Keynesianism is an ass

[Augustine2004]

Let's suppose you are correct. Then Keynesian ideas became mainstream because mainstream economists also desired justification for their previous love of big government?

But then why does the mainstream still hold onto Keynes' framework? Even Milton Friedman, who was for the most part a libertarian and apologist for small government, bought into this basic framework and made a case for small government from within the framework instead of refuting the framework itself. (For this reason, Friedman still couldn't reject monetary policy.) From what I understand, Friedman and some neoclasical "mainstream" economists argue for smaller government, but still hold onto ideas such as a consumption function and the multiplier, an aggregate demand function, liquidity preference theory of interest, the IS-LM model. But economists like Friedman have no reason such as you suggest for buying into this framework. So why do they? (After reading General Theory, I read this http://legacy.ncsu.edu/classes/ec348001/TextLinks.htm which I think gave me a better understanding of what "mainstream" economists think today. On the other hand most of the derivation of the models in this book consists of unsupported assertions, thus leaving me in the dark as to why they believe these things.)

Do you know of any debates between 'Austrian' and 'mainstream' economists (say, over something like the IS-LM model), that would help illuminate their point of departure? Or it would also be helpful if I had one or more 'mainstream' economists with whom I could dilalogue.

Governments prefer Keynesists or economists that support big government. Austrian School economists have had to struggle against the government. Incidentally, did you hear about the TSA detaining a Ron Paul aide? Either that or they go over to the dark--and materially comfortable and safe--side.

Certainly Austrian School people have criticized other economists such as Keynes and Paul Krugman. I think there were debates, but I can't recall one offhand.

[joel]

It may be useful to also discuss modern ideas derived from Keynesianism. I thought I could share some thoughts I had regarding what I learned about modern 'mainstream' macroeconomics from the McElroy text (that I linked to in my last post). But I don't want to be guilty of back-to-back posting, so if you are interested, you need to contribute with some discussion.

In this post, I'd like to discuss the modern interpretation of the "consumption function" and the multiplier. The analysis focuses entirely on the spending side of things (as opposed to production). Every dollar spent is received by someone else as income. Thus if you add up all the items of spending, it must equal total income Y. As before, if we divide all spending into consumption C and investment I, then

Y = C + I

Okay so far. Now what the 'mainstream' economists do is try to express C and I as mathematical functions of other quantities. They think of consumption as a function C(Y) of income, and investment as a function I(r) of the interest rate. Now there are various reasons to believe that C and I are not functions of these quantities. And conceiving of them in this way makes it seem as though they were independent, as if one could be increased independently of the other, as if there were not a tradeoff between them, such that you would have to reduce your spending on investment if you wanted to increase your immediate consumption. Y in the expression above is the income generated by the spending, but we seem to be ignoring the question of where people got the funds to spend in the first place, and that those funds are limited.

But even if we suppose that there exists such independent functions, we run into other problems. As a simplification they treat the function C(Y) as being (at least locally) linear

C(Y) = C0 + C1*Y

Thus we have

Y = C + I = (C0 + C1*Y) + I(r)

This brings us to the idea of the multiplier. Although Keynes seemed to think of the multiplier as something that is always mathematically true at every instant of time, it seems that the modern way to think of it is as an iterative phenomenon. Notice that there are two Y's in the above formula. The one on the left is the income generated by (and equal to) the spending on the right. The C1*Y implies, on the other hand, that a certain percentage of income previously received is consumed. It may be easier if we think of dividing up time into periods, where income Y2 (at time/period 2) is generated by spending out of income received in the previous time period Y1 (and perhaps out of cash saved--not spent--in even earlier time periods). Thus something like:

Y(t+1) = C0 + C1*Y(t) + I(r)

Where an equilibrium or 'steady state' (or 'evenly rotating economy') would require an unchanging income from period to period: Y(t+1) = Y(t).
For illustration, consider a numerical example. For simplicity, suppose C0=0, C1=0.9, I(r)=100, and thus

Y = 0.9 * 1000 + 100 = 1000

This describes the a steady state where people as a whole consume 90% of their income and invest 10%. Now let's suppose that the C and I above were only private spending and there wasn't any government spending. And then suppose the government comes along and taxes and spends 100. Now we are going to have:

