Circular Flow of Income theory

csbrown28 said:

But I think that imports result in the export of our dollar, trillions of dollars of which are saved by foreign interests.

Click to expand...

You can't cut out the mechanics behind the identity because they don't match the theory... savings = investment within the NIPA framework. All the identity tells us from imports > exports is that this differential will be used to fund subsequent investment. If there is a fiscal deficit, some of this imbalance will be lent to the public sector at the behest of the private sector. Sometimes this can be desirable. Sometimes it is counterproductive.

Point being, crowding out of private investment is a detriment to long term economic growth. This cannot be identified from within JFC's (or your own) framework.

Deficit spending/ debt is money that replaces the money that's currently "leaked" out of our economy.

Click to expand...

Again... your definition of leakage doesn't fit within the framework. You can't pick and choose what aspects of the NIPA identity that are to adhered to and ignored at will.

...a lot of investment these days isn't productive. Speculating on real estate or energy removes as much money as it adds and does very little relative to the volume of money that moves through those markets to put people to work. Investment should be broken into subcategories IMO as not all investment is of the same quality.

Click to expand...

Nobody has put forth the claim all investment is equal or yields equally productive/profitable returns.

I just think that imports create a fiscal problem that if misunderstood will cause an accounting problem that could be totally avoided if only we understood the nature of our currency.

Click to expand...

What are these potential fiscal and accounting problems?

csbrown28 said:

The idea that inflation is by default bad is simply wrong. (...) Without trying to sound like I'm being argumentative, good in this context is pretty subjective. When you say good or bad, you have to share with me what you think the goal of economic policy should be in the context of this convo.

Click to expand...

First of all, it's not bad by default in the sense that (a) it is not an assumption and (b) it's not even obvious given typical assumptions that the costs of inflation would turn out to be large enough to even matter. However, you are correct that I need to be more specific about this kind of argument. So, let me be more specific.

Most of our relevant macroeconomic models are built around a representative household. You can think of this as if we bothered only looking at the average household. The choices of that household are modeled using constraints (budget, liquidity, timing, etc.) and preferences, as you would find in an introductory microeconomics textbook. Those preferences impose an ordering on combinations of consumption and leisure across time. The interesting part is that this gives you a normative criterion (it tells you implicitly what is preferred), but you cannot twist those preferences however you want because you need them to replicate many key features of macroeconomic data.

Now, the idea is this: I'm solving my macroeconomic model with some inflation target (say, 2%). Then, I'm solving my macroeconomic model with another inflation target (say, 4%). Then I do the following thought experiment: if I had to reduce the consumption of the 2% economy household by a fixed amount every single period in a way that would make that household indifferent between his 2% economy and the 4% economy, what would that percentage be? And that is how you measure the welfare costs of inflation. In fact, all cost-benefit analysis you ever see are made by economists using this type of reasoning.

There are valid objections to this methodology. The obvious one is that you might not like my criterion. I'm suggesting a way to get something like a Pareto improvement (formally, it's Kaldhor-Hicks that I am invoking) and a way to make the criterion less arbitrary (by requiring that its empirical implications be realistic), but you might not like it. Another more subtle problem is the realism of the model: is it good enough an approximation? Well, the paper I linked above offers a stunning match of empirical patterns. It is by no means perfect, but we're talking about a model that matches dozens of correlations, cross-correlations, autocorrelations, and volatility patterns... and it's not just the sign that is correct because the magnitudes are also correct. Of course, I mentioned above that this is a representative agent model. What about models with many households? Here, the question would be: maybe the inflation is costlier to some than others and someone on the left would certainly not be as alarmed if most of that cost was borne by wealthier people. Unfortunately, having many types of households makes things vastly more complicated and, to my knowledge, we cannot yet make models that are as sophisticated as the ones I showed above in that context, but a lot of work is being done on it nowadays.

