One of the things that annoys me about the concept of the equity premium is that it is an academic creation that does not grasp the structures of the markets. Send the academics to be bond and equity portfolio managers for a time, and maybe we would get a better theory than Modern Portfolio Theory [MPT].
Here is the first thing that is wrong with MPT - it doesn't understand the bond market. The best estimate of what bonds will return over time is the current yield less expected losses from defaults and optionality. Hold a bond to its maturity, and the standard deviation of returns is low, over the full time horizon.
Thinking about bonds in the current environment, virtually nothing is earned with high-quality short-dated debt. The yield curve is still relatively steep, as people expect the economy and lending to pick up.
Think for a moment. what is a longer asset, a corporate bond, or the stock of the same company? The stock is the longer asset, because the cash flows of the business in question potentially stretch far longer than the maturity of the corporate debt, at least in most cases.
Also think, in a bad scenario, where insolvency is possible, who has the better claim: the equity or the unsecured debt? The unsecured debt, of course.
Longer assets in general possess more risk and should carry higher yields to induce people to take those risks. Inverted yield curves are exceptions. Also in general, longer corporate bonds have higher spreads over Treasuries most of the time, than shorter corporate bonds.
The one significant advantage that equities have over corporate bonds is that of control. Increases in earnings go to the stockholders. Buyouts go to the stockholders. Bondholders get paid off at best.
That said, in the losing scenario, bondholders get back 40% of par on average, while stockholders get little if anything.
I believe that the equity of a company needs to be priced to return more than the longest unsecured debt or preferred stock of the company.
Thus when I think about MPT, I think they are positing an asset-liability mismatch, comparing T-bills versus a long asset, common stocks. The comparison should be broken down into several spreads:
- T-bills vs T-notes/bonds of the longest maturity issued by companies like them.
- Corporate bond yields minus Treasury yields at the same maturity.
- The earnings yield of the stock minus the corporate bond yield.
This takes apart the seemingly simple MPT calculation, revealing the complexity within, helping to explain why beta doesn't work. It embeds an asset-liability mismatch. Stocks are long term, T-bills are not. There is no reason why their returns should be considered together, without a model of yield curve spreads, corporate spreads, and equity financing spreads.
That's a sketch of the correct model, now who wants to try to build it out?