Abolish the Equity Premium Now!

This post is slightly whimsical. I can't decide myself if I'm being totally serious. My argument is certainly less than watertight. But it's not (to me) obviously wrong either. So I'm just throwing it out there.

On average, stocks outperform government bonds. This is called the "Equity Premium". Nobody understands why (yes, we know stocks are riskier than bonds, but that doesn't seem to be enough to explain it). This is called the "Equity Premium Puzzle". (See Brad DeLong and Konstantin Magin (pdf) or Wikipedia for a discussion).

Lets say that the equity premium is around 5% for a ballpark estimate, which is good enough for my purposes here.

"Stock market capitalisation to GDP ratio" is the market value of all equities traded on stock exchanges as a ratio of annual GDP. This ratio varies across time and across countries.

A very quick and dirty Google gives me a ballpark estimate of around 100% for that ratio for advanced countries like Canada. Total stock market capitalisation is around the same size as annual GDP.

Suppose the government could wave a magic wand, make stocks as attractive to hold as government bonds, and abolish the equity premium. Multiplying my two ballpark estimates together, the benefits of waving the magic wand would be worth 5% of GDP, for each and every year the magic wand was waved. That's a big deal. And (assuming my two ballpark estimates are correct) that's a lower bound estimate of the benefits of waving the magic wand. Because if the wand were waved, and the equity premium were abolished, firms would presumably issue more stocks and households would hold more stocks. (Just like if you could wave a wand and make transportation costs disappear, the benefits would exceed current transportation costs, because more goods would be transported unless the supply or demand curves were perfectly inelastic. There's a triangle, as well as a rectangle, between the supply and demand curves.)

All we need is a magic wand.

We have a magic wand. It's monetary policy. All we need to do is have the central bank target the stock market total return index. If the central bank used monetary policy to target the total return index at (say) 7% per year, then anyone investing their savings in that stock market index would be guaranteed a return of 7% each and every year. That's a nominal return, of course, unadjusted for inflation. But it would be exactly the same as buying a government bond earning 7% per year nominal that could be bought or sold at face value any time you felt like it. Both returns would be guaranteed by the government. The stock market index portfolio, and government bonds, would become perfect substitutes. The equity premium would disappear.

Would having monetary policy target the stock market total return index be good for macroeconomic stabilisation? I don't know. But 5% of GDP (lower bound) is a benefit that's not to be sneezed at.

And maybe, just maybe, there would be a lot less bond-finance and bank-finance of investment if we abolished the equity premium, which would reduce the size and probability of financial crises too.

[Brad DeLong's recent post "The Economic Costs of Fear" inspired me to write this one.]

I'm off to buy myself a new canoe, to do my bit to increase Aggregate Demand, and because I deserve it. Back later this weekend.

Comments

[Frances Woolley]

Nick, given how the markets are doing, I'm figuring that the equity premium was abolished some time ago, it's just that no one is willing to admit it yet. (See my post on "Does the equity premium still exist").

"Because if the wand were waved, and the equity premium were abolished, firms would presumably issue more stocks and households would hold more stocks." This makes sense if the equity premium exists because stocks are intrinsically undesirable investments, and thus people are not prepared to pay much for them at the time of IPO, and thus they're a poor way for firms to raise capital. So abolishing the equity premium would be equivalent to abolishing whatever it is that makes equities intrinsically undesirable investments.

On the other hand, if people believe that an equity premium exists and will continue to exist, and are able to manage risk through portfolio diversification, shouldn't they favour stocks over bonds? This is, after all, the standard financial advice given to younger investors looking for growth. In which case I don't see that abolishing the equity premium would cause households to hold more stocks.

Excellent news about the new canoe.

[Nick Rowe]

Frances: we won't know for sure whether the equity premium has been abolished until many years after the fact. Brad DeLong (the blog post I linked above) says it's 7% in the US right now. That's a forecast, of course, because he doesn't know for sure what the stock market is going to do in future, and he's basing it on historical ratios. But, given the dividend yields and P/E ratios of (say) Canadian bank stocks right now, compared to the much lower than normal government bond yields (mostly negative in real terms for short bonds), my guess is that the equity premium is as big as ever.

The weird thing about the equity premium: when the ex ante equity premium falls, that causes stock prices to rise, so the ex post equity premium rises! And vice versa. So my interpretation of the last few years is that the ex ante equity premium increased, as the demand for liquidity increased, and fear increased, which caused the ex post equity premium to fall.

[ptuomov]

Your scheme might make bonds and stock priced the same way by making nominal bonds riskier and not by making stocks less risky. The amount of money printing needed to compensate for a bear market would be large and lead to inflation that would erode the real value of the bonds. A lot of the adjustment would come from the bonds being less safe and therefore discounted at a higher discount rate.

More generally, your scheme would be the opposite of tranching. Tranching assets or pools of assets to different risk level tranches generally increases the value that people are willing to pay for the asset by a small amount. I would think that your scheme of untranching the stocks and bonds would lead to a corresponding small loss.

[Simon van Norden]

Nick; Your comment reminds me of arguments I've had with central bankers who talk about the virtues of forward-looking monetary policy and/or exchange-rate targeting. When I ask them what they would have against taking their own arguments to the next step and targeting the stock market index, they usually just change the subject. (Like you, I wasn't sure whether I was being serious or not.) I don't want to suggest that it would be a panacea (anyone want to try to figure out the Non-Accelerating Inflation Rate of Stock Returns? and what makes it vary over time?), but if we take seriously arguments about the importance of efficient capital allocation and forward-looking policy, then I think we need to better understand why we shy away from this.

But your post also gives me an excuse to ask you a question that's been bugging me for a while. Last time I looked, a basic condition for efficiency in a dynamic economy is that the interest rate on real investments needs to be higher than the long-term growth rate. (If not, Ponzi games become sustainable and welfare-improving, among other things.) But looking up the TIPS yield, I see it is currently negative for 10-yr US bonds. I think I know that's below the expected growth rate of the US economy. You teach macro: how do you tell your grad students to think about that?

[Ian Lippert]

The equity premium is likely just an artifact of many statistical biases. Here's a good explanation:

http://falkenblog.blogspot.ca/2009/07/is-equity-risk-premium-actually-zero.html

[Lord]

Doesn't gdp already include the growth of stocks, at least to the extent those returns are consumed which I assume would amount to about half, so that half of 5% or 2.5% is the gdp growth rate?

[Peter N]

News flash: Charles Koch has been buying university economics departments. He currently has three (that are known).

http://www.tampabay.com/news/business/billionaires-role-in-hiring-decisions-at-florida-state-university-raises/1168680

They don't seem to be very expensive, so he has many times the amount of money needed to buy every economics department in North America, though I expect some won't sell. Still, no more useless controversy, the standard will be Kochenomics.

[123]

Nick, do you want to target real or nominal returns? You need to target real returns if you want to abolish risk premium, but then there is no nominal anchor.
Anyway, you need to have a very deep pockets if you want to abolish risk premia ( which is another way of saying your scheme will blow up ).

[123]

Nick,
Here is an alternative title for your post: "Peg the Chicago VIX at zero".

[rsj]

The central bank cannot target equity returns to be whatever you want. Other than the law, a desire that we are in a private ownership economy so that the government does not own private capital, there are two things working against you:

By lowering the short rate, you are forcing a divergence between long duration and short duration assets. Yields between short and long dated assets must increase when rates fall. This is an arbitrage-free condition having nothing to do with risk per se.

Adding risk, as stock dividend yields fall, the duration increases. As the duration increases, the stock becomes more sensitive to changes in dividend yields. A 1% change in yield has a bigger effect on the value of the stock. Assuming that the underlying volatility in earnings is constant, a stock with a lower dividend yield will be much more volatile, and so will require a higher discount.

