meaningful correlation. I've already told you this before: take a data set with NO correlation, for example (0,-1), (0,1), (1,-1) and (1,1), and find the trend line between these data sets. It will give you a trend line of y=0, but this is not a correlation (you already know there is NO correlation in this data set), because the trend line has such a poor degree of correlation.

Why didn't you pick a linear trend line? It would be flat, so wouldn't support your argument at all

It's actually slight descending. I put a polynomial line since I prefer to see peaks of performance rather than a linear trend.

I have also already explained to you why this would be the case. If the private sector is in severe contraction, the size of the government relative to total GDP will rise so long as the change in government expenditure is smaller than the change in total GDP. This explains why there might be a very slight downward trend. High private sector GDP growth leads to government expenditure being smaller relative to total GDP, ceteris paribus, and the relationship is not the other way around. I have already explained this to you. You're also demonstrating great incompetence at statistics by saying you 'prefer polynomial lines', paying no attention to the coefficient of determination in choosing your line of best fit. I'm getting tired of having to go through this conversation explaining to you the same thing over and over again when you clearly don't know much about statistics beyond 'the computer made a line and I like this line so this line is correct'.

[Orostan]

>what is reading the article

EPI's thesis is that the scissor is partially caused by the decoupling of productivity from wage; this would be big if true. Turns out it is not, it's a statistical artifact from using different deflators. The wage inequality is fundamentally a productivity inequality: a lot of workers are stuck in more or less solved jobs where there is little expected improvement.

The LSE paper directly challenges the notion that there is, in fact, such a net decoupling. In the UK it does not exist and in the US it is overshadowed by explainable differences between compensations and deflators by a factor of five. You read the article.

http://oecdinsights.org/2012/02/20/do-w ... ty-growth/

How do you explain the part of the EPI graph where wages and productivity are close to each other?

[Arkolon]

EPI's thesis is that the scissor is partially caused by the decoupling of productivity from wage; this would be big if true. Turns out it is not, it's a statistical artifact from using different deflators. The wage inequality is fundamentally a productivity inequality: a lot of workers are stuck in more or less solved jobs where there is little expected improvement.

The LSE paper directly challenges the notion that there is, in fact, such a net decoupling. In the UK it does not exist and in the US it is overshadowed by explainable differences between compensations and deflators by a factor of five. You read the article.

http://oecdinsights.org/2012/02/20/do-w ... ty-growth/

How do you explain the part of the EPI graph where wages and productivity are close to each other?

Although I haven't been paying much attention to this conversation, Frances Coppola has written a very interesting post on the matter of wages and productivity never having really followed each other.

[Coconut Isle]

No problem:
(Image)

Hi, doesn't that curve imply that 20% government size is the optimal government size for "private GDP" growth? Why the choice to use private GDP as opposed to actual GDP? Is the regression overall significant?

[The Liberated Territories]

No problem:
(Image)

Hi, doesn't that curve imply that 20% government size is the optimal government size for "private GDP" growth? Why the choice to use private GDP as opposed to actual GDP? Is the regression overall significant?

"Actual GDP" is often misleading. A government could probably raise it's GDP artificially very quickly by say, putting all government spending towards creating new ships who will never set sail. Nonetheless it'll show that GDP and jobs grew.

This is why the Soviet Union's GDP in particular is misleading, much of the GDP growth shown is hot air by people working on useless government projects while its average citizen starved to death because that market wasn't liberalized. Even though the SU's GDP was half of the US's in 1990, no person would think that if you went to a Soviet supermarket then and compared it with a US supermarket.

[Arkolon]

Hi, doesn't that curve imply that 20% government size is the optimal government size for "private GDP" growth? Why the choice to use private GDP as opposed to actual GDP? Is the regression overall significant?

"Actual GDP" is often misleading. A government could probably raise it's GDP artificially very quickly by say, putting all government spending towards creating new ships who will never set sail. Nonetheless it'll show that GDP and jobs grew.

This is why the Soviet Union's GDP in particular is misleading, much of the GDP growth shown is hot air by people working on useless government projects while its average citizen starved to death because that market wasn't liberalized. Even though the SU's GDP was half of the US's in 1990, no person would think that if you went to a Soviet supermarket then and compared it with a US supermarket.

