1. Did a financial crisis cause a fall in expected and actual aggregate demand? (With central banks being unable or unwilling to do enough to stop it).
2. Or did a fall in expected and actual aggregate demand cause (or worsen) a financial crisis? (With central banks being unable or unwilling to do enough to stop it).
Obviously, there is a positive feedback multiplier between financial crises and expected and actual aggregate demand growth. (If central banks are unable or unwilling to prevent there being positive feedback.)
I started to write a post on this, but scrapped it, because I couldn't think of any way to provide good evidence one way or the other.
Now David Beckworth has a very important post which provides evidence for the second hypothesis. Go read it. Or simply look at the picture. US households' expected dollar income growth started to fall well before the US financial crisis.
[Update: see also Marcus Nunes' post and graph.]
The purpose of this post is simply to draw attention to David's post, and to raise a few minor points:
1. It's not enough just to look at the US. The same thing happened in a lot of other countries at the same time.
2. Households' average expectations of their own nominal household income growth isn't exactly the same as expectations of nominal GDP growth.
3. You can see the 1982 and 2009 recessions clearly foretold in David's picture. But you can't see, or can only just see, the 1991 and 2001 recessions.
4. David's graph starts to head south in 2005. It's just a bit too early for comfort, if you want to say that declining expected nominal income is what caused the 2008 financial crisis and 2009 recession.
5. Putting 3 and 4 together: while it is a commonplace in intertemporal macroeconomic theory that a decline in expected future nominal income would cause a decline in current desired nominal expenditure ("aggregate demand"), that can't be the only thing that affects aggregate demand. Interest rates, for example, would be another thing. Can those "other things" fill in the small gaps between David's picture and what actually happened?
6. "Never reason from an expectations change". What caused/allowed expectations to change like that?
In any case, Scott Sumner's crazy idea that declining actual/expected NGDP is what caused (at least a large part of) the financial crisis, rather than vice versa, is looking a lot less crazy. And maybe it's time to put macro into finance, rather than (just) vice versa.
Gotta go grade exams, and do a lot of admin stuff.
Worthwhile Canadian Initiative 
If I loan you $1000, I create an $1000 asset, the loan, in place of a $1000 cash asset and you create a $1000 cash asset and a $1000 liability for the loan. In a sense nothing has happened, since both sides have the same net worth before and after.
If I were a bank, however, you would two get pieces of paper (conceptually). One would say you owed me $1000 and the other that I owed you $1000. You would give me the same. However since I was a bank, my liability (your deposit) is money. You could, of course, ask me to satisfy it in cash, but then you would deposit it in some other bank and they would have the liability (deposit).
For each borrower there is a lender because financial assets and liabilities are created pairwise. The difference is that if I weren’t a bank, my IOU to you wouldn’t be very good money (would not be very liquid).
Moreover, as a bank, I can lend as much as I like subject to some not very restrictive reserve requirements and a more restrictive capital to assets at risk ratio. A bank will lend up to it’s capital limit (with a small margin for prudence) provided that it can find credit-worthy borrowers willing to pay a satisfactory premium.
When banks are lending less than they could, it means either that there is insufficient demand for loans at the premium or that there is a lack of borrowers who meet the banks’ credit standards.
The Fed can control bank lending by changing the rate it charges for lending reserves, but it doesn’t directly restrict the amount of reserves it lends, and banks can borrow reserves elsewhere. So money is endogenous, but the Fed can restrict the amount created to the extent it controls the cost of funds and is willing to raise short term interest rates.
Note that as we go up the money hierarchy from M1 to M4, the Fed’s control becomes clumsier. It can be unwilling to go to the necessary lengths to control something like M4, for both economic and political reasons.
The important point is that increased credit becomes increased investment and adds to GDP, but the way GDP is calculated hides this effect. GDP = GDI = GDE is an accounting identity, and it’s dangerous to draw conclusions from it about aspects of the economy it deliberately ignores.
For this stuff I highly recommend Monetary Economics by Godley and Lavoie, which is stock and flow correct. The software to implement their models is freely available.
As someone once said:
“There’s something really wrong with the way we do short run macroeconomics. We focus all our attention on the output of newly-produced goods and services. That’s what we call “Y”. We talk about Aggregate Demand and Aggregate Supply, and what we mean by AD and AS is the demand and supply of those same newly-produced goods and services.
Keynesians then go on to divide Y into C+I+G, and C+S+T. Monetarists talk about MV=PY. Both agree that a recession is a fall in Y, caused by a drop in demand for Y.