Y2 = C(Y1 - T) + I(r) + G

where T is the tax and equals G, the government spending. Thus

Y2 = 0.9 * (1000 - 100) + 100 + 100 = 1010

Woah, starting from a steady state, the government taxed 100 and spent 100 and total income increased by 10! This is completely surprising and counterintuitive, because one would think that the tax must reduce private spending power by exactly 100. (What's more, income would have increased by even more if T=0, and the government still spends the same 100, running a deficit. In fact income, according to this model, would have increased by the full amount of the government's spending!) Instead of questioning this absurd result, the 'mainstream' economist considers this the "expansionary" power of government spending. McElroy, at least, does not stop to consider or explain where this extra 10 comes from. Let's see. In the steady state, Y was 1000, and C1 was 90% of that, so consumption was 900. But in period 2, consumption was decreased due to the tax:

C = 0.9 * (1000 - 100) = 810

Thus consumption decreased by only 90, less than the amount of the tax! But private spending power was reduced by 100. Something had to be reduced by the extra 10. McElroy never explains this. If reduction of consumption fell by only 90 and we assume that investment did not decrease, then if it didn't come into being by magic (or the government's printing presses), it could have come only from cash balances that people had saved up prior to the steady state and were not spending in the steady state. If people were spending all of their money every time period--which is possible--then the above result is outright contradictory, and thus the above formula has an obvious flaw. But why would people begin to consume their cash balances just because of the government's tax and spending. By what reason can we suppose this is necessarily true. Before considering this closer, let's get back to the multiplier. Y2 according to the above model was larger (1010). Now plug this in for the next time period, assuming the government continues its behavior of taxing and spending 100 each period:

Y3 = 0.9 * (1010 - 100) + 100 + 100 = 1019

Income got even bigger (but by a smaller increment). But then

Y4 = 0.9 * (1019 - 100) + 100 + 100 = 1027.10

And total income keeps getting bigger and bigger. According to this model, it will continue growing until Y(t+1) = Y(t). We can solve for this algebraically.

Y = C1*(Y-T) + I + G
solving for Y gives
Y = (1/(1-C1)) * (I + G - C1*T)

Thus (1/(1-C1)) is the multiplier. In this case, 1/(1 - 0.9) = 10. And
Y = 10 * (100 + 100 + 0.9*100) = 1100, thus we supposedly reach a new steady state where the income is 1100 each period. This is the magic of the multiplier. Algebraically, this is actually the same as Keynes' multiplier (as I described in an earlier post), because C1 is really ΔC/ΔY. It is merely interpereted differently here, as an iterative, instead of instantaneous, phenomenon.

But still no reason is given to believe that people necessarily behave in this way. Also it has all the same failings as Keynes'. For example, C1 could be 1 (that is, a change in income could coincide with a change in consumption by the same amount). But as C1 approaches 1, Y goes to infinity! Something is wrong here.

Also consider that Y, being total spending, must be equal to the sum of all the prices paid for everything purchased:

Y = p1 + p2 + p3 + ...

The only way Y could increase is for either more goods to be produced and purchaced or for the prices of the purchases to increase, or both. But government taxing and spending does not magically produce more goods to be purchased--that takes production, which takes time and scarce resources. Thus if it does cause an increase in Y (which there is reason to suspect), it can happen only by increasing purchase prices overall. But this is the same thing as a fall in the purchasing power of money (i.e., inflation). So this means that the increase in Y is offset by a fall in the value of that money, and thus the real value of Y is unchanged or even reduced.

Furthermore, if the purchasing power of money falls but the money supply remains the same (as we are assuming here), this means that the real value of the total cash holdings of everyone is shrinking. But this can happen only if the demand for money (i.e., the demand for cash holdings) has fallen. If it hasn't then there will be a shortage of money, in which case people will reduce their purchases and increase their sales in order to try to gain more money. This behavior will tend to drive prices downward, counteracting the above supposed increase in Y. There is no reason to believe that additional government taxing and spending will decrease the demand for money, let alone decreasing it simultaneously and by the exact amount predicted by this mathematical model. In fact, the uncertainty generated by the government increasing its taxing, spending, etc., may be likely to cause people to want to increase rather than decrease their cash balances.

In order to get around this last problem, the 'mainstream' economist proposes a brand new theory of interest, throwing out all the advancements in the study of interest over the past 200 years. This new theory of interest is severely flawed. Though I should leave that for another post, assuming anyone is interested in continuing the discussion.

[Augustine2004]

In a free market, a shortage of money would never develop. If perchance the people saved more money than usual, or somehow the gold mines output less gold so that grams of gold per capita decreased, the economy can easily adjust the prices so that each gram of gold would have greater purchasing power, i.e., it can buy more goods or services than before.

What could be meant by consumption being a linear function of total income? That can be true only if it is good every time. It's got to be at least partially a function of time. People used to not save money--well, not much. Now, suddenly they've got saving religion! Hallelah, a saved dollar saves!

