Thursday, March 21, 2013

Steve Roth — Scott Sumner Does Not Understand that S ≠ I

This is basic sectoral accounting, a subject in which neoclassical (and “market monetarist”) economists seem to have received no training.
Asymptosis
Scott Sumner Does Not Understand that S ≠ I
Steve Roth

We knew this already but Steve reminds us of it.

Ramanan posts a clarification of different uses of S = I that can lead to ambiguity and confusion.


The Saving = Investment Identity



24 comments:

Ramanan said...
This comment has been removed by the author.
Ramanan said...

Actually S = I even for the world as a whole with many governments and currencies.

This is because the S here is the sum of saving of all sectors of the world _unlike_ the S used in sectoral balances where it is private saving.

Tom Hickey said...

One of the big no-no's in use of technical language is the use of the same symbol in different contexts giving it different meaning without specifying the differences. It's called "ambiguity."

Tom Hickey said...

I updated the post on S ≠ I with a link to Ramanan's clarification at his place.

Anonymous said...

Ram's post doesn't clear anything up for me. Even after you put the subscripts in so that the glabal S is not confused with the domestic private sector S, I have no idea why the sum of all saving in the world must equal the sum of all investment in the world.

Ramanan said...

Dan,

I wasn't meant to be a full explanation. It was just meant to state that "S=I" isn't wrong as people assert it is wrong. This is done by showing how S - I = G - T with appropriate subscripts, becomes S = I with different subscripts.

Jose Guilherme said...
This comment has been removed by the author.
Jose Guilherme said...


I have no idea why the sum of all saving in the world must equal the sum of all investment in the world

It all follows from the definitions. We're talking accounting tautologies here.

For the world as a whole there is no foreign sector.

There remain only two sectors: the government sector and the private sector.

Seen together, the private sector and the government sector become a single sector. Ergo the overall S = I identity.

Seen together (adding them up) they either consume or save. Either C or S. Y = C + S.

Seen together, they spend on either consumption goods or investment goods. Either C or I. Y = C + I

It follows that S - total savings of the private sector(s) and the government(s) - must equal I - total investment by the private sector(s) and the government(s) - for the world as a whole (or for a country with a zero current account deficit).

It's all in the definitions. Income received is equal to income spent. Income not spent as consumption (savings) must have been spent as investment.

Private sector savings can differ from (can be higher or lower than) private sector investment; government sector savings can differ from government investment. But, absent a foreign sector, Sprivate + Sgovt = Iprivate + Igovt

No great economic truths to be found in these identities. For a meaningful economic model that is consistent with accounting rules we'll have to go to something like Paul Krugman's cross diagram and the Fullwiler paper derived from it.

Tom Hickey said...

Stated conceptually, this is abut transactions in aggregate, i.e. all entries in journals in a period.

All expenditure shows up as someone's income. This means that aggregate expenditure on "product" in a period is aggregate income in that period (flow).

This is symbolized as Y (income) = GDP (expenditure on product)

Some product is consumed during the period and some not consumed in the period but held over. For producers, the former is the consumption goods (c) and the later capital goods (including inventory).

This is symbolized as GDP = C + I.

For consumers, the former is spending on consumption goods (C) and what remains after consumption as a residual, which is called "saving" (S).

Since consumption is equal to consumption goods, then the value of goods not consumed but paid for (I) must be equal to the value of what is saved (S).

Transactions of governments are essentially the same as everyone else in this regard in terms of income and expenditure, so they can just be included in the aggregates.

The key point is that what is left over of production after consumption in the period has to balance on the LHS and RHS of the accounts.

Anonymous said...

Ramanan, I'm not making myself clear. Here's my issue. You say:

"Government’s saving is T − G(current)"

And then derive the rest. Now if, as you say:

G = G(current) + I(govt)

which seems entirely reasonable, then we can derive:

S(govt)= T - (G - I(govt))

and S(govt) = (T - G) + I(govt)

But that just seems like a weird result. If T - G is the government's net income for the period - it's tax revenues minus its expenditures - why would we say that the government's "saving" is that number plus its investment for the period?