But a moment’s reflection tells you this is wrong. It’s not just new stuff that is harder to sell in a recession; it’s old stuff too. New cars and old cars. New houses and old houses. New paintings and old paintings. New furniture and antique furniture. New machine tools and old machine tools. New land and old land.”
http://worthwhile.typepad.com/worthwhile_canadian_initi/2011/12/why-y.html
Peter N,
Y is divided into Personal Consumption Expenditures + Business Investment + Government Expenditures + Net Exports ( Keynesian style )
Personal Consumption Expenditures = Business Investment * Markup (Value added, profits, etc) + Purchase of existing goods from prior periods
The problem is not in Y, it is a problem in separating the nominal, real, AND depreciated value of Y.
Using the car example. It is relatively easy to track the prices of new cars and guage an inflation rate on new cars. Used cars are trickier because of depreciation. And so you really have three prices for a used car – it’s nominal price, its inflation adjusted price, and its depreciation adjusted price. A car manufactured in 2010 sold today will typically sell for less in a year’s time even if it has the exact same mileage, wear, and tear.
Used cars don’t exist. The car itself is an existing good. A car can only be produced once. GDP only takes account of added value, where value is basically labor cost. That allows GDP to equal GDI, since all goods are investment when they are produced. The investment value is the cost, which is simultaneous. Since cost = cost, you have an identity, even though the goods produced can’t possibly be the same goods for which the costs were incurred. The methodology ignores the fact that production takes time.
The closer you look at GDP, the more you realize what a poor basis it is for macroeconomics.
Peter N.
“GDP only takes account of added value…”
No it doesn’t.
http://www.bea.gov/iTable/iTable.cfm?reqid=12&step=1&acrdn=2#reqid=12&step=3&isuri=1&1203=17
This is the BEA’s (Bureau of Economic Analysis) report for Personal Consumption Expenditures (First quarter 2011 through First quarter 2013).
Line #11 – Used Autos
Line #15 – Used Light Trucks
Nick wrote: “. David’s graph starts to head south in 2005. It’s just a bit too early for comfort, if you want to say that declining expected nominal income is what caused the 2008 financial crisis and 2009 recession.”
You should know better than to admit something like that in public, Nick, as it inspired this sophomoric post.
With used cars, what’s counted is the dealers added value. The dealership is a business, so its output has to be part of GDP, but its sort of a service.
Everything in GDP eventually maps to the cost of the labor that produced it. Otherwise GDI = GDP wouldn’t be an identity. Since the used car dealer is a business, its payroll direct and indirect makes a contribution to GDP.
If you as an individual sell your car, it doesn’t contribute to GDP. If I’m reading this correctly, the amount the dealership pays for the used car isn’t income to the seller, so it won’t appear in GDI or GDP. Likewise it shouldn’t appear in GDE. It couldn’t be a consumer expenditure to the buyer from the dealership unless it was consumer income for the seller to the dealership, otherwise the accounts wouldn’t balance.
The dealer’s added value has to be an expenditure for the buyer so the dealer’s labor costs can be income for the dealer’s employees. Again, otherwise the books wouldn’t balance.
GDI = GDP = GDE is an identity by definition.
Peter N,
Most used vehicles orginate as trade ins. And so the number of people selling their car on their own is insignificant compared to the number of people that trade in their car for a newer model. The used cars are then sold back to the public by used car dealers. The sale of the used car does not show up in the income account for the seller because the net of the transaction (new car price paid – used car price received) is positive.
When a business sells anything from inventory to a consumer, this is consumer expenditure and business disinvestment. This mostly doesn’t affect GDP. I say mostly because the rules are complicated. This what the BEA says:
”
Net transactions
Net transactions between persons and other sectors of the economy primarily consist of the wholesale value of purchases by persons from dealers less sales by persons to dealers (either directly or as trade-ins). In addition, transactions may occur between persons and businesses other than dealers (such as the sale of scrapped vehicles), government, and nonresidents. Transactions among persons are intrasectoral and so do not affect PCE.
For both benchmark and nonbenchmark years, estimates of net transactions are developed by valuing the annual change in unit stocks of used motor vehicles held by persons, rather than by explicitly taking into account each type of transaction listed above. Year end unit stocks of used autos and of used light trucks are estimated for each year of original sale (vehicles greater than 11 years old are grouped together) using annual data on new motor vehicle purchases and retention information developed from R.L. Polk & Co. data on vehicles in use by model year. Unit stocks held by business are based on business purchases of new motor vehicles and on retention rates for rental vehicles (6–18 months), leased vehicles (2–4 years), and other business vehicles (1–9 years). Unit stocks held by government are based on government purchases of new vehicles and on assumed retention rates. Stocks held by persons are then calculated as the residual.
Changes in the unit stocks of autos and of light trucks held by persons reflect purchases of new vehicles, scrappage of old vehicles, and net unit transactions other than scrappage. Purchases of new autos and of light trucks by persons are estimated separately (see the section “New motor vehicles”). Scrapped units are calculated by age of vehicle as a proportion of total vehicle scrappage; this proportion is assumed to be equal to the ratio of the unit stock held by persons to the total unit stock. Net unit transactions other than scrappage is then calculated as the residual.