The net result of that is that there is a limit, below which dividend yields will not fall, and further attempts to lower the yield only add more volatility. Adding more volatility is economically harmful. You don't get bubbles in stock prices when nominal rates are at 10%, even if inflation is at 10%. But you do get bubbles when short rates are at 2%, even if inflation is at 0. People are less likely to make mistakes estimating the earnings of a particular firm over a 10 year period of time rather than over a 50 year period of time. The firm may not even exist in 50 periods. And remember -- each individual firm must be valued. You cannot buy "the market".

DeLong is crazy. Bonds have been outperforming stocks over the last two decades, and every indication is that this will continue.

Dividend yields are already at 2%, almost a historic low, and DeLong thinks that this is still too high, and that yields should be at 1%, meaning a duration of over 100 years. There is a reason that yields will never fall so low -- people are unable to make predictions that sensitive that far out for individual firms. It is all well and good for an economist to make predictions about the market as a whole, but in the stock market we have to price individual firms. Market cap is determined from the bottom up.

Even if there is only a 1% chance that the firm will go bankrupt every year, if the firm is only paying out 1% of its market cap each year, then its a bad deal for investors. The fact that some new firm will take its place is of little concern to the pricing of the firm with the 1% dividend yield.

[Andrew F]

^ Great point.

[K]

Rsj: " if the firm is only paying out 1% of its market cap each year, then its a bad deal for investors"

No. Not if it's retaining earnings that can always be paid out later. If it takes too long the firm can be LBOed or otherwise privatized. Cash can also be returned via equity buy backs. Apart from transparency, the dividend yield is not very relevant.

Nick,

I think the basic idea is right, though I'm in favour of backing with stocks (and bonds), not targeting them. It's not the same thing since, as we discussed recently (Woolsey suggested I was nuts, if you recall) targeting can fail as it needs to be backed by sufficient taxation power. So, done correctly, I think your idea is the same as the capital asset backed money we've discussed before, but with a 7%/year split rate.

[rsj]

K,

The dividend yield + cap gains is all that matters. The *only* way that investors can take delivery of re-invested earnings or share buybacks is via capital gains. The re-invested earnings are used to purchase more activity specific capital that is always at risk of becoming worthless or declining in value. That occurs should the firm be re-priced, in which case investors will no longer be able to take delivery of the re-invested earnings. Those earnings are gone.

Let me give you two scenarios. In the first scenario, I propose a speculative venture and promise to pay you 10% interest. But you will only get 1% interest in cash. The other 9% interest consists of me making additional speculative ventures in other areas. You have no control of when or if you ever get paid anything more than 1% -- that depends on the additional speculative ventures succeeding or not.

In the second scenario, I propose a speculative venture and promise to pay you 10% cash each period.

Do you really believe that you should apply the same risk discount to the two ventures? Is it "irrelevant" whether or not you get paid in cash or in promises?

[rsj]

And when DeLong says that the risk premium is 9%, he is claiming that, with a 2% dividend yield, he expects on-going capital gains of 7%. Does anyone -- including Brad -- really believe this?

[rsj]

A perfect example of the above is Adobe. Adobe has two nice monopolies with Photoshop and Illustrator, that generate massive profits. Those profits correspond to earnings. Yet Adobe has never paid out a single dividend, but has used its earnings to make a series of poor investments. Even though it has engaged in share buy backs, it has issued more shares than it has bought back.

From the time that Narayen took charge of Adobe in Dec. 2007 until now, Adobe has maintained an average earnings yield of 4.4%.

But how did investors benefit from this yield? They received zero dividend payments and suffered capital losses equal to 1/3 of their investment.

Being paid in more promises is not the same thing as being paid in cash. There is a reason why firms with high dividend yields are less volatile than firms with a lot of "earnings" but few dividends. That additional volatility requires an additional premium be charged.

[K]

Risk premium, BTW, is usually defined relative to the risk free *real* rate, since utility is usually thought to be a function of real, rather than nominal quantities. I think there's good reason to suspect, though, that people have an extreme form of nominal fixation, *whatever* the definition of the numeraire. May as well make the numeraire something that can actually transport consumption and which people are going to have to hold anyways, ie capital assets.

[rsj]

When you are subtracting two nominal rates, inflation doesn't matter. Again, what is being confounded here is skepticism over future dividend growth per additional dollar invested versus risk intolerance. The two are different.

Investors today do not believe that re-invested earnings will result in similar profit growth that the firm currently enjoys. Firms agree, which is why they are sitting on hoards of cash. Both firms and investors see a dearth of profitable future investment opportunities, but investors cannot force firms to dislodge those profits and pay higher dividends. They are forced to watch as the firms sink more and more money into poorly performing investments. HP purchases Compaq and then later announces they are shutting down their consumer electronics division. Then they change their mind. They purchase Palm, invest another billion, and then shut it down as well.

Investor's only option is to re-price the stock so that money not paid to them in the form of dividends is heavily discounted, because they assume it will be poorly invested by the firm.

This is a completely separate issue from their ability to bear volatility.

[K]

Rsj:

You said that if the dividend yield was less than the default rate then it's a bad investment. Like retained earnings are worthless. That's what I was responding to. If you are saying we need to discount earnings growth, then yes, clearly. That's partly why the P/E of stocks is 14 and the P/E of the 30-year bond is 40.

[rsj]

You said that if the dividend yield was less than the default rate then it's a bad investment.

Huh?!! No, I didn't. I said all that investors take delivery of are dividends plus capital gains.

Like retained earnings are worthless.

Didn't say that either.

But you *implied* that retained earnings are 100% equivalent to cash paid out. That is the only way you can argue that "dividends are irrelevant". I'm not the one making crazy statements here.

Do you no longer believe that? If so, what is an appropriate discount for the earning not being paid out?

Is it conceivable that this discount may change over time?

Do you believe that if it does change, then this must be due to an increase in risk-intolerance rather than a change in expectations of either future capital gains movements and/or future dividend growth?

[Scott Sumner]

Nick, Neat idea, but people might point to the US where until recently stocks were doing pretty well despite a sub-par economy.

Your canoe purchase will not boost AD. Any upward pressure on AD will be offset by tighter monetary policy, as the BOC attempts to keep inflation at 2%. Admittedly the policy comes later, but current AD depends on expected future monetary policy.

[JP Koning]

Nick: "But it would be exactly the same as buying a government bond earning 7% per year nominal that could be bought or sold at face value any time you felt like it."

It seems to me that are assuming away one of the real-world risks of holding bonds. In general, bonds can't be bought or sold at face value any time you feel like it. Rather, you get the face value returned at maturity, but it may be that (for personal reasons, say the need for liquidity, whatever) you are forced to sell the bond before maturity at a discount to face, upon which you have to endure a capital loss.

In order to make a government bond that can be "bought or sold at face value any time you felt like it", the government would have to modify its existing bond structure so as to offer bond investors a put-option feature at face-value. But this would increase the government's cost of capital, since at the moment markets do not require a put-at-face feature to make government bonds attractive.

So using your setup, abolishing the equity premium wouldn't be free, rather, it would penalize the government by raising its capital costs. The upshot is that the premium wouldn't disappear via equity yields falling to government yields. Rather, it would disappear via required returns on equity falling and required returns on government debt rising.

[K]

rsj: "Even if there is only a 1% chance that the firm will go bankrupt every year, if the firm is only paying out 1% of its market cap each year, then its a bad deal for investors."

I never called it crazy. I just disagreed. Anyways, tangential point. Forget it.