A last problem is that the costs are channeled through wage dispersion in both Calvo-style and Taylor-style wage rigidities in this model. It is also true for prices (it's the same type of mechanism that mimics pricing inertia), but the costs there are very small. The problem is that we just put our hand on data to test this dispersion effect, but only for prices on consumption goods (you need prices for many types of goods going all the way back to the 1970s to compare high versus low inflation rate environments) and it's not true. So, the model uses an empirically false mechanism for prices (but it doesn't matter for inflation costs) and we have no idea if wage dispersion does increase in a high inflation rate environment. So, there is a very subtle theoretical point here: the results may rely on things that might not work. On the other hand, it's the best answer we have at the moment.

csbrown28 said:

Here I agree again, but efficiency and inefficiency measured against what? Productive output or social mobility. I don't mean to sound like someone who only cares about people at the cost of corporations or the "wealthy", but I think that the scales are being tipped in a direction that favors efficiency at the cost of human welfare. I think we need to strive for a balance and I don't think neoclassical economics considers this at all.

Click to expand...

It concerns neither production nor social mobility, but Pareto efficiency. I was trying to explain how higher inflation induces economic costs in medium-scale New Keynesian models. In economics, prices often act as coordinating mechanisms. Under some idealized conditions, price ratios reflect marginal rates of substitutions from individual prefers and from technological constraints. In New Keynesian models that represent nominal wage and price rigidities through Calvo pricing and monopolistic competition, price ratios do not perfectly reflect how each type of goods and each type of labor can be substituted and that in turn distorts how household and firms make their choices. Higher inflation rates happen to exaggerate those distortions.

As I said in a previous post, we could potentially use what we call a heterogeneous agent model to make meaningful concerns such as yours. We could ask how rapidly households can move across income strata, or if some households are hit more than others by different types of policies. So, it might please you to learn that your comment about neoclassical economics is wrong: the mathematics we use to deal with questions regarding inequality and social mobility is more or less the same we have been using since the late 1970s.

If you are curious, the mathematical difficulty of dealing with heterogeneous agent models is that the choices of everyone in the model depend on the state of the system. In a normal DSGE model, the state of the system is the value of a handful of variables: aggregate capital stock, maybe past consumption if we include habit formation in the model and a handful of random variables we call shocks. So, you have to keep track of maybe a dozen things at most. Now, move on to a heterogeneous agent DSGE model. Over time, the distribution of people over things like employment status, wealth, leverage, etc. changes, so that distribution is a state variable. If you ever took a statistics course, a light just off in your head: hey, but hold on, a distribution is a function. Yes, but functions are vectors and the whole problem is that they are infinite-dimensional vectors and that doesn't fit in a computer for solving my model numerically. So, you need clever ways to approximate that and that makes the work considerably harder.

csbrown28 said:

I've spent a LONG time studying econ. That said I realize there are large gaps in what I know. You obviously have knowledge that I do not and I conceded that I could learn a lot from someone who, based on some of your posts here, like you. That said, I think that there is also something that people like JohnfC and I have to add to this conversation. Believe me, I would take great pleasure in meeting with groups of people, like your self, in person and hashing this stuff out to fill gaps in my knowledge.

I for one would like to thank you and Kush for taking the time to have this conversation,

Click to expand...

Answering noneconomists and taking account of their comments is a good way for me to work through what I learned.

csbrown28 said:

Do people have money to borrow because other people save it, or do people have savings because people have the capacity to borrow?

Click to expand...

In theory, what we would want in a model is a supply and a demand for savings. That is how you would think about it: some people want to save, others want to borrow and interests are how you coordinate both of them. You can make this very, very complicated (e.g., they do it through banks, or through many types of investment vehicles like securities and simple deposits, some people have more information than others, etc.), but the gist is always the same. Some people want to transfer funds into the future, others would prefer to have funds now and both of those behaviors need to be consistent.

In practice, what happens is that unless you get in the heterogeneous agent framework, the savings dynamics are more abstract because there is only one household. For some vehicles of saving, the equilibrium conditions are that the quantity of some assets is 0 in equilibrium. Why? Because the household is the one who both supplies and demands the asset, so on net balance they must have none. Trading volumes thus are often undefined, though the price of those assets is always well defined. Your question makes more sense when many different households are involved.

Kushinator said:

Point being, crowding out of private investment is a detriment to long term economic growth. This cannot be identified from within JFC's (or your own) framework.

Click to expand...

Ok, I grabbed this from Investopedia....