The changes in unit stocks, grouped by age, are then valued at wholesale prices.The average wholesale value for each age group of used autos and of used light trucks is based on average auction prices by model year from ADESA. Scrapped units by age are valued at 8 percent of the wholesale price.
Current quarterly and monthly estimates of net transactions are extrapolated from the annual estimates, using data on retail sales of used car dealers from the monthly retail trade survey. The estimates of real net transactions are prepared by deflation, using the PI for used autos and trucks.
“For each borrower there is a lender because financial assets and liabilities are created pairwise.”
That is one way. I’d like to know what Nick thinks about that.
“If I loan you $1000, I create an $1000 asset, the loan, in place of a $1000 cash asset and you create a $1000 cash asset and a $1000 liability for the loan. In a sense nothing has happened, since both sides have the same net worth before and after.
If I were a bank, however, you would two get pieces of paper (conceptually). One would say you owed me $1000 and the other that I owed you $1000. You would give me the same. However since I was a bank, my liability (your deposit) is money. You could, of course, ask me to satisfy it in cash, but then you would deposit it in some other bank and they would have the liability (deposit).”
So are you saying borrowing $1,000 from a friend is the same as borrowing $1,000 from a bank?
“The Fed can control bank lending by changing the rate it charges for lending reserves, but it doesn’t directly restrict the amount of reserves it lends, and banks can borrow reserves elsewhere. So money is endogenous, but the Fed can restrict the amount created to the extent it controls the cost of funds and is willing to raise short term interest rates.”
I’d change “can control bank lending” to “can attempt to control bank lending”.
Plus, what is your definition of money?
“So are you saying borrowing $1,000 from a friend is the same as borrowing $1,000 from a bank?”
No. The accounting works the same way, but the moneyness of the two liabilities is different. You can’t write a check on my liability as a lender and have it universally accepted, whereas you can do so with the bank’s liability, which is usually known as a deposit.
Money is anything that possesses enough moneyness to have economic significance. You can measure moneyness by the calculating the asset’s liquidity premium (which isn’t always possible). J P Koning is the blogging guru for thia topic: http://jpkoning.blogspot.com/
The aggregates M0 – M4 are commonly used to track the supply of money. Divisia M3, M4- and M4 are a welcome addition to the family of aggregates, since the Fed killed off the old M3 (which had problems, in any case).
I think that “can attempt to limit bank lending” is better. The Fed’s ability to encourage bank lending appears to be much weaker. It is worth noting that trends in the larger aggregates tend to lead M0, not trail it.
IIR Minsky said something like: “Anybody can issue money, the trick is to get other people to accept it.”
Peter N,
Governments figured out how to do that a long time ago. They levy taxes.
This is change in debt as a percent of GDP flow charted against change in GDP. Note the close correlation.
It’s not the stock of national debt that has an immediate effect om GDP, it’s the flow of total debt.
TMCDO is total credit market debt owed
HSTCMDODNS is total credit market debt owed by the household sector
TCMDODNS is total credit market debt owed by the domestic non-financial sector
TCMDODFS is total credit market debt owed by the domestic financial sector
“No. The accounting works the same way, but the moneyness of the two liabilities is different. You can’t write a check on my liability as a lender and have it universally accepted, whereas you can do so with the bank’s liability, which is usually known as a deposit.”
What if you borrow $1,000 in currency from a friend?
W. Peden,
“The bank looks to borrow reserves from other banks to meet its liabilities.” So the textbooks say, but in reality? In reality reserve requirements aren’t very restrictive. Banks have other sources of funds than demand deposits. The real limits on lending are the ratio of capital to assets at risk, the cost of funds, the availability of credit-worthy borrowers. And, of course, banks sometimes would rather invest than lend.
I have papers from economists about this going back 15 or 20 years. The Fed influences bank lending by using the discount rate to increase or decrease the cost of funds. Assuming a sloping demand curve for loans, and a fairly stable bank margin, increasing the discount rate will decrease the demand for loans as long as banks don’t have cheaper sources of funding.
And, of course, as you go up the hierarchy of aggregates from M1 to M4, the Fed’s fine tuning ability gets progressively weaker.
I’m curious why imcnoe inequality is not mentioned — it seems to be the elephant in the room, both in the U.S. and in China. In that situation, you are not going to get rid of excessive patience. It’s almost impossible for the top 1% to consume that much — they are more or less immune to interest rates. But if you believe in a consumption function (a.k.a Carroll and Kimball — 1996) that is strictly concave in wealth, then re-distribution should be the cure for excessive patience, just as (I believe) the structural shifts towards top earners was the cause of the excessive patience. And better imcnoe security would also help.