[Bernhard]

Nick, thanks for the interesting post.

When you state that “the benefits of waving the magic wand would be worth 5% of GDP, for each and every year the magic wand was waved”, are you implying the economy would be 5% larger every year going forward? If so, how would this work in the real economy unless the elimination of the equity risk premium somehow increases the level of overall productivity?

Perhaps, as you state, with the elimination of the equity premium, firms will issue more equity capital and fewer bonds. But thinking about the Modigliani–Miller theorem, this does not change the overall value of a firm, simply its capital structure. I wonder whether the elimination of the equity risk premium would simply adjust the relative price of a firm’s equity and debt, but not increase the firm’s overall value, eliminating the 5% gain to the economy.

[rp1]

The Fed has been using this policy for years. It's called the Greenpan Put, or now the Bernanke Put. How'd it work out?

[dlr]

Two questions.

(1) Do we really know enough about the equity premium to say whether eliminating it would be a welfare benefit? Having trouble explaining it isn't the same as knowing it's a welfare cost. I know it's a puzzle that has withstood a lot of attacks, but it's also a capital structure phenomenon that has been incredibly persistent across different markets with different characteristics.

(2) Are you really comfortable with a formulation that talks about the 5% benefit as a starting point supporting this idea and then the macro stabilization proprties as a supplemental question? If targeting the stock market turns out to be a poor AD stabilization method, say because of excessive volatility in time preference, risk or risk aversion (due to sunspots or RBC causes), that becomes a world with recursive implications for the real cost of capital, and perhaps renders unisurable systemic risk higher than it was before the policy. Wouldn't it make more sense to say that such a policy would provide a 5% GDP benefit (assuming the equity premium is actually too high) if and only if it was also a good macro stabilization target?

[Nick Rowe]

ptuomov:

I think you may have a good point there. If targeting the stock market total return made inflation much less predictable, then bonds would be riskier in real terms. So the gap between bonds and stocks would be closed, but partly by making bonds more risky (bad) and not just making stocks less risky (good). Whether or not that's a big deal depends on how much more volatile inflation would be.

"More generally, your scheme would be the opposite of tranching."

Sorry, but you lost me there.

Simon: "...but if we take seriously arguments about the importance of efficient capital allocation and forward-looking policy, then I think we need to better understand why we shy away from this."

Agreed. Actually, I think that's a much better rationale for my post than I gave. We ought to consider apparently whimsical or even "daft" policy ideas, just to make sure we really do understand why they are daft. And we might also learn something about sensible policy even if we do eventually conclude these daft ideas really are daft.

"You teach macro: how do you tell your grad students to think about that {r less than g]?"

Good question. One that bugs me too, and that i turn over in my mind every so often. Here are two posts I wrote 2 years ago, trying to get my thoughts together. Basically, if r is less than g, the economy "needs" a bubble/ponzi scheme. That "demand" for a bubble will create a supply. That bubble could be supplied by the government (unfunded pension plan a la Samuelson, or just rolling over the debt, because the long run government budget constraint is invalid if r less than g), or it could be supplied by the private sector. The trouble is, it's all unstable, because: there will be competition to be the supplier of the bubble; the bubble will cause r to increase until r=g, at which point the bubble/ponzi becomes unsustainable.

I think those were possibly my two most important posts, but I still can't fully figure out what it all means. Part of the problem is that we can't know for sure in practice whether r less than g will hold for certainty forever. But there ought to be some way to eat the free lunch provided by dynamic inefficiency even if r less than g holds only in expectation, and for a long period even if not forever. But I can't figure it out.

[Nick Rowe]

Ian: Dunno. There are two questions: is there an equity premium?; is there an equity premium puzzle?

People who know more than me say there is a puzzle. But this post is mainly based on the assumption that there is an equity premium, whther it's a puzzle or not.

Lord: I'm not sure I understand. GDP doesn't per se include the growth in stock prices. It does include the growth in output. Growth in stock prices may partly cause or be caused by growth in output, but these are not the same thing.

123: I'm talking about targeting nominal returns on stocks. Monetary policy can only target nominal stock prices, and the nominal returns on stocks. But since government bonds are (mostly) nominal assets too, and the equity premium means the premium of nominal returns on stocks over nominal returns on nominal bonds, monetary policy could eliminate that difference between stocks' and bonds' nominal yields.

I hadn't thought about the VIX. Targeting the total return index to a fixed number presumably implies making the VIX zero. I'm not sure if the reverse is true.

rsj: I'm afraid I don't follow you at all. Remember, the total return index is a nominal variable, not a real variable. I agree that monetary policy cannot target the real rate of return on the stock market. I'm talking about targeting the nominal rate of return. Central banks can target the nominal rate of return on: gold; the CPI basket of goods, etc. Why not the basket of stocks?

K: "So, done correctly, I think your idea is the same as the capital asset backed money we've discussed before, but with a 7%/year split rate."

Hmmm. Interesting way to think about it. So, the central bank issues "currency" 100% backed by stocks (instead of government bonds as at present), convertible on demand, but with a convertibility ratio falling at 7% per year. But I don't think that's the same as a 7% split rate, because when you do a split, the new currency would be paid (like interest) to existing holders of currency. In my view, the central bank's 7% profits (after costs) would be handed over to the government as seigniorage.

Scott: "Your canoe purchase will not boost AD."

Unfortunately you are right. But my puritan brain needed some excuse to be convinced that it was OK for me to go out and buy something I really didn't need to buy. Damn! What can i tell myself now?

Aha! I've got a new slogan, to refute you! "Never reason from a canoe purchase"! My buying a new canoe was an endogenous response to the Bank of Canada's monetary policy. The Bank of Canada wanted me to buy a new canoe, to keep AD where it wanted it to be. It was my duty to buy a new canoe, and if I hadn't done so I would have upset the Bank of Canada's carefully laid plans!

[Nick Rowe]

JP: "It seems to me that are assuming away one of the real-world risks of holding bonds. In general, bonds can't be bought or sold at face value any time you feel like it."

You are correct. I slid over that bit. But I think it just strengthens my case. My proposed monetary policy makes the return on holding the stock portfolio even better than the return on holding (existing) government bonds. It's like having a savings account, which you can deposit or withdraw from at any time, with the rate of interest guaranteed fixed indefinitely.

"So using your setup, abolishing the equity premium wouldn't be free, rather, it would penalize the government by raising its capital costs."

But that's just a pecuniary externality. The government loses, through higher borrowing costs, but those lending to the government gain an equal amount. If I wave a magic wand and make all canoes faster, that creates a cost on producers of kayaks, who will have to cut prices to make kayaks competitive with canoes. But buyers of kayaks gain what sellers of kayaks lose.

Bernhard: "When you state that “the benefits of waving the magic wand would be worth 5% of GDP, for each and every year the magic wand was waved”, are you implying the economy would be 5% larger every year going forward?"

Not really. GDP is not a good measure of welfare. Welfare can increase even if GDP stays the same. What I'm saying is that the increase in welfare *would be equivalent to* a 5% increase in GDP. The benefits would have the same value as 5% of GDP. It's unlikely that GDP would increase by 5%, though it might increase a bit, for the reasons you gave (increased investment).

rp1: "The Fed has been using this policy for years. It's called the Greenpan Put, or now the Bernanke Put. How'd it work out?"

If we judge it simply by what happened when the Fed stopped implementing the Greenspan put, and let both the stock market and economy tank, I would say that the Greenspan put worked very well, while it was in place.

dlr: "1) Do we really know enough about the equity premium to say whether eliminating it would be a welfare benefit? Having trouble explaining it isn't the same as knowing it's a welfare cost."

