The American Enterprise Institute recently had a symposium on QE after 10 years. It began with Desmond Lachman interviewing Ben Bernanke. At one point Lachman asked about the influence of asset price increases on demand. Lachman referenced studies that estimated an impact of 4 cents on the dollar, presumably meaning that an extra dollar in aggregate wealth leads to an extra 4 cents of aggregate demand.
Should we take those estimates seriously? I’d say no, for a wide variety of reasons. Most importantly, they overlook two serious problems, which make any sort of wealth effect unreliable. These are “never reason from a wealth change” and “monetary offset”. Let’s start with never reason from a wealth change.
It’s impossible to know the impact of a change in wealth without first knowing why wealth has changed. Thus suppose that wealth increases because real interest rates have fallen (a very plausible assumption, BTW, regarding asset prices increases during the 21st century.) In that case, it’s not clear that people would have any incentive to spend more than before.
Think about the following example. We start with a country with $1 trillion in wealth, all in land. Interest rates are 4% and the land yields an income of $40 billion per year. People consume 70% of that figure. Now assume that the global real interest rate falls to 2%. The value of the land doubles, to $2 trillion. But the flow of income from the land does not change, staying at $40 billion. So why would the public spend more? After all, their flow of income has not changed. One could do a similar example with common stocks and a changing P/E ratio.
The lesson is that one should never reason from a wealth change; first consider why the total stock of wealth has changed.
Now let’s assume that wealth goes up because our economy is becoming more productive, yielding a higher expected future flow of income. In that case can we expect more nominal spending? Maybe, but not because wealth has risen, rather because aggregate supply has risen and the central bank is targeting inflation. Thus real GDP will rise as inflation stays stable–which leads to more NGDP. On the other hand, if the central bank were targeting NGDP, then a positive supply shock would not boost nominal spending, due to monetary offset. Aggregate demand should be stable under NGDP targeting; hence under that policy regime any wealth effect estimates are nothing more than an estimate of central bank incompetence.
The US has seen an enormous growth in wealth since 2011, and yet productivity growth remains sluggish. Thus no one should be surprised that aggregate demand continues to grow quite slowly. As wealth rises, the Fed raises interest rates fast enough to keep inflation close to 2%.
PS. The symposium on QE had four panelists. Three gave talks full of vague warnings about possible financial imbalances, but with no plausible explanation as to why this should impact the stance of monetary policy. In contrast, Joe Gagnon gave an excellent defense of QE, and also several themes that market monetarists have emphasized. Here is a slide from his presentation:
Nominal GDP targeting, level targeting, and target the forecast. If we had done those three things in 2008-09, the recession would probably have been far milder.
READER COMMENTS
ChrisA
Jun 15 2018 at 5:45am
Scott – in your first example, isn’t the case that if you are a holder of land and the value of the land doubles (and you expect this increase to be stable) by the permanent income hypothesis you should increase your spending? Maybe I am naive but I would say that should my bond fund double in value, even if due to falling interest rates, I would increase my spending rate?
Benjamin Cole
Jun 15 2018 at 6:15am
Great post!
michael pettengill
Jun 15 2018 at 8:31am
I’m old fashion and stick with Adam Smith’s almost off hand definition:
Changing the price of something has no impact on real wealth.
Only changing the product of labor can change real wealth.
It’s easy to reduce the product of labor, which reduces wealth.
What debt is is time shifted labor payment. When workers aren’t paid to work, they can’t pay for the past labor in assets they are currently using by way of debt, whether it be a car or house.
The loss of wealth is not from the falling prices of assets, but from the falling product of labor. When the worker with debt is not producing, he can’t repay the worker whose savings he borrowed, and he cant use his saving to pay for the product of other worker’s.
The Fed does play a role in ensuring the worker with savings can get his savings back to pay other workers to work. But it’s of no help to the unemployed worker. Lacking income from work, his wealth has fallen and thus he can’t consume the product of other worker’s, so they produce less. The Fed can’t give him a new job.
No new jobs, no increase in wealth.
Defining wealth in terms of the quantity of assets, like gold, silver, warehouses of grain, is mercantilism.
Scott Sumner
Jun 15 2018 at 2:45pm
ChrisA, To make it simple, suppose you have a 30-year bond that pays $1000/year for 30 years. You consume $800 of that interest and save $200. Now suppose the current market price of the bond doubles, but it continues to pay $1000/year. I don’t see why you would not keep spending $800 and saving $200.