Since the result follows from the two initial assumptions, it seems to me that the issue is with the claim that S(gov) = T − G(current)"

In some sense people are entitled to define these terms however they wish, but if they use misleading terms then they create misleading impressions. For example, if someone tried to convince me that in the world as a whole births during a given period always equaled deaths, I would be extremely skeptical. If they then proved it by noting that they defice "births" to be the sum of "life births", "cremations" and "burials", then I would say they are using "birth" in a weird and idiosyncratic way. I think we should feel the same way about attempts to define saving and investment into equality.

Anonymous said...

Tom, how does that accounting system take into account the fact that some income is not the result of someone else's expenditure at all?

I guess I'm looking at this very classically. Imagine a world consisting only of self-sufficient farmers with no exchange. Each year, each farmer has a harvest. From the crops harvested, they separate the seeds from the grain. That harvest is their annual income. They can then do one of three mutually exclusive things with the products of the harvest, seeds and grain combined: (i) they can consume it, (ii) they can store it for a period extending into the next year, (iii) they can plant it in the ground to grow next year's harvest. It seems to me that these are the three traditional foundations of the concepts of consumption, saving and investment.

Now you can use a broader concept of saving to include not just "storing", but all unconsumed income, so that saving = storing + investing. That seems OK. But even in that case, it will be rare that saving = investing across the whole economy.

So in a world without exchange, there is still income. If you add exchange to the economy, you then take into account that the grower of apples might exchange some of his harvest for a portion of his neighbor's wheat harvest. If the the two growers exchange 100 apples for a bushel of wheat, and then they both eat what they got in exchange, then macroeconomically, that is the same as if the the original growers had eaten their own product. All we care about is the ultimate disposition of the product after the exchanges have taken place.

Tom Hickey said...

Dan, if you think about it, you will get confused. I finally figured that out. I'ts really just about how the accounting is done, this means that in double-entry, the LHS and RHS has to balance. How these balance has to do with the accounting rules. It's just fule following.

Of course there are issues arising from this as you point out. One really big issue is taking all transaction in an economy over a period as contributing to GDP regardless of whether they are productive of actual goods (stuff).

This means that a lot of things that actually non-productive or counter-productive get included and the entire informal economy is ignored, even though much of it is highly productive or at least contributory.

For such reasons, I have concluded that the usefulness of economics in policy formulation is limited, and its uncritical use is potentially misleading.

Ramanan said...

Dan,

It is true there are a lot of misleading terminologies.

Actually my point of writing the post was because - forget behavioural equations/models - people mix accounting identities themselves.

And mixing accounting identities is the beginning of the process of mixing things up more in more complicated analysis.

For example if one wants to talk of planned investment etc, one should be using two symbols and so on ... etc.

About your question. Imagine the government is in a surplus of $100bn and its investment is $500bn in one particular year. We are assuming gross here and zero depreciation for simplicity.

At the end of this period, The government's net worth has increased by $600bn because the government's liabilities have reduced by $100bn and its stock of fixed assets has increased by $500bn.

But saving is equal to the increase in net worth (assuming away capital gains/losses).

So the $600bn makes sense.

Of course none of this means the government should save etc etc.

The term net income is a bit ambiguous. Imagine a corporation which purchases fixed assets. You do not subtract the associated expenditure to reach its "net income".

Anonymous said...

Tom, I'm more inclined to believe that if an accounting system is misleading, we should propose new ones that are less misleading.

Let me say why I think it is important to get this one right. There is a long tradition of recognizing that some individuals have a higher marginal propensity to consume their income than others. It is generally believed to be the case that the more income one has, the more one is disposed to save it at the margin, and so taxes on the wealthy are more likely to lay hold of surplus savings, rather reduce consumption.

One way in which the rich have tried to defend themselves against these taxes is to argue that what they do not consume they invest, and so taxing them heavily reduces the country's capital development and and inhibits growth. Of course, some people don't use their monetary income either to buy consumption goods or capital goods directly. They buy financial assets. But then the question is what the seller of the asset does with the monetary income. They might, for example, use it to buy consumption goods or capital goods themselves.