You are exactly right. That's why many economists say that we need to understand/model the economy if we want to do policy analysis properly. That's one of the reasons I said my argument wasn't 100% watertight.

"2) Are you really comfortable with a formulation that talks about the 5% benefit as a starting point supporting this idea and then the macro stabilization proprties as a supplemental question?"

No, I'm not. That's the second reason I said my argument wasn't 100% watertight!

[123]

Nick,
So basically you are talking about the stockmarket level targeting, NGDP is volatile, but stocks are stable. Not a good idea. And it actually does little to reduce the real equity risk premium ( excess return of stocks over TIPS)

[rsj]

Nick,

I made several points :)

1. As asset prices are set on the margin, the CB would need to credibly threaten to purchase all private capital in order to set its price. But we live in a private ownership society, which is why it is illegal for the CB to do this. We don't want the government owning (and therefore controlling) private business.

2. Arbitrage. The CB can only target *one* rate. The other rates are set from the one rate via arbitrage. If it tries to target two rates, then it creates an infinite stream of arbitrage. That is fiscal policy, so it is also illegal. Better to pay out more generous unemployment benefits rather than supply arbitrage income to capital holders, if aggregate demand increases via hidden fiscal policy is your goal. The specific example I gave is that wil low nominal rates, changes in the short rates force the spread between longer and short rates to go up. I believe Ross was the first to point this out -- e.g. that only the local EH is compatible with no arbitrage under general conditions.

3. You cannot set equity yields to be whatever you want, because as the dividend yield falls, volatility increases. A yield of zero, in the limit, means *infinite* volatility (as long as there is any uncertainty). If there is any form of risk aversion at all, that means that there is a lower bound beyond which you will not be able to drive dividend yields. What matters here is the nominal rate, not the real rate. The nominal rate is the one with the zero bound and the volatility asymptote.

To see this, imagine that you are trying to price a firm that sells music. You are not sure what the earnings of this firm will be -- perhaps it will be made obsolete and perhaps not. As new information comes out -- say new technologies for distributing music -- you revise your estimates of the earnings potential of this firm.

As you do for all firms.

To keep things simple, assume that the dividend yield will (permanently) grow at a rate of 1%. Suppose the rate of return for equities, which by definition will be the cost of capital, is 2%, so a share paying out $1 of dividends now is worth $100. This is all nominal. Now you get some new information and decide that the long run growth rate of this firm is actually 0%. What is the price of a share now? It is $50, because it needs to pay 2% now to compensate for 0% growth.

A 1% decline in the estimated growth rate of dividends forces a 50% decline in the value of the firm.
That is a lot of volatility, if you assume that future growth rates are constantly re-priced as new information comes out.

Suppose, however, that the rate of return for equities was 4%, and not 2%. Now, the same firm, with an estimated growth rate of 1%, will be priced at a 3% dividend yield, so a share paying out $1 will cost $33. With the same news, a change in the estimated growth rate to 0% will result in a new market price of $25/share. So the price only declines by 25%, not by 50%. You've cut the volatility in half!

When the return on equities is low, you are bubble land. All of this is of course nominal. There is nothing special about a 0% real return, but everything is special about a 0% nominal return. A 0% nominal return requires infinitely high capital values that are subject to infinite volatility, regardless of the inflation rate. A 0% real return does not require such undefined behavior.

When nominal returns are low, minute revisions to forward growth rates cause *massive* swings in capital values. That is not only welfare reducing, but investors begin to charge a premium for this added volatility. At some point, you will not succeed in forcing down the return any more, because any additional attempts to lower the yield cause the risk premium to increase by even more.

Look at the Nikkei to see what happens when governments try to force up equity prices or enter a secular period of low nominal rates. The dividend yield stayed low, but capital values took a horrible beating.

[Nick Rowe]

123: "So basically you are talking about the stockmarket level targeting, NGDP is volatile, but stocks are stable."

Yes.

But the main question, and the one I ducked, is "how volatile would NGDP be?" and would that volatility be big enough to outweigh the benefits of (say) 5% of GDP?

rsj: If I were arguing that the central bank should target real stock market returns then all your criticisms would be valid. You can't use monetary policy to target a real variable, because of long run monetary neutrality. But I'm not. I'm saying the central bank should target nominal stock market returns.

Think of it this way:

Suppose the central bank wanted to double (say) farmland prices relative to the CPI. Then the central bank would probably have to buy up most of (all?) the land that exists and keep it off the market. We are talking fiscal policy. That's not what I am talking about.

Now suppose the central bank wanted to double land prices in terms of money. That's easy. Just double the supply of money, and double all prices, including the price of land, by halving the value of money.

Or compare it to the old gold standard.

If a central bank wanted to control the real (relative to the CPI) price of gold, it would need to totally dominate the total market for gold, including all the gold mines. If the central bank wanted to double the price of gold, and if the elasticity of demand for gold were (say) one, it would need to buy up half the gold in existence anywhere in the world, and shut down half the gold mines or buy up half (or more) of all the newly mined gold. But if a central bank wanted to double the nominal price of gold, in terms of its money, all it needs to is double the supply of its own money.

Now, just switch "farmland" or "gold" in the above examples with "the stock market index".

[Nick Rowe]

rsj: this difference between real and nominal variables is absolutely central to monetary economics. It is totally unsurprising that different "schools of thought" are sometimes unable to understand each other if some of those "schools" don't get this distinction. Of course we will be arguing at cross-purposes.

Go back to the historical gold standard as an example. Different central banks around the world were able to set totally different nominal gold prices (in terms of their own monies). What they were really doing is setting the prices of their own monies in terms of gold. Even a small country's central bank could peg the nominal price of gold at whatever it wanted. It didn't have to dominate the whole world supply or demand of gold to hit this target. It just had to dominate the whole supply or demand for its own money. All those different nominal prices of gold were compatible with each other, because they were measured in different units: some in US dollars, some in UK pounds, etc.

If different central banks around the world had tried to set different real prices of gold, this would have been impossible. One central bank would be trying to buy up all the gold on the market to try to raise its real value, while a second central bank would be trying to sell more gold than was demanded to try to lower its real value.

[rsj]

1)

You can only control one interest rate. if the CB says that the price of overnight cash is 1%, but the price of 3 year cash is 7%, then people will arbitrage that, and the resulting arbitrage is fiscal policy. The return on equity is an interest rate. You want to keep control of the overnight rate (as you are not going to have the CB walk away from its role backing the payments system and acting as a LOLR) and *also* control rate of return of equity. You can't do that. You have to pick one interest rate and let the market price all the others. Not pick one interest rate, and say "I think the other one is also too high". But your whole post is basically "I think this other interest rate is also too high".

If the overnight interest rate is zero, then that's your interest rate. As soon as you change a second interest rate, you are now making a shift in relative prices and need to start buying up real assets.

2)

Moreover, once you add risk, a decline in the *nominal* return of equity adds volatility to those firms whose payout is partly in the future.

That's just simple arithmetic as per the example I gave. It is just duration effects.

With a lower nominal rate of return for equity, the market punishes firms whose earnings are re-invested and rewards firms whose earnings are in the present. In the limit of zero equity returns, there is zero re-investment and we are in some form of feudal society. Low nominal returns create more volatility of capital values and reduce welfare. That is also a real effect.

[Nick Rowe]

Suppose I reframed my policy. Instead of saying "the central bank should make the value of the stock market total return index rise at 7% per year, in terms of its money", let me instead say "the central bank should make the value of its money fall at 7% per year, in terms of the stock market total return index".

It's the same thing, of course. But rewording it may make it clearer.

