A risk premium is the additional return demanded by an investor in exchange for investing in a risky asset. This premium is measured as the total expected return minus the return on a risk free asset, such as a treasury bill issued by the United States government. Logically, an investor should demand a large risk premium on riskier investments. This is why bonds that have been rated close to or at junk status trade at very high effective interest rates. Conversely, securities issued by large, stable corporations can usually be sold at very low interest rates, since investors are quite sure that they will be paid; thus, there is a low risk premium.
Risk premiums are not theoretical - they can be quite justified. For example, a credit rating agency assigns a junk rating to a government bond because its analysts believe there is a notable probability that the issuing entity will repudiate its obligation. Investors then buy the bond at an effective interest rate of 25%, thinking that they will earn an exceptional return on their investment. However, the rating agency proves to be correct, and the bonds are repudiated. Investors then sell their shares to speculators for pennies on the dollar. In short, the risk premium was justified, because there was a capital loss.
Wealthier investors may have such large cash reserves that they can afford to speculate on riskier investments that carry very high risk premiums. By doing so, they are placing a bet that a few of the investments will pay off, resulting in rich returns.
An inherent problem with the risk premium is that a demand for an excessively high premium places an undue financial strain on the entity soliciting the investment. If the financial burden is too high, paying a return on the invested funds no longer makes sense for the entity. Consequently, there is a natural cap on the amount of the risk premium, over which no investments should be available.