"Private GDP" is often misleading. The private sector could probably raise its GDP artificially very quickly by say, making many speculative investments in housing or producing lots of yachts and mansions for its CEOs while its average citizen starved to death due to increasingly poor wages, rising rent prices and increasingly dismal living conditions. Nonetheless it'll show that GDP and jobs grew.

And they weren't useless government projects in the Soviet Union, they were huge industrialisation efforts to produce more coal, steel and other crucial construction materials for military purposes, but obviously this was at the expense of consumption goods (cf. yachts and mansions). There is no doubt though that a Russian person in 1990 was a lot richer and 'had a better supermarket' than a Russian person in 1917, before all the industrialisation efforts.

[Forsher]

Yes, you can obtain an r-squared from a polynomial trend, it is polynomial regression, and I know Google Drive calculates it automatically. If a trend line has such a dismal degree of correlation, yes it is fake news. Just because Excel drew a line on your graph doesn't mean that that trend is in any way useful. It just means Excel found the line of best fit at whatever degree you chose – it does NOT mean there is a meaningful correlation. I've already told you this before: take a data set with NO correlation, for example (0,-1), (0,1), (1,-1) and (1,1), and find the trend line between these data sets. It will give you a trend line of y=0, but this is not a correlation (you already know there is NO correlation in this data set), because the trend line has such a poor degree of correlation.

A slight correction... it's still just a linear regression, the difference is that it now incorporates a polynomial term. In terms of the underlying regressions, y = b₁x + b₂x² + e would be best practice, but I worry whatever program he's using would just be using y = b₃x² + u as I don't know what exactly it is that they do.

Anyway, from what I have been skimming through, it sounds like Arkolon's saying that if we want to get a more meaningful graph we'd compare the growth rates of the absolute figures for public and private sector GDP (presumably in real terms).

If GM is interested, I suggest downloading the free statistical language R (and also R Studio because I think that's an easier environment to work in) and saving his data set as a .csv file. Assuming that the data set consists of two columns with row 1 using headers of "capita" and "govt" (without quotes) saved in NSGData.csv you could then use the following code:

data.df = read.csv("NSGData.csv", header=TRUE)
reglin.lm = lm(capita~govt,data=data.df)
summary(reglin.lm)
regpoly.lm = lm(capita~govt+I(govt^2),data=data.df)
summary(regpoly.lm)
interact.lm = lm(capita~govt*I(govt^2),data=data.df)
summary(interact.lm)
#names are arbitrary, can be changed if you don't like them
#chucked in interaction as well just for kicks

We're interested in the Adjusted R² figure, but also whether or not the coefficients are significant (indicated with asterisks) using cut offs (i.e. p-value = 0.049 and p-value = 0.051 tell us different things despite being hardly different). For real results there are some checks we should run but this will do for our purposes I think? Ah, what the hell... download package s20x, and run for each .lm model eovcheck(model.lm), normcheck(model.lm) and cooks20x(model.lm). We want the first to look boring (patternless... anything else is a BIG problem), the second to match up (that'll make sense when you see it) and the last? Well, we don't really want any much taller spikes and as a rule of thumb 0.4 is not where we want to be reaching.

[Orostan]

Hi, doesn't that curve imply that 20% government size is the optimal government size for "private GDP" growth? Why the choice to use private GDP as opposed to actual GDP? Is the regression overall significant?

"Actual GDP" is often misleading. A government could probably raise it's GDP artificially very quickly by say, putting all government spending towards creating new ships who will never set sail. Nonetheless it'll show that GDP and jobs grew.

This is why the Soviet Union's GDP in particular is misleading, much of the GDP growth shown is hot air by people working on useless government projects while its average citizen starved to death because that market wasn't liberalized. Even though the SU's GDP was half of the US's in 1990, no person would think that if you went to a Soviet supermarket then and compared it with a US supermarket.

I don't think the USSR had famines after WWII. And in addition "the market not being liberalized" is an extremely simplistic explanation that ignores the real reasons why there food shortages in the USSR. One of which was their method of economic planning, which only allocated as much resources as was minimally neccesary to doing things. Another was American sanctions on the USSR that forced the cost of food up.