The price of eggs in terms of money, which is just the reciprocal of the price of money in terms of eggs, doesn't just depend on the supply and demand for eggs. It depends just as much on the supply and demand for money. This is absolutely central to all monetary economics. (More strictly, one is the flipside of the other, rather than two different things.)

Let me now stick my neck out and conjecture that all disagreements between (say) MMTers and (say) monetarists are fundamentally the result of MMTers not getting this basic point. (That's almost certainly an exaggeration, but there must be some truth in it.)

[rsj]

nstead of saying "the central bank should make the value of the stock market total return index rise at 7% per year, in terms of its money", let me instead say "the central bank should make the value of its money fall at 7% per year, in terms of the stock market total return index".

OK, fine. You've walked away from the LOLR role and so all the banks fail. A bank wants to borrow money, but you say "sorry, the stock market is where we want it now, so no more money for you".

Oh, you forgot about the banks and the need to set overnight borrowing rates *as well* as rates for the stock market? How to set both?

That's the problem with the monetarists -- no banks! Add banks, and the quantity of money is now demand determined, with the central bank not supplying money, but supplying reserves to the banking system in support of a policy overnight interest rate.

One policy rate, which is the rate necessary to purchase and sell reserves overnight.

There is no more room to set any other price.

Any attempt to set two nominal prices are really fiscal policy in disguise.

[rsj]

FYI,

The MMTers also believe that the CB can set the entire term structure of rates to whatever they want -- Bill Mitchell said that repeatedly -- and it was just me and another guy in the back saying "no, you only set the short rate and everything else follows from arbitrage/preferences. And to the degree that you (temporarily) succeed in moving two rates, you are really doing fiscal policy and in a bad way as it corresponds to subsidies for investors".

The MMTers are a big fan of 'euthanizing the rentiers' with zero rates, and I was trying to argue that this actually enriches incumbent capital holders, causes bubbles, and deters investment.

But the other point is that the arbitrage conditions are not homogeneous due to duration effects. I think Keynes pointed this out, and Jan Kregel has written several papers about it -- it doesn't seem that you've responded to this point.

[Nick Rowe]

rsj: 1. "You can only control one interest rate."

Basically agreed. Let me be more precise and say that monetary policy can only target the time-path of one variable, and that variable must be a nominal variable. And I'm proposing (whimsically) that that one variable should be the stock market total return index. Of course the central bank can't set that time path at (say) 7% and at the same time choose any second rate of interest to be anything it wants. Just as the Bank of Canada can't do what it is doing now, which is target the time path of the CPI to grow at 2%, while at the same time doing anything it wants with any rate of interest. It must allow all other (nominal) rates of interest to fall in line with that 2% inflation target. Which is exactly what the Bank of Canada does. It must adjust the overnight rate "target" (allow the overnight rate "target" to adjust) if it wants to hit its CPI inflation target.

Again, this is well-understood, and all it goes to show is that the Bank of Canada's overnight rate "target" is not really a target at all (except for 6 weeks). The Bank of Canada's target is the 2% CPI inflation target, not the overnight rate. Switch "CPI" to "stock market total return index", and switch "2%" for "7%", and you've got my proposal.

Yes, acting as lender of last resort might conceivably force the Bank of Canada to have to temporarily deviate from its inflation target. It hasn't happened yet, but it might happen. The same would presumably be true if the Bank targeted the stock market index rather than the Consumer Price Index.

2. Risk. IIRC, the historical returns on the stock market have been around 7%, so I'm not talking about lowering the nominal return.

Plus, even if I were talking about lowering or raising the nominal return on the stock market, standard super-neutrality of money says this shouldn't affect the real variables like the volatility of real stock prices. Now, it's true that super-neutrality does not apply to (0% interest) currency itself. So it is conceivable that, if the central bank targeted the stock market total return index, the lower the nominal target rate of return, the greater might be the implied volatility of the equilibrium price level. I would need to think about that one, to be sure. But it doesn't sound right to me.

[Nick Rowe]

rsj: "The MMTers also believe that the CB can set the entire term structure of rates to whatever they want -- Bill Mitchell said that repeatedly -- and it was just me and another guy in the back saying "no, you only set the short rate and everything else follows from arbitrage/preferences. And to the degree that you (temporarily) succeed in moving two rates, you are really doing fiscal policy and in a bad way as it corresponds to subsidies for investors"."

My apologies for lumping you in there. You were right and they were wrong.

[K]

rsj,

On MMT:

Yeah. Given that they consider the role of banking central to their theory (rightfully so!), many of them display a remarkable level of ignorance/misunderstanding of finance. Fulweiler is one of a few notable exceptions.

On "duration effects":

I'm not totally sure I understand what you mean, but I'll take a shot at it. Imagine that money is simply the units of a large fund which holds a broad cross-section of stocks. The units are fully redeemable for a share of the underlying basket, and new units are created via are splitting at a rate of x%/year. The CB commits to continuing to do so forever. Nick says x=7. x could just as well be zero. I don't see anything unstable.

[J.V. Dubois]

Very interesting article. While it may be fun to play with these ideas, I think there are other considerations. So far monetary policy was used mainly for targeting "objective" variables. CPI basket of goods, unemployment even price of gold can be accepted as a measure widely accepted by public. Targeting stock prices of few hundreds of corporations, I am not so sure about this. Just getting you into this selected club would mean that the price of your shares will have additional monetary component. If for instance there would be some Enron-like situation that top 10 world companies in terms of market capitalization were involved in a fraud, your policy would basically mean that loses would be partly shared via inflation.

There are many other forms of private equity, mostly shares and partnerships in non-publicly owned corporations and limited liability companies or even personal wealth (mostly real-estates). "Equity premium" for these forms of companies would be largely left out of your policy. But then I am not that good in this whole equity premium debate and it may very well be so that all these things were already considered.

[JP Koning]

Nick, like Bernhard above, I'm having trouble getting my head around how welfare increases in your example. It seems to me that you'd have increased investments in stocks, but decreased investments in bonds. But maybe I am missing something. If GDP is not the relevant metric, what element of the National Accounts would improve if the equity premium was abolished? Would it show up in the wealth accounts, for instance national worth?

[K]

Nick,

I'm also having trouble understanding the welfare benefits. Or rather, I think there are big welfare benefits, but I don't see where you've made the case for them, or how you arrive at your 5%/year benefit. It's just a nominal change, after all, and as I pointed out above, the risk premium, in the standard form, is relative to risk-free *real* returns. You haven't made a case that you've changed that.

The reason *I* think it's beneficial is that I think people do, in fact, have an extreme form of nominal fixation, whatever the CB is targeting. Except under hyper-inflation scenarios, people count their wealth in dollars. E.g. they prefer nominal over real return bonds, they refuse to spend "capital," even under deflation, but happily spend coupons during inflation. Focusing this money illusion on the consumer basket is a very bad idea because the economy cannot *in aggregate* guarantee a particular level of future consumption. In times of panic, allowing people collectively to channel their demand for nominal safety into an impossible consumption guarantee, at the expense of the capital investment which is the only thing that can provide that future consumption is extremely dangerous. If money illusion is strong you don't want to direct it at aggravating the paradox of thrift.

If instead, we channel our nominal fixation at a broad class of capital assets, then a flight to money will *not* be a flight from investment. Perhaps there will be a flight from consumption, but it will be much less likely that we will simultaneously attempt to stop both consumption *and* investment. But I don't believe in targeting. Targeting is necessary for the CPI because you can't store the consumption basket (it's a unit of a flow), which is part of what makes it an impossible promise. But, like gold, you *can* store capital assets. So is there any reason why you want to "target" rather than "back"?