Some would therefore have us believe that all monetary income is being used, whether directly or indirectly through the kind of financial intermediation just described, for consumption or investment. Thus, if we define saving as all income not laid out for consumption, they are saying that saving = investment.

But this is clearly not true. First there are old-fashioned ways of saving that have nothing to do with consumption or investment - for example, converting it to cash that you simply put in a safe. Also in the contemporary world you can buy financial assets that ultimately derive their interest flows from government provisions of reserves, not from any share in the profits of capital development. And there are also assets that are based on speculative and ponzi-like systems in which the interest-payoffs are juts derived from additional asset purchases.

So I don't regard S=I as just an innocent piece of accounting conventionality. I see it as a bit of flim-flam that attempts to redefine and twist previously existing concepts in subtle ways so as to bake a certain policy recommendation into the accounting cake.

Jose Guilherme said...

the issue is with the claim that S(gov) = T − G(current)

It's all a matter of definitions, but it does have a certain internal logic.

If "Saving" is defined, in general, as "Income not spent on consumption", then Govt. saving should logically be its income (taxes) minus its consumption (current) expenditures.

Whereas the govt. net saving is (Sgovt - Igovt), that is, the budget surplus.

Just like the private sector net saving is (Sprivate - Iprivate).

For a closed economy (or an economy with a balanced current account) there is no net saving, because S(private + govt.) = I(private + govt.) by definition.

Again, all accounting tautologies.

For accounting-consistent insights about the way the economy works when planned Investment and planned Saving schedules change (not to be confused with the ex-post accounting identities), one should go to the Krugman-Fullwiler simple, yet powerful and accurate, model.

Tom Hickey said...

Tom, I'm more inclined to believe that if an accounting system is misleading, we should propose new ones that are less misleading.

I think there are two issues. The first is that accounting is not transparent other than to people that understand it fairly well, which excludes most people and which is why they have their taxes prepared by a professional, for instance. It's just not worth the time and effort to learn the rules. Often professional systems and literature are purposely opaque to ensure that professionals are employed to interpret it.

For example, companies with large and influential MIS staffs often made hardware purchasing decisions based on recommendation of the staff that lead to purchase not of the most appropriate equipment but equipment that would ensure than the MIS staff was necessary and indeed privileged. This allow extraction of rents due to artificial exclusivity.

The second issue is inappropriate use of accounting to misrepresent of deceive. Of course, embezzlement comes to mind, which John Kenneth Galbraith called "the bezzle," and Bill Black calls "control fraud." Creative accounting is notorious as not only an extractive device but also for putting lipstick on a pig.

In politics all kinds of accounting devices are used to mispresent and sow disinformation, like "unfunded liabilities' that will "bankrupt the nation."

It's all about duping the rubes, and the media are generally not smart enough to figure it out. Greenspan shrugged off the FBI fraud division, too, when the lawyer/accounts warned of massive mortgage fraud in 2004. This was because of ideological blinders, which he implicitly admitted later.

So I agree that all this needs to be made much more transparent, which, I think, MMT is attempting to do and I suspect that is a reason that many professionals are pushing bank on the exposure.

Jose Guilherme said...

One way in which the rich have tried to defend themselves against these taxes is to argue that what they do not consume they invest, and so taxing them heavily reduces the country's capital development and and inhibits growth

This is where the complete statement of the S = I identity may be useful - to counterbalance such claims.

Since Sprivate + Sgovt = Iprivate + Igovt, someone should tell said "rich people" that govt. investment may well be higher or socially more useful than private investment.

"Malinvestment" can be attributed to either the private sector or the government. It's certainly not the case that only the government may make bad investments.

For instance, the recent bubble and its collapse were a direct consequence of a gigantic case of malinvestment by the private sector.

So, contrary to the claims of "rich people" accounting identities are not neoliberal: they are politically neutral.

It's only their distorted presentation by certain individuals and groups that sometimes carries a clear neoliberal bias.

vimothy said...

Steve writes,

This is basic sectoral accounting, a subject in which neoclassical (and “market monetarist”) economists seem to have received no training.

But ironically, doesn't seem to understand the identity either. No, it doesn't hang around whether we're in a barter economy with no government sector.