Even if there are second order effects I missed, surely you wouldn’t start spending $1600/year in response?
Thaomas
Jun 15 2018 at 4:37pm
There are good theoretical reason to thing that a change in wealth could be ambiguous. Take the fall in real interest rates this will depend on the reason for the fall. Suppose we suppose that the economy has become more prone to recession because the central bank abandons its de facto ngdp target for an inflation rate ceiling. Seeing slower growth in income in the future might lead to greater saving to compensate the lower return on existing wealth or to substitute current consumption for future consumption whose relative price has fallen.
Given theoretical ambiguity why not look to data controlling of course for the cause in the increase in wealth.
Seppo
Jun 15 2018 at 5:07pm
If I’m about to retire and my retirement savings suddenly double in value from 1 million to 2 millions and interests that I earn on them plummet from 4% to 2%, I’m sure it will increase my willingness to spend.
Spending 50k on a new car reduces my interest income by $1000 instead of $2000 and insurers would for sure offer better annuities because in my expected 20 remaining years of life the amount of capital is more important for annuity payment than the interest income.
—
In opposite scenario (halving of savings + doubling of interest rates) I would probably delay retirement and increase savings for last years at work to top up my savings before retiring.
bill
Jun 16 2018 at 3:35pm
@Sepppo, I think you make a good point. I think also that the price increase/lower interest rate would make the case of saving for retirement more difficult for people that are say 20 years from retirement. So in aggregate the changes lead to no wealth effect (as Scott has laid out).
Sven
Jun 15 2018 at 6:19pm
So why would the public spend more? After all, their flow of income has not changed.
I think this scenario is to easy. There are several ways why this kind of wealth increase would lead to higher spending, one of them being the “balance sheet” effect. This will make credit more easily accessible by those who gain a better equity position by that kind of wealth change. This in turn will increase consumer spending and investment spending (and therefore maybe even productivity).
See the very old Bernanke paper “Inside the Black Box: The Credit Channel of Monetary Policy Transmission” where he identifies that the balance sheet channel (and therefore also the wealth effect) is the main reason why central bank induced interest rate changes will lead to more aggregate demand by easier access to credit.
Scott Sumner
Jun 15 2018 at 9:56pm
Seppo, Recall that debt is both an asset and a liability.
Sven, I don’t deny that there may be some second order effects, but that’s very different from the “wealth effect” that Keynesians have in mind.
Ethan R.
Jun 27 2018 at 11:17am
Hi Scott, excellent post (as usual).
I have a slightly related question. Do you believe that redistribution can make the business cycle more stable?
To be clear, I’m talking about a particular kind of redistribution:
1. I’m not referring to counter-cyclical fiscal policy. I’m wondering about a secular change in the long run amount of redistribution that the government does.
2. I’m also not trying to get at an automatic.fiscal stabilizer idea like unemployment insurance or something like that.
3. Perhaps more specifically, what I’m thinking about would look something like Oren Cass’s proposal for wage subsidies that you often speak about. Wage subsidies might have a slight counter cyclical effect in that wages tend to decrease during recessions (though I suspect they will be ineffective because wages are sticky in the short run, meaning that there will be a significant lag), but again, I’m trying to isolate that effect out.
Let me explain my thoughts on this. From this post, you argue that there is no significant wealth effect for the aggregate macroeconomy. But, if I understand you correctly, there is a significant wealth effect on micro scales – if you give a poor person more money then they will almost certainly increase their consumption.
I know that consumption isn’t the only part of aggregate demand. But I’m thinking that because poor people have a higher marginal propensity to consume, a greater proportion of NGDP will come from consumption. From my understanding of Friedman’s permanent income hypothesis, absent any exogenous changes in expected income, consumption will always be fairly stable. Does this imply that the business cycle will be more stable in an economy with more redistribution?
A more speculative argument I’m thinking of has to do with the economic diversity of different regional areas in a currency area. If we redistribute from wealthier regions to poorer regions, perhaps the wealth effect can increase the number of firms in poorer regions that currently might have a very small number of firms. This might have an effect on the business cycle by mitigating the effect of asynchronous regional business cycles.
I hope to hear your thoughts!
Comments are closed.