[123]

Nick: "But the main question, and the one I ducked, is "how volatile would NGDP be?" and would that volatility be big enough to outweigh the benefits of (say) 5% of GDP?"

Yes, NGDP would be too volatile. And I am afraid the excess volatility of NGDP might cause the real stockmarket risk premium to increase.

And it is the real stockmarket risk premium that we want to reduce, not the nominal one.

" Let me be more precise and say that monetary policy can only target the time-path of one variable, and that variable must be a nominal variable."
I disagree. Monetary policy has two targets: a time path of some nominal variable and a real risk premium that is related to that nominal variable. As I wrote in the email, I support NGDP level targeting and I support lowering the price of NGDP options to the extent it is possible to lower it.

While it is possible to peg VIX at almost zero, it is not possible peg the real VIX at zero (real VIX would be implied volatility of inflation adjusted S&P 500). Good monetary policy can reduce the NGDP risk premium, and VIX would be reduced via arbitrage...

[rsj]

K,

Think of it this way. Your basket has two firms. One pays out 100% of earnings as dividends. The other pays out 50% of its earnings as dividends. They are each 1/2 of the basket. When the central bank lowers or raises the average return on equity, the relative market cap of these two firms will shift. If the nominal return falls, for example, the firm that pays out all of its earnings will be worth more than the firm that re-invests 1/2.

You have just changed relative prices, and created a real effect.

To avoid changing relative prices, the central bank would need to control the dividend payout policies for all firms. But these are firm specific and related to real investment opportunities for each firm -- one firm may need/want to invest more while another firm does not.

Btw, the same is *also* true for the consumption "basket". The mix of auto/food changes when inflation changes and nominal rates change. You assume that duration problem away in the optimization problem by assuming that everyone is purchasing auto-rental services instead of buying cars. But people take delivery of auto-rental services by purchasing cars, and so duration effects are important here, too.

It is only by some fast-talking that you can try to make the optimization or arbitrage equations homogenous. They are not homogenous because of duration effects.

[dlr]

rsj, that's not right. "duration" effects from changes in expected nominal price changes require fixed nominal earnings streams. there is no real price change here except the one you are incorrectly assuming between the two firms. increases in nominal prices also cet. par. mean increases in nominal earnings and the nominal discount rate to no net effect. this is also true for the auto/food basket. nominal rates don't create duration effects on real consumer durables like cars versus food. what matters is real rates, and this happens only because of the liquidity effect -- something that certainly doesn't always even show up to dominate the fisher effect when nominal rates increase.

it is only when you talk about whether this policy impacts real rates that you can talk about duration -- but even then you are wrong about dividends and duration. dividends do not shorten duration -- the only thing they do is decreased agency risk -- otherwise they do not represent fixed reinvestment commitments.. imagine a company which owned nothing but fixed rate treasuries. it had two share classes. one share class paid the coupon out. the other was called a "zero" because whenever the firm received a coupon it would be reinvested into TIPS (at market). these share class do not act any differently to changes in either nominal or real rates. that is your dividend paying versus not dividend paying firm economy wide. reinvestment of unpaid dividends comes amid market rates not some preset real rate like it does with a normal zero. buying a consumer durable is kind of like buying a zero in that its future return on marginal investment is fixed so its value does suffer relative to rental prices when real rates (but not nominal rates) change. certainly for some firms agency risk (the belief that management will fail to adjust to new required returns in their reinvestment models) but for the market as a whole that is just an MM violation via anecdote (i.e. HP making bad acquisitions). expected marginal corporate returns on equity as a whole are going to move with the real rate.

[Lord]

Stocks have a return. That return does not just disappear into the aether. About half of those returns are consumed, either through more investment to grow the company or through more consumption by its holders, both of which are in gdp. The other half goes into increasing stock prices via buybacks and reinvestment or retained earnings which is akin to disappearing into the aether. So half of this return is already counted in gdp, (and I would theorize the return of stocks to be expressly twice the growth rate of gdp to allow for growth, investment, consumption, and gain). It is a little different if companies merely piled up their earnings in cash, like Apple did until recently, but that more the exception than the rule and the coffers would runneth over otherwise.

[Phil Koop]

Completely off-topic: I just noticed that Ryan Avent has written a post that reads like it was written by you, Nick! (http://www.economist.com/blogs/freeexchange/2012/06/business-cycles.) Separated at birth or assimilated by the Borg?

[rsj]

dlr,

Do you 1) disagree with the math -- in which case, pls point out the error.
2) Disagree with the assumptions -- constant dividend yields. In which case, do you really believe the conclusions are reversed if the growth rates follow a curve? It makes no difference -- I was just trying to keep things simple.

But if you think you have a result in that direction, rather than philosophy, let's see the math.

I will help you. Pick whatever inflation rate you like. Pick whatever utility function you like (except quadratic -- it must be strictly concave).

For each dollar invested by the firm there is a probability that the dollar will earn a profit and a probability that it will not. If you object to constant probabilities each period, then make up your own distribution as you see fit. It will make the math more complicated, but it wont change the conclusion.

Firm A pays out 50% of its earnings and re-invests 50%, growing the capital stock.
Firm B pays out all profits as it receives them on a constant capital stock.

The market cost of capital is 4% (nominal). What is the share of market cap of both of these firms?

Now, the central bank succeeds in reducing that to 2% (nominal). What is the share of market cap of these firms? Is it identical?

In my example, it wasn't. A change in the nominal rate forced a change in the relative values of the firm. The *magnitude* of the change depends on utility and inflation. But the *sign* depends only on nominal values. Firm B is more valuable vis-a-vis Firm A.

If you think this is purely an artifact of assuming constant nominal growth rates, then let's your calculations for the above problem. It should be easy to determine whether nominal rates make a difference or not.

[K]

rsj:

I'm having a hard time getting my head around your assumptions. To me, in order to compute the optimal dividend payout and leverage ratio, you need a Merton-type capital structure model with parameters like asset value and volatility, expected losses on hitting the debt default boundary, and tax efficiency of interest vs dividend. You also need an interest rate term structure model.

I see two principal benefits of the proposed system. First, by aligning returns on real and nominal investments, it removes, in a systemic way, the mismatch between corporate assets and nominal liabilities. I.e. it ought to reduce systemic default risk. This is a direct result of the redefinition of the unit of account. Since bankruptcy costs are a dead weight loss in the economy, this is a significant gain. Secondly, I think it stabilizes demand by changing the optics of investment risk. If the numeraire *is* the market, then the market is no longer risky in nominal terms. This could potentially significantly reduce risk premia.

The question of what happens to optimal dividend policy under such a scenario is quite complicated, since there are obvious implications for asset volatility and default risk, which makes it extremely hard for me to reverse engineer the behaviour and intentions of your two companies. There is too much I don't know about them and their motivations for me to understand why your dividend assumptions reflect optimal behaviour, both before and after the change of the unit of account.

[dlr]

rsj,

If you think this is purely an artifact of assuming constant nominal growth rates, then let's your calculations for the above problem. It should be easy to determine whether nominal rates make a difference or not.

The problem is easy if it's well specified and doesn't confound real and nominal rates. Pick (randomly) expected inflation of 2% . Neither firm's assets have excess systematic risk so their risk adjusted real cost of equity is 2% (4% "nominal market cost of equity capital" less expected inflation). Firm B then is expected to merely maintain its real dividends by paying out 100% of earnings into perpetuity and nominal dividend growth is 2% with inflation. It is worth 50X earnings and will provide a 2% real return. Firm A is also worth 50X earnings. It will reinvest half its earnings, which will have an expected real return on equity of 2%. Firm A nominal dividend growth will be 4% and with 50% reinvestment and a 4% nominal cost of capital it is also worth 50X earnings (or 100X dividends).