For any aggregate entity S=I unless there can be cross sector or country flows, in which case, S=I at the top level. Barter doesn't have anything to do with it.

Say that the economy is closed. Then S=I for the whole economy. I.e., national saving equals aggregate investment. Or say that the economy is open. Then we can have S>I or S<I, but (ignoring any statistical discrepancies introduced by measurement error) the sum of the differences across all countries equals zero. I.e., S=I for the whole world.

vimothy said...

Dan,

The answer to your question is that you have defined saving and investment in your self-sufficient farming scenario idiosyncratically.

Storing and planting grain is investment (corresponding to inventory investment and investment in fixed capital). Saving is not consuming the grain. Unconsumed grain is equal to planted grain plus stored grain. Hence, saving equals investment.

Tom Hickey said...

The way I understand the analysis — please let me know if I do not have it quite right — is that national income and product accounting (NIPA) results in S =I wrt national income and product as the two sides, however it does not account for saving of net financial assets in aggregate, which the flow of funds account does.

The NIPA accounting shows that change in nominal of firm expenditure (flow in period) is equal to the nominal value of change in saving (flow in period) in that the change in ownership is reflect on both sides as being equal. that is to say, firms are owned by households, so the change in value of firms gets reflected on household income statements. In the (closed) private sector, this nets to zero due to double entry rules.

However, in the flow of funds account, flow of financial assets is included in a way that they is not reflected in NIPA. When govt is included, net financial assets held by non-govt in aggregate enters the picture, and this amount changes within periods affecting the equation. JKH breaks this out as S = I + (S-I).

vimothy said...

To my knowledge, the NIPA and FoF give somewhat, but not entirely, equivalent measures of private saving--so that both will give a measure of JKH's quantity (S - I) (i.e., MMT's net saving), but this measure will be slightly different.

A comparison of the two can be found in this St Louis Fed paper: http://research.stlouisfed.org/publications/review/07/11/Guidolin.pdf

But I think that's a bit tangential. Pace Steve Roth, it's not necessary to understand esoteric facts about national accounting to understand basic economic relationships like saving and investment.

Why there is ambiguity is that lots of different economic entities can save and invest. For example, we could imagine that there are only two people in the world, A and B.

If A and B are unable to lend to one another, then both will have S=I. If they are able to lend to one another, then A could have S>I, S=I or S < I. Because there are only two people in this world, B is the complement of A, meaning that if S>I for A, S < I for B, and so on.

But it's not actually necessary for A and B to be people. This holds if A and B are sectors or countries--or whatever. As long as A and B are complements, then the net saving of one is the net dissaving of the other.

It's not even necessary for A and B to be the same sort of entities. As long as they are complements then A and B's combined saving must equal A and B's combined investment. A could be a person and B could be the rest of the world. A could be a country and B could be the rest of the world. A could be planet Zorg and B could be the other planets in the Galactic Federation.

Point is that you have a global level of analysis for which S=I. Then you can decompose that into entities (who are disjoint and whose union equals the global set) who do not have to have S=I, but who are constrained in terms of having to maintain consistency so that S=I globally at all times--ie, (S - I) = 0 when summing over all entities.

Tom Hickey said...

In the U.S., the Bureau of Economic Analysis (BEA) produces income and capital accounts: National Income and Product Accounts (NIPAs)

The Federal Reserve Board (FRB) produces financial accounts: Flow of Funds Accounts (FFAs)

BEA and FRB are currently working to integrate the accounts to create a balance sheet that meets international guidelines

The NIPAs show nonfinancial accounts and the FFAs feature financial accounts; together they provide a comprehensive picture of the government sector.

While the concepts in both accounts are the same, differing data sources and estimation methods employed cause the results to differ

NIPA – Flow of Funds Accounts Integration


BEA Government Statistics Users Conference
Charlotte Anne Bond

vimothy said...
This comment has been removed by the author.
vimothy said...

Yes, but they can both be used to give you a picture of private saving net of investment--the NIPA measure is coming from the income and expenditure and the FoF is coming from assets and liabilities. This is because changes in wealth equals the difference in net disposable income and consumption. The NIPA gives you the right hand side of this equation and the FoF gives you the left.