The choices that people and the actions that they take, as I've said more than once, are not predictable. Hence, the function of time must not be knowable.

Modern theory, phooey. Keynes, phooey.

[joel]

In a free market, a shortage of money would never develop. If perchance the people saved more money than usual, or somehow the gold mines output less gold so that grams of gold per capita decreased, the economy can easily adjust the prices so that each gram of gold would have greater purchasing power, i.e., it can buy more goods or services than before.

Yes, although couldn't we also say that actually there is always a small shortage or surplus of money because we are in a constantly changing economy? Certainly at final prices and in the evenly rotating economy there is no surplus or shortage of money, but we are never quite there, always moving toward a moving target. This is what makes entrepreneurial profit and loss possible. In mainstream terminology this would be called the "short run." Suppose we start out with no shortage or surplus, but then the demand for money quickly increases (or decreases). Prices don't adjust everywhere instantly or at the same time (or to the same degree). Mises called this a cash-induced change to the money relation. Redistribution effects will occur in the process of prices across the economy adjusting. In this "short term", before prices have finished adjusting, people will have smaller cash balances than they want, no? But any shortage or surplus is always tending toward zero.

What could be meant by consumption being a linear function of total income? That can be true only if it is good every time. It's got to be at least partially a function of time.

If you suppose that consumption C(Y) is a differentiable function of Y, then you can treat it as being locally linear via a first-order Taylor expansion (http://en.wikipedia.org/wiki/Taylor_series). For example, suppose Y =1000, C=900. Then we would say that C(1000) = 900. Suppose also that we can know the slope C'(1000) of the function C at Y=1000. Let's say the slope is 0.8. Then C(Y) is approximately

C(Y) = 1000 + 0.8 * (Y-1000)
or
C(Y) = 200 + 0.8 * Y, (Thus C0=200, and C1=0.8)

at least in the neighborhood around Y=1000. Of course the problem is that we don't know that there is a functional relationship between C and Y at any point in time, let alone that it is continuous and differentiable. Even if there were such a function, we couldn't measure C0 and C1 because they would constantly be changing. We would have no way of knowing whether a change over time was a movement along the function or if the function itself changed or both. So yes, like you say, you would need to know it as a function of time too:
C(Y, t). But then what does the function even mean when you vary Y but hold t constant? Are we then speaking of counterfactuals?--What would consumption be (have been) at time t if Y were other that it actually is (was) in the real world. One can argue that there is no such thing as a true counterfactual--especially when talking about beings with free will.

The choices that people and the actions that they take, as I've said more than once, are not predictable. Hence, the function of time must not be knowable.

True. The amount people consume depends on changing personal preferences and fact of what actually gets produced in what proportions and offered at what prices. There is never a single, definite, level of consumption determined by a given level of income.

So much for the consumption function and the multiplier. Perhaps the next thing to discuss is investment and interest. What do you think about the 'mainstream' idea that the total amount of investment I is a function I(r) of the rate of interest? For example, "Given expectations about returns on fixed investment, every level of interest rate [r] will generate a certain level of planned fixed investment and other interest-sensitive spending" (http://en.wikipedia.org/wiki/Islm If one looks at that page, one will see that the
Y=C(Y-T) + I(r) + G that we have been considering defines the IS curve in the IS-LM model).

[Augustine2004]

I think that given enough data points that are reliable and cover a given t,Y area reasonably well, we can fit a nonlinear C(Y,t) to the data. No need for a Taylor expansion. The problem is to get the reliable data timely. In practice, the data is collected in weeks or even months. Even at that, the data may not be reliable anyway.

But, suppose we do have a C(Y, t) to play with anyway. What general conclusions be made from that? We really can’t assume it is really continuous and differentiable. There may be steep slopes that are effectively discontinuous, but that the data, as discrete as it is, does not show. We certainly can’t assume that the function applies to all time intervals anyway.

As for the relationship between the total amount of investments and the interest rate(s), there is no such thing as the interest rate. Every time the mix of different interest rates may differ from any other time. One could create an index, but it depends on how the index is created. It would have to be consistently applied to the data. How can that be, when the mix of interest-bearing instruments may vary from time to time. Even so, the amount-rate relationship may vary from time to time.

[joel]

As for the relationship between the total amount of investments and the interest rate(s), there is no such thing as the interest rate.