Now, the central bank succeeds in reducing that to 2% (nominal). What is the share of market cap of these firms? Is it identical?

You haven't specified whether this rate reduction is also real. First, say it is only nominal and the Fed merely eliminates expected inflation leaving real risk adjusted market rates unchanged at 2%. Expected nominal dividend growth drops to 0% for Firm B and 2% for firm A, but nominal discount rates drop by the same amount. Both relative and absolute value are unaffected. Alternatively put, Firm B is still expected to grow real dividends at 0% discounted at 2% (real), while Firm A is still expected to grow real dividends at 2%, reinvesting 50%, and discounted at 2% (real). They are both worth 50X earnings.

Let's say it is instead half nominal half real, i.e. the 2% market rate now reflects 1% expected inflation and 1% real rates. Now the existing assets are worth more, because the price of either risk or time has declined. Firm B is now worth 100X earnings and dividends, since its produces the same flat real dividend stream but discounted at 1% real (in nominal terms, dividend growth is now 1% and the nominal discount rate is 2%). Firm A is also worth 100 times earnings (200X dividends). Its expected real return on marginal investment has declined to 1% (remember, this is a microcosm for the corporate sector, the risk adjusted real cost of capital is the expected returns on capital -- we can add excess systematic risk but it would serve no purpose). In nominal terms, dividend growth is now 3% for Firm A and reinvestment is still 50%.

The only way you can conclude that "Firm B is more valuable vis-a-vis Firm A" is if you make some strange assumption about expected real marginal returns on reinvestment of firm A coming below the risk adjusted cost of capital. That is fine for anecdotes you mention like Adobe or HP but isn't sensible on average or overall. Think of the many examples over time of companies with two share classes, only one of which paid a dividend or paid differential dividends, that doesn't support the valuation discrepancy or reactions to changes in nominal interest rates you think should exist.

[Nick Rowe]

I think I now get rsj's point about risk. Let me restate it my way:

If I'm right about my "proposed" monetary policy eliminating the equity premium (assume I am right), then a likely consequence is that the real yield on stocks will fall. If so, that would mean that given changes in (e.g.) the growth rate of real earnings would cause larger percentage changes in stock prices, given the formula: P=E/(r-g) , where P is price of stocks, E is current earnings, r is equilibrium real rate of return on holding stocks, and g is expected real growth rate of earnings.

That's correct. Is that an argument against the policy? Not so sure.

Phil: thanks, yes, I noticed that RA post. Sounds like he agrees with my old "concrete steppes" post.

Lord: OK, but that's a purely demand side theory of GDP. In the long run, I would look at supply side theories of (real) GDP.

123: "Yes, NGDP would be too volatile. And I am afraid the excess volatility of NGDP might cause the real stockmarket risk premium to increase."

That's certainly possible. It might. I think it would depend on parameter values. Yep, monetary policy cannot peg real VIX.

JP: "If GDP is not the relevant metric, what element of the National Accounts would improve if the equity premium was abolished?"

Ah, the tyranny of accounting!

Assume the demand for canoes was perfectly inelastic (one new canoe per Canadian per decade, regardless of anything). And assume the supply of canoes was perfectly elastic (horizontal MC curve). Then if I waved a magic wand and made canoes twice as much fun, all canoists would be better off. But nothing would change in the national accounts, either income or wealth, because neither the quantity nor price of canoes will change. But the increase in welfare would equal the amount of GDP spent on canoes (even though that is unchanged), because they are now twice as much fun.

K: "I'm also having trouble understanding the welfare benefits. Or rather, I think there are big welfare benefits, but I don't see where you've made the case for them, or how you arrive at your 5%/year benefit."

I haven't really made a *watertight* case for them. I'm just assuming that people would get as more "fun" out of holding stocks, and as much "fun" as they now get out of holding bonds. See my canoe example in response to JP.

[123]

Nick: "Yep, monetary policy cannot peg real VIX."
And monetary policy cannot peg the volatility of NGDP, but it can (and should) sharply reduce it from current levels.

[rsj]

Nick, exactly, but the increase in volatilty means that even though the fed has reduced the risk premium -- e.g. the cost of bearing a unit of volatility -- it may have actually increased the total risk and therefore the total spread between riskless bonds and equities, because there is more volatility in capital prices now.

Moreover, as some stocks are more dependent on future growth rates to justify their market cap, those stocks would experience more volatility and be discounted more. So the relative price of growth to value stocks changes, causing real effects and possibly real distortions.

And I think there are other distortions, but am in the office now :)

[rsj]

Dlr,

You are assuming perfect certainty of future earnings, in which case there is no rationale for a risk premium. But if you believe that $1 of capital pays out 1% with 50% probability and 2% with 50% probability, then then the P/E multiple of the firm that re-invests half its earnings to purchase capital and pays half out will be different from the P/E multiple of the firm that sits on the same capital stock and pays out all earnings as dividends each period.

The ratio will change as the *nominal* short interest rate changes, even assuming that the above payouts are real. The relative price is not independent of the nominal short rate.

Moreover, you will not get high P/E multiples in the presence of risk, with any reasonable utility function, at some point lower nominal rates just translate into so much more volatility that *any* non-zero risk premium will result in P/E multiples much lower than you believe they should be in an environment of perfect certainty. There is a nominal return below which you cannot push equity, unless you can credibly commit to buying it all.

[JP Koning]

"I'm just assuming that people would get as more "fun" out of holding stocks, and as much "fun" as they now get out of holding bonds."

But wouldn't holding bonds get less "fun"?

Just prior to adopting a policy that targets 7% equity returns, bonds are willingly held by investors because of their unique range of features (low risk, senior to equity etc), features that stocks don't have. After the policy is adopted, these features are no longer unique to bonds. In order for investors to willingly hold existing bonds, don't bond prices have to fall to some lower level, thus providing a better return to compensate for features lost?

[K]

Nick: "That's correct."

I don't think so. You're assuming profit growth varies but the real rate stays unchanged. To first approximation, I'd assume they move in tandem. I.e. changes in the outlook for future growth are completely offset by changes in the real rate curve, so asset volatility is *not* driven by systemic changes in outlook.

[rsj]

The profit growth rate of each firm varies in lock-step with the real rate?

What if the profit growth of two different firms aren't identical?

[K]

rsj,

I don't see that relative prices matter to this conversation. I'm saying the growth rate on the income generated by the market portfolio moves in parallel with the real rate (risk premium being constant). And I'm not talking about just stocks. Stocks, bonds, all capital assets, are ultimately claims on the fraction of future output that doesn't go to labour or government. That's why I use the word income, rather than earnings or dividends.

With the current targeting regime, I believe that almost all *systemic* asset volatility is a result of expectations of failure of the target. In particular, it is expected that there will be periods of excess real rates (relative to income growth rates) at the ZLB resulting in demand deficiency. It is expectations of divergences between growth and real rates that cause asset volatility, *not* expectations of growth alone. A credible target will not permit divergences and will therefore stabilize asset prices. Unfortunately no targets, unless explicitly backed by the thing being targeted, can always work. That's why I advocate *backing* by capital assets.

[Nick Rowe]

JP: "But wouldn't holding bonds get less "fun"?"

If I waved my magic wand and canoes became more fun, that doesn't make kayaks less fun in absolute terms, only relative to canoes. So the price of kayaks might indeed fall (depending on the elasticity of supply for kayaks), but that fall in price is simply a change in the distribution of income between kayak sellers (worse off) and kayak buyers (better off). It's not a true negative externality from the benefits of me waving my wand to make canoes more fun. The gains and losses in the kayak market cancel out.