[Taihei Tengoku]

"Actual GDP" is often misleading. A government could probably raise it's GDP artificially very quickly by say, putting all government spending towards creating new ships who will never set sail. Nonetheless it'll show that GDP and jobs grew.

This is why the Soviet Union's GDP in particular is misleading, much of the GDP growth shown is hot air by people working on useless government projects while its average citizen starved to death because that market wasn't liberalized. Even though the SU's GDP was half of the US's in 1990, no person would think that if you went to a Soviet supermarket then and compared it with a US supermarket.

I don't think the USSR had famines after WWII. And in addition "the market not being liberalized" is an extremely simplistic explanation that ignores the real reasons why there food shortages in the USSR. One of which was their method of economic planning, which only allocated as much resources as was minimally neccesary to doing things. Another was American sanctions on the USSR that forced the cost of food up.

The last famine was in 1947, and it's not like Russia needed US imports: under the Tsars it accounted for over a third of world wheat exports.

"Actual GDP" is often misleading. A government could probably raise it's GDP artificially very quickly by say, putting all government spending towards creating new ships who will never set sail. Nonetheless it'll show that GDP and jobs grew.

This is why the Soviet Union's GDP in particular is misleading, much of the GDP growth shown is hot air by people working on useless government projects while its average citizen starved to death because that market wasn't liberalized. Even though the SU's GDP was half of the US's in 1990, no person would think that if you went to a Soviet supermarket then and compared it with a US supermarket.

"Private GDP" is often misleading. The private sector could probably raise its GDP artificially very quickly by say, making many speculative investments in housing or producing lots of yachts and mansions for its CEOs while its average citizen starved to death due to increasingly poor wages, rising rent prices and increasingly dismal living conditions. Nonetheless it'll show that GDP and jobs grew.

And they weren't useless government projects in the Soviet Union, they were huge industrialisation efforts to produce more coal, steel and other crucial construction materials for military purposes, but obviously this was at the expense of consumption goods (cf. yachts and mansions). There is no doubt though that a Russian person in 1990 was a lot richer and 'had a better supermarket' than a Russian person in 1917, before all the industrialisation efforts.

The five year plans did no better than the extrapolation of lower bound Tsarist trends. The five year plan was tractors at the expense of people and food--the Stalinist policy deliberately suppressed food prices in the city to get people off the farm, forcing them to make machines to make up for lost labor. Unfortunately that attempt to create an economic perpetual motion machine led to the Holodomor.

Housing investments and yachts account for peanuts. When home prices collapsed US GDP shrank...5%. Famine conditions would be a much larger contraction if the median citizen was starving.

[Taihei Tengoku]

EPI's thesis is that the scissor is partially caused by the decoupling of productivity from wage; this would be big if true. Turns out it is not, it's a statistical artifact from using different deflators. The wage inequality is fundamentally a productivity inequality: a lot of workers are stuck in more or less solved jobs where there is little expected improvement.

The LSE paper directly challenges the notion that there is, in fact, such a net decoupling. In the UK it does not exist and in the US it is overshadowed by explainable differences between compensations and deflators by a factor of five. You read the article.

http://oecdinsights.org/2012/02/20/do-w ... ty-growth/

How do you explain the part of the EPI graph where wages and productivity are close to each other?

Simply put employers hadn't found that paying people in health insurance was better than paying them in actual money. The biggest part of the wedge is pay vs comp, with the accumulated overestimation of inflation that RPI produces filling the rest in the UK. The article admits that once you apply those corrections the wedge disappears.

The "lower half" is stagnating because there isn't a great productivity revolution in the jobs that the lower half works--your wage still tracks your productivity and what determines how productive you are is the job you perform. The problem is that a lot of jobs are solved problems where it is harder for rank-and-file labor to enjoy productivity improvements (unlike a pre-automated factory).

[Chestaan]

[citation needed]

No problem:

Also, of course Capital accumulates faster without the government regulating things.

Thanks for corroborating my correlation with a causation thesis. :^)

My issue with "prosperous" times is that they often are not prosperous for the vast majority of people.

Adjusted GDP per capita is heavily related with employment figures in the US. Nonetheless to say, it is also related with an income increase. Under most of the periods when rGDP/capita grew, blue collar wages followed and even outstripped it.

Did you calculate the p-value for this? If so, what were they? And what other factors did you include?