I think that seals its fate right there. And I think there are other problems besides. For the sake of argument, let's suppose that there were just "the interest rate." Take that quote:

"Given expectations about returns on fixed investment, every level of interest rate [r] will generate a certain level of planned fixed investment and other interest-sensitive spending"

They are taking returns on fixed investment as given, and varying the interest rate on business loans, as if those two things were independent of each other. In reality, however, those two things tend toward each other. If returns on capital goods are expected to be greater than the interest rate, then people will refrain from lending and directly invest in order to capture the greater expected return. This will drive the interest rate up and the return on those capital goods down (because it will drive the price of them up and the price of their product down). Likewise if the expected return on capital goods falls below that of the interest rate on loans, then people will choose to lend their money at interest instead of investing directly in capital goods. This will drive the interest rate down and the return on the capital goods up, until expected returns are equal. So the premise of the quoted statement is erroneous.

The assumption of the IS curve and the multiplier is that consumption will iteratively adjust until savings and investment balance at whatever the interest rate may be. But how can this be? The formula Y2=C(Y1-T) + I(r) + G does not include savings anywhere--at least not in the sense that they mean it. Savings is defined as Y1-T-C. But this doesn't appear anywhere. There is no mention in the formula of any tradeoff between consumption and saving. In fact, we might point out that lending money is an alternative to consumption spending. The higher the interest rate, the higher the opportunity cost of consuming vs lending (it also makes it more expensive to borrow to consume). But the 'consumption function' does not take this into account. Perhaps it should be C(Y, t, r). And likewise investment (here restricted to mean only the purchase of capital goods) surely depends on the funds available to invest, which are limited by things like past income and consumption. But no mention of this is made in the function I(r) which varies only with the rate of interest.

Specifically, the 'mainstream' economist says that the relationship between the quantity of investment and the interest rate on loans is an inverse one. If the interest rate increases, then the volume of investment will decrease.
But is this really true? Note that if we take into account an inverse relationship also between consumption and the interest rate (as described above), this would seem to further support the idea that a high interest rate is bad because it drives down both consuming and investing, thus driving down total income Y! But if the rate of interest increases, this must also mean that the expected return on capital goods is tending to increase too, because the two are always tending toward each other, as explained above. Thus we should expect an increase in motivation to invest in capital goods.

The 'mainstream' economist also ignores time preference here. Both the expected return from capital goods and the interest rate on loans are part of the larger category of the 'time market'--the exchange between present goods and future goods. If the demand for present goods (in exchange for future goods) increases while the supply schedule remains the same, then the rate of return (on both 'investment' and loans) will increase, and the volume of lending/investing will also increase. Likewise if the government were to impose a maximum rate of return below that which would occur on the unhampered market, then investment would decrease, and there would be a shortage of investment. Thus it is not necessarily the case that there is an inverse relationship between investing and the rate of interest.

But if this relationship does not necessarily hold, then their whole theory breaks down. The IS-LM model is a theory of the interest rate. It says, for example, that if government spending increases, then the IS curve shifts to the right. Consumption drops by less than the amount of the tax (if any) and this is compensated by a rise in the interest rate causing investment to decrease by just the amount needed to supposedly compensate for the given demand for money, based on a liquidity-preference theory of interest. And so we see that according to this model, interest must be determined exactly by demand. The interest rate needs to adjust to the point that balances demand for consumption and the demand for money and the demand for investment (and government 'demand'). This is supposedly the point where the IS and LM curves cross. But this completely leaves out the obvious cause for the rate of interest: the supply and demand for present goods (as exchanged against future goods). It ignores a couple hundred years of advancement on the understanding of interest.

And ather thing that doesn't make sense to me: Before the increase of government spending, we assume we are in a state of equilibrium. In particular, the supply of funds for loans must equal the demand for loans. Then the interest rate increases in order to equilibrate demand. But the higher rate of interest makes the supply of loans increase (as indicated by the LM curve)--people become more eager to loan money--and the demand for loans to decrease--people become less eager to borrow. But surely this means that there is now a surplus of lendable funds! People want to lend more than people are willing to borrow. We cannot possibly be at a new equilibrium.

[Augustine2004]

I think that seals its fate right there. And I think there are other problems besides. For the sake of argument, let's suppose that there were just "the interest rate." Take that quote:

"Given expectations about returns on fixed investment, every level of interest rate [r] will generate a certain level of planned fixed investment and other interest-sensitive spending"

They are taking returns on fixed investment as given, and varying the interest rate on business loans, as if those two things were independent of each other. In reality, however, those two things tend toward each other. If returns on capital goods are expected to be greater than the interest rate, then people will refrain from lending and directly invest in order to capture the greater expected return. This will drive the interest rate up and the return on those capital goods down (because it will drive the price of them up and the price of their product down). Likewise if the expected return on capital goods falls below that of the interest rate on loans, then people will choose to lend their money at interest instead of investing directly in capital goods. This will drive the interest rate down and the return on the capital goods up, until expected returns are equal. So the premise of the quoted statement is erroneous.