[rsj]

K,

Assume *zero* systemic risk. Everyone knows with completely certainty that the market return is 1% in every single period.

But each individual firm is subject to massive swings in earnings, all of which cancel out. Firms are driven out of business and their revenues are absorbed by incumbents, etc.

The investor can earn a premium by diversifying and getting more than the risk free rate, because the individual business owner (or manager), is exposed to the returns of the firm, by definition of their job role. To not be exposed to the earnings would be to have performance disassociated from pay. Even ignoring things like stock options and stock ownership, your reputation and future earnings are determined by your success or failure at managing a firm.

The one deciding whether to spend a billion to open a new chip fab in the case of Intel, or $100 to add a new counter to the coffee shop in the case of a small business, will still be risk-averse. If there is a 50% chance of 0 earnings and a 50% chance of earning $2, then they will not pull the trigger and invest. They will always prefer to just buy the 1% bond (which is what businesses are doing now).

And therefore each individual firm's market cap will be discounted at a higher rate than the risk free rate regardless of what the government does. And if that is true of each individual firm, it will be true of the market as a whole.

You are always going to have a risk premium, even if market risk is always zero, and stock indexes can be predicted with perfect certainty. If you really can't stand this arbitrage, then start levying taxes on investors rather than trying to buy up all the capital.

[K]

rsj,

"therefore each individual firm's market cap will be discounted at a higher rate than the risk free rate regardless of what the government does. And if that is true of each individual firm, it will be true of the market as a whole."

How the managers of a firm choose to make investment decisions has no impact on the market's discounting of the firm's liabilities. The market discounts securities based on the probability distribution of future income, *whatever* the decision making process that generates that income. The risk preferences of the managers are irrelevant to securities pricing.

"You are always going to have a risk premium, even if market risk is always zero, and stock indexes can be predicted with perfect certainty"

No. If both assets are risk free, they both earn the risk free rate. Two securities. Both pay $1 in a year with absolute certainty. One yields more than the other. Are you saying you'd buy the more expensive one?

[rsj]

How the managers of a firm choose to make investment decisions has no impact on the market's discounting of the firm's liabilities. The market discounts securities based on the probability distribution of future income, *whatever* the decision making process that generates that income. The risk preferences of the managers are irrelevant to securities pricing.

I would buy the one the one that yields more!

But riddle me this: if all the decision makers (for real investment, not purchases of paper claims on profits) are risk averse, which they should be given their outsized stake in the success of the firm, and demand a premium of x% over the risk-free rate, then you will have a discrepancy between return on invested capital and the risk-free rate, which is the case. Now if you bid up the share price of firms so that the return on equity is the risk-free rate, where the former is always greater than the latter, then you are in a situation in which the market price of a firm is always greater than the book value of the firm.

So there are two opposing arbitrage relations, and which of them will win out? I can think of a lot of reasons why neither arbitrage is real -- e.g. you don't know with certainty that $1 in lost earnings by one firm will result in $1 of increased earnings by another. Book value is not liquidation value, and privately held equity is greater in value than publicly traded equity, etc.

But the argument that there should be an equity premium of zero is equivalent to saying that the first arbitrage should always dominate the second, or that book value should be permanently less than the market value of a firm in all cases, provided that preferences are unchanged, and that also doesn't make sense.

[ptuomov]

The opposite of tranching comment.

Tranching is one of the oldest tricks in finance. Let's take a company that is owned by a person, who just owns the equity and there's no debt. Then a banker comes and tells the person to borrow money and pay himself a dividend. Surprise! We've just tranched the company to two tranches, the safe debt tranche and the more risky (now levered) equity tranche.

Most of the security design is about tranching some cash flow rights in a way that creates very safe (looking) securities and riskier securities. The market, for many reasons, pays more for the tranches than the whole and "value" is created.

Your scheme is "untranching" the economy, turning low risk government bonds and high risk stocks into medium risk untranched securities.

[K]

rsj,

"So there are two opposing arbitrage relations, and which of them will win out?"

There is no contradiction in arbitraging both since both the relative returns of risky securities vs riskless ones can change, and so can the relative returns of risky securities vs corporate assets (Tobin's Q).

The more difficult arbitrage, Tobin's Q, consists of two parts:  

1. There's the real economic difference between the market price of liabilities and the cost of assembling the assets. That part *is* fairly arbitrageable, which is the job that's done by MBOs, LBOs, merchant banks and some hedge funds. I remember, for example, hedge funds building sheet rock factories while short selling the stocks of sheet rock companies when there were building supply shortages during the housing bubble.

2. Then there's the other part, which you are talking about, which is simply the marginal cost of skilled management. This cost, which depends both on the supply of managers and their aversion to risk, can be very high. Having to pay them is what closes the gap between the low return on the liabilities and the high return on assets (not including management). To an arbitrageur, management needs to be thought of as just another (possibly expensive) asset.

But either way, arbitraging tobin's Q should not have any impact on the returns of liquid securities which are determined by investor preferences. It will just reduce the rents that flow to management and reduce the returns to the non-management corporate assets.

[rsj]

Thanks, K, but I don't understand the argument in part 2. If managers are risk averse when making individual investments because their own wealth is disproportionally exposed to the returns of the firms that they are running -- including their future labor income -- how is paying them more going to help? It seems it would hurt, as you are increasing the stakes.

The only thing that would help would be to hedge them from the consequences of their investment decisions, that is, from whether they perform well or not in their jobs. You would need to include assurances that whether they are deemed to be skilled or unskilled is independent of the outcomes of their investment choices. I guess you could pay them solely on their educational credentials and years of experience. That wouldn't be much different than having the government buy up all the capital and appoint unionized civil servants to make strategic decisions about which projects the firm undertakes.

[K]

"It seems it would hurt, as you are increasing the stakes."

I don't think so. You're only increasing the upside. The downside is zero (in case of default). If I pay you more in some states of the universe and never pay you less that's an unequivocal good (it's actually a definition of you doing arbitrage).

[rsj]

..and the point being that in the long run -- if managers just don't undertake a return unless it yields the risk-free rate +P%, on average -- then who gets the P%? The managers can pocket it, equity investors in the secondary market can pocket it, or the VCs that sell firms to the secondary market can pocket it.

I think, in the real world, none of these arbitrages are perfect and there is a tug of war as to who earns the rent. I think the case can be made that in the last 20 years or so, it was the insiders (e.g. managers) who pocketed the bulk of the premium, as equity investors have not been doing too well. VCs have been doing OK.

[rsj]

K, I missed your post.

OK, basically you are saying that "paying them more" means letting them earn the entire rent, or a large portion of it.

[rsj]

I don't think so. You're only increasing the upside. The downside is zero (in case of default). If I pay you more in some states of the universe and never pay you less that's an unequivocal good (it's actually a definition of you doing arbitrage).

Hmm, if you don't need to eat, or you don't mind your next job being flipping burgers, then it is an unmitigated good. Others would would use their current salary/position as a baseline of zero, and think of how they can marginally improve or reduce from their baseline when evaluating risk.

[rsj]

From my experience with management, most of them are definitely risk averse and interested in protecting their position. They would never undertake a risky project whose expected payout is just the risk free rate. And that is certainly true of small businessmen.

In the world of rock-star CEOs, I think they are driven by non-monetary considerations to a large degree. It could be legacy, ambition, whatever. It's hard to say in that case. Nevertheless, they wont undertake a risky project with the expected payoff only the risk-free rate. They would rather buy a bond.

Finance, of course, is a different world. There, sure, you are picking up nickels in front of a steam roller to get a 1% spread.

But you don't do that in the non-financial world.