One of the most common forms of crowding out takes place when a large government, like that of the U.S., increases its borrowing. The sheer scale of this borrowing can lead to substantial rises in the real interest rate, which has the effect of absorbing the economy's lending capacity and of discouraging businesses from making capital investments.

Click to expand...

Yet, the data does not support this commonly held position.




And I believe this is true because lending does not rely on savings it relies on capital which is supplied by investors. When people take loans they aren't creating scarcity in savings or capital, because loans create new deposits. The only thing that scarce and can be "crowded out" are the things of real value. Resources, labor, land, etc. This makes inflation the only real constraint.

Do you disagree?

Did I misunderstand what you were saying?

Kushinator said:

Again... your definition of leakage doesn't fit within the framework. You can't pick and choose what aspects of the NIPA identity are to adhere to and ignored at will.

Click to expand...

Step away from the framework for just a sec... Please explain to me how foreign interests saving US dollars does not result in reduced demand in the domestic economy.

csbrown28 said:

And I believe this is true because lending does not rely on savings it relies on capital which is supplied by investors. When people take loans they aren't creating scarcity in savings or capital, because loans create new deposits. The only thing that scarce and can be "crowded out" are the things of real value. Resources, labor, land, etc. This makes inflation the only real constraint.

Click to expand...

Have you been paying attention to what's been going on in repo markets over the past month? The continued issuance of Treasury securities in conjunction with a modestly growing economy and (at the time) a central bank inclined to normalize monetary policy has created considerable strain on the financial system... so much that the Fed has opened a liquidity facility to purchase enough Treasury securities to keep reserves from their continued decline.

Government spending can crowd out private investment, especially as deficits grow in an economy that creates 2+ million jobs per year. Without the Fed, interest rates for private debt would be at substantially higher levels.

Do you disagree?

Click to expand...

Absolutely.

Did I misunderstand what you were saying?

Click to expand...

Yes.

Please explain to me how foreign interests saving US dollars does not result in reduced demand in the domestic economy.

Click to expand...

These proceeds of domestic dollars fund domestic investment.

Kushinator said:

...crowding out of private investment is a detriment to long term economic growth.

Click to expand...

Just to clarify, when you say investment=real goods and services?

csbrown28 said:

Just to clarify, when you say investment=real goods and services?

Click to expand...

I don't understand what your asking. Are you asking me what i consider investment?

Kushinator said:

I don't understand what your asking. Are you asking me what i consider investment?

Click to expand...

Yes, when you say investment, what do you mean?

csbrown28 said:

Yes, when you say investment, what do you mean?

Click to expand...

It can be real estate purchases or leases, facility construction, maintenance, research, training, financial securities, etc.... We can even go by the NIPA definition or refer to economics texts, but it doesn't change my point. People, businesses, or even countries do not typically hold large cash balances outside of their needs for operation. It is naive to believe foreign entities (or anyone for that matter) just sit on and just accumulate piles of cash.

TheEconomist said:

My intention was not to drown your claims but to be precise about how economists work out the relationship between real variables and nominal variables.*If you're going to make the case that it is wrong, the first thing to do is to be very clear about it is you criticize.

And I don't think of DSGE models as exactly correct, as much as I think they are useful tools to organize arguments. I know that Lawrence Christiano, Martin Eichenbaum, and Paul Krugman have expressed this kind of view in the past. John Cochrane also has a similar view and it's the opinion all macroeconomists under which I studied share.



Let me try it again.

The most fundamental concepts in economics are preferences and option costs. Here, the important bit concerns option costs. The idea is that when you can't have your bread and eat it too, so one must be sacrificed to obtain the other. That which you sacrifice is the option cost. Now, in a context where you have prices, this means that what will matter are price ratios: the rate at which the market allows you to sacrifice, say, little American flags for some beer on every July 4th is what matters.

This feature is not a bug, but a fundamental feature of economic theory. It's everywhere, including in our most sophisticated models and because it's the ratio of prices that matters, you get the core of what makes money neutral in some models: if you multiply all prices by 10, no ratio has changed (because you multiply the numerator and denominator of each ratio by 10), hence, if nothing else has changed, the behavior of people is the same. The way you work around this is obvious: you want some, but not all prices to be multiplied by 10 so that some ratios change and the behavior of people therefore also changes, even if nothing else happened. That's what price stickiness does in New Keynesian models: it creates drag in average price adjustments (sometimes also in wage adjustments) that have consequences on real variables like output, consumption, investment, etc.