Not only that, but government 'investment' is like a joker. Who could forecast its amount and where it would be invested?

Specifically, the 'mainstream' economist says that the relationship between the quantity of investment and the interest rate on loans is an inverse one. If the interest rate increases, then the volume of investment will decrease.
But is this really true? Note that if we take into account an inverse relationship also between consumption and the interest rate (as described above), this would seem to further support the idea that a high interest rate is bad because it drives down both consuming and investing, thus driving down total income Y! But if the rate of interest increases, this must also mean that the expected return on capital goods is tending to increase too, because the two are always tending toward each other, as explained above. Thus we should expect an increase in motivation to invest in capital goods.

I recall reading somewhere that as the economy expands, the interest rates tend to rise. However, the volume of loans and other investments may expand also, because business seems to be so good.

The 'mainstream' economist also ignores time preference here. Both the expected return from capital goods and the interest rate on loans are part of the larger category of the 'time market'--the exchange between present goods and future goods. If the demand for present goods (in exchange for future goods) increases while the supply schedule remains the same, then the rate of return (on both 'investment' and loans) will increase, and the volume of lending/investing will also increase. Likewise if the government were to impose a maximum rate of return below that which would occur on the unhampered market, then investment would decrease, and there would be a shortage of investment. Thus it is not necessarily the case that there is an inverse relationship between investing and the rate of interest.

It may be better to put what you said about the volume first, not last, in the quoted passage above.

But if this relationship does not necessarily hold, then their whole theory breaks down. The IS-LM model is a theory of the interest rate. It says, for example, that if government spending increases, then the IS curve shifts to the right. Consumption drops by less than the amount of the tax (if any) and this is compensated by a rise in the interest rate causing investment to decrease by just the amount needed to supposedly compensate for the given demand for money, based on a liquidity-preference theory of interest. And so we see that according to this model, interest must be determined exactly by demand. The interest rate needs to adjust to the point that balances demand for consumption and the demand for money and the demand for investment (and government 'demand'). This is supposedly the point where the IS and LM curves cross. But this completely leaves out the obvious cause for the rate of interest: the supply and demand for present goods (as exchanged against future goods). It ignores a couple hundred years of advancement on the understanding of interest.

if the government will really do more good than bad, I would be all for letting the government manage our resources--all of it.

And ather thing that doesn't make sense to me: Before the increase of government spending, we assume we are in a state of equilibrium. In particular, the supply of funds for loans must equal the demand for loans. Then the interest rate increases in order to equilibrate demand. But the higher rate of interest makes the supply of loans increase (as indicated by the LM curve)--people become more eager to loan money--and the demand for loans to decrease--people become less eager to borrow. But surely this means that there is now a surplus of lendable funds! People want to lend more than people are willing to borrow. We cannot possibly be at a new equilibrium.

Would the people really be more eagar to lend their money, especially when the Fed drives short-term interest rates so low?

[joel]

if the government will really do more good than bad, I would be all for letting the government manage our resources--all of it.

How do you reconcile this with Mises' argument that economic calculation is impossible under socialism? --that even if the government agents are perfectly virtuous and wise and do their best to do more good than bad, socialist control of the means of production would result in capital consumption and the disintegration of the social order, because economic calculation would be impossible.

Would the people really be more eagar to lend their money, especially when the Fed drives short-term interest rates so low?

I was specifically referring to the case of an increase in interest rates.

[Augustine2004]

How do you reconcile this with Mises' argument that economic calculation is impossible under socialism? --that even if the government agents are perfectly virtuous and wise and do their best to do more good than bad, socialist control of the means of production would result in capital consumption and the disintegration of the social order, because economic calculation would be impossible.

I was being sarcastic.

I was specifically referring to the case of an increase in interest rates.

OK, but the tendency of the Fed is to drive down interest rates. Only when it feels the need to moderate inflation (think Paul Volcker) does it deign to raise interest rates.

Suppose the economy came low to a point where it had no capital, but the people finally shucked off the government.

At first we would have Y = C + S, C being consumption and S being savings.

S eventually becomes in part investment S = S1 + I + Int, I being investment in capital goods and Int being investment in interest-paying instruments. The return from I is not instant, of course. It could take years for part of our savings to finally return to Y. Let that be R. Let Rint be the interest paid by the instruments.

Y = C + S, still, even though there’s plenty of capital. Y would have grown, though:

Ynew = W + R + Rint + other, other being gifts and inheritances (we have to take care not to double-count) and W being salary or wages.

It’s getting too complicated for me.