differ based to their respective dividend yields? The correct (and intuitive) answer is, they shouldn't. In fact, there is an equilibrium condition that determines long-term dividend yield. In the long run, dividend yields should equal earnings growth rates. Why? If dividend yields were higher than real earnings growth, the transfers from the corporate sector would exceed its economic growth and corporations could not remain a constant proportion of the economy. The opposite is clearly also the case. Therefore, equilibrium dividend yields equal long-term earnings growth. All that remains to determine an expected real return on equity, then, is long-term real economic growth. Since 1947, compound real annual GDP growth averaged 3.4 percent per year.4 Taking a slight haircut from this to reflect survivorship, it seems reasonable to expect future real economic growth in the 2.5 to 3.0 percent range. This implies a real equity return of 5 percent to 6 percent. Taking the midpoint of this range, along with Ibbotson and Chen's real risk-free return estimate of 2 percent, yields an ERP estimate of 3.5 percent. The decomposition outlined above offers a useful tool for understanding the current debate about the level of the ERP. Since most researchers would agree on the basic structure of the decomposition, the debate can be centered on both the levels of each component (e.g., the real economic growth rate) and the underlying economic fundamentals. For example, Arnott and Bernstein argue forcefully that we are in a bubble, and that P/E ratios will decline substantially from their recent levels. To answer this question, we can reasonably ask what the equilibrium relationships are between equity valuations and the real economy. Thus, in addition to exploring the historical record, we should also include a theoretical understanding of the equity premium. EQUILIBRIUM ESTIMATES OF THE EQUITY RISK PREMIUM Gauging the equity risk premium from the demand side requires a model for investor preferences. Two early academics studying the ERP, Mehra and Prescott (1985), report a rather surprising estimate of 0.4 percent for the ERP! This is so low relative to realized equity returns that they called their finding "the equity premium puzzle." Their results spawned a generation of new academic research attempting to rationalize their findings, some of which we describe in this chapter. Mehra and Prescott's work is an application of a standard dynamic macro-economic model generalized to allow for asset pricing (see, for instance, Lucas 1978). At its core, this model makes the commonsense assumption that what investors really care about is not investing per se, but rather the consumption stream that such investing will support. That is, an investor's well-being (or utility) depends on the path of current and future consumption. Investors are willing to defer current consumption and invest only if they believe that the return on investing will generate sufficient future consumption to make them feel better off. 4According to the U.S. Federal Reserve Board, 1947-2001, Flow of Funds Accounts of the United States, Washington, DC: Federal Reserve Board.