So, to respond to a comment you made previously, money doesn't come first in this view. What is fundamental is how many hours I need to work to buy a bag of oranges and prices are just instruments that convey this information, even if I gave above an explanation as to how this information might be distorted by delays in pricing adjustments. You don't care about dollars, except through what dollars can get you if you prefer. That's the point of view of contemporary economics, though I left out some subtleties, hopefully to make more accessible.



This will shock you, but the arguments made by Krugman and Summers are not esoteric. Krugman's Wonderland analogy, his arguments about the ineffectiveness of monetary policy at the zero lower bound, about how unconventional monetary policies should play out, about the power of fiscal policy during a zero lower bound spell, etc. ALL of that without a single exception comes out of a New Keynesian DSGE model. Krugman uses the IS/LM diagrams to get his point across to a lay audience, but the point can be even more forcefully be made in a New Keynesian model.

Click to expand...

Very impressive series of posts here from you, John and Kush. I understand the need for modeling and why macro ends up becoming a mathematical discussion fairly quickly once you get past the ideology. What I continue to be skeptical about though is the usefulness of mathematics to predict future outcomes in a field such as macro. Way back in the 70s when I took macro in college I distinctly remember being the kid in class that asked the simple question challenging the assumptions being presented. That question was based upon human nature not the slope of a curve or the linear regression in favor at the time. At the very heart of all three of these fine posters comments is the complex nature of money and how it is created to facilitate a modern economy. Its an endlessly fascinating subject and if one can just keep ones political biases out of the discussion as these three have done here, the rest of us can learn something of value. Thank you all three of you.

Vadinho said:

Very impressive series of posts here from you, John and Kush. I understand the need for modeling and why macro ends up becoming a mathematical discussion fairly quickly once you get past the ideology. What I continue to be skeptical about though is the usefulness of mathematics to predict future outcomes in a field such as macro.

Click to expand...

The possibility of forecasting depends on what kind of variable you have in mind. If you are talking about production and employment data, it is not impossible to have a good performance. In fact, although theoretical models are not designed specifically for producing forecasts, more recent developments have made their performance improve for these kinds of variables. Of course, they aren't the best tools we have for forecasting macroeconomic activity. They are designed to organize discussions about policy, so their focus is getting forecasts conditional on policy changes right whereas forecasting future economic activity is about getting unconditional forecasts right. An example of a model that does very well:

1. The principal components of a large macroeconomic dataset is a simple dimension-reduction tool that helps summarize the variations of large datasets in just a handful of variables. It's just a smart way to pick weights to build new variables as weighted means of many other variables. You can either see this as a statistical trick, or you can also see it as a core implication of approximate solutions to DSGE models (because DSGE models basically say that a handful of latent variables drive all economic activity and that the rest is just sector-specific noise);
2. You put lagged values of those PC and lagged values of the variables you want to forecast in a random forest. It is extremely hard to beat that and the forecasts are quite good, even around turning points (moving in and out of recessions) and even at horizons of 12 and 24 months ahead.

With that being said, my tune would change if your point was to forecast stock market returns. Yes, we know methods that will beat what we call a random walk, so it is not totally unforecastable. However, your best model will have trouble capturing more than 4 or 5% of what happens in the future. In other words, it's practically useless to make point forecasts in financial markets. Moreover, I have my shares of doubt regarding the value of point forecasts period. If you use point forecasts to make choices, your success hinges on the accuracy of your forecast... Even for more predictable economic variables, it's a bad plan. Often, it's easier to think in terms of conditional responses to very bad cases and try to make choices that are robust to mistakes -- or, better, that can make you benefit from mistakes -- than it is to try to make a good forecast. I like DSGE models because they help think through complicated ideas, anticipate different kinds of troubles and compare policies, not because they do well at pinning down the interest rate in Bulgaria in three months.

Vadinho said:

Way back in the 70s when I took macro in college I distinctly remember being the kid in class that asked the simple question challenging the assumptions being presented. That question was based upon human nature not the slope of a curve or the linear regression in favor at the time.

Click to expand...

Mathematics is useful if you understand it's just a good way to think. Otherwise, it will provide a false sense of knowledge and understanding. The difference between good and bad economists is the difference between these two sentences.

TheEconomist said:

The possibility of forecasting depends on what kind of variable you have in mind. If you are talking about production and employment data, it is not impossible to have a good performance. In fact, although theoretical models are not designed specifically for producing forecasts, more recent developments have made their performance improve for these kinds of variables. Of course, they aren't the best tools we have for forecasting macroeconomic activity. They are designed to organize discussions about policy, so their focus is getting forecasts conditional on policy changes right whereas forecasting future economic activity is about getting unconditional forecasts right. An example of a model that does very well:

1. The principal components of a large macroeconomic dataset is a simple dimension-reduction tool that helps summarize the variations of large datasets in just a handful of variables. It's just a smart way to pick weights to build new variables as weighted means of many other variables. You can either see this as a statistical trick, or you can also see it as a core implication of approximate solutions to DSGE models (because DSGE models basically say that a handful of latent variables drive all economic activity and that the rest is just sector-specific noise);
2. You put lagged values of those PC and lagged values of the variables you want to forecast in a random forest. It is extremely hard to beat that and the forecasts are quite good, even around turning points (moving in and out of recessions) and even at horizons of 12 and 24 months ahead.

With that being said, my tune would change if your point was to forecast stock market returns. Yes, we know methods that will beat what we call a random walk, so it is not totally unforecastable. However, your best model will have trouble capturing more than 4 or 5% of what happens in the future. In other words, it's practically useless to make point forecasts in financial markets. Moreover, I have my shares of doubt regarding the value of point forecasts period. If you use point forecasts to make choices, your success hinges on the accuracy of your forecast... Even for more predictable economic variables, it's a bad plan. Often, it's easier to think in terms of conditional responses to very bad cases and try to make choices that are robust to mistakes -- or, better, that can make you benefit from mistakes -- than it is to try to make a good forecast. I like DSGE models because they help think through complicated ideas, anticipate different kinds of troubles and compare policies, not because they do well at pinning down the interest rate in Bulgaria in three months.



Mathematics is useful if you understand it's just a good way to think. Otherwise, it will provide a false sense of knowledge and understanding. The difference between good and bad economists is the difference between these two sentences.

Click to expand...

Well said and I have to commend you for your contributions to this thread. Unlike you and others, I am a generalist and not very good at the details or the specifics. I guess its a matter of genetics because I am the same way about a lot of topics that interest me. I can see from your posts that you are not a dogmatist. My rule is that if you constrain yourself to a dogma, you are very likely to miss the next truth.

Apologies if I'm reviving a dead thread, but being new here and having just read through the whole thing, I'd like to post a few thoughts.

...the purchase of imported goods and services has no direct impact on GDP.

Click to expand...

This is true from a purely accounting standpoint. Instead of calling out M separately, I could net it out of C and GDP wouldn't change. However, if you consider what C represents--consumption--a reasonable person could conclude that there was demand that was not (but could have been) satisfied domestically. For example, say I am going to buy a $30,000 car (I'm deliberately using the FRED example scenario). No matter if I buy one from the US or Japan, C remains unchanged at $30K. If I bought a domestically produced car, that $30K gets added to GDP, but if I buy the one from Japan I have to subtract that $30K. In that sense, imports do impact GDP in terms of opportunity cost.

Imports and output move together:

Click to expand...

To me this graph doesn't prove that imports don't lower GDP. Instead, it just shows that rising aggregate demand increases demand for both domestic and imported goods.

Getting back to the original post, I think I do agree with JohnfrmClevelan. In order to have growth in real production, there first must be increases in demand. Since overall, the private sector likes to net-save and we export more than we import, we basically have two options to help demand. We either have to continually run a government deficit or we have to ensure that private debt continues to increase. What other course is there? The work I've read by Steve Keen on private debt has me more worried about that, than continuing to run government deficits.