Risk parity is the new "new thing" in portfolio allocation

Risk parity has become the new thing for asset management. Do not allocate your risk by dollar allocations, but by the contribution to risk. A dollar allocation that is 60/40 between stocks and bonds is not the same as having a 60/40 risk allocation. Since stocks are riskier than bonds, the risk allocation is higher to equities than is the case for a dollar allocation. Under this type of model, there is the assumption that diversification is the only free lunch that still exists and risk management is more important than dollar exposure management. The simplest case is that you equalize the risk across asset classes and then lever to set the overall volatility target. 

JP Morgan will call this approach of looking at risk parity more uncertainty management. When there is more uncertainty on the parameters of return, volatility and correlation there is a reason to equalize risk exposures. Put another way, since we have so little faith in the expected returns determined historically and used for optimization, a better approach would be to equalize the returns and focus on risk or other aspects of asset allocation. A simple early version of this type of work is the results showing that 1/n allocation is not a bad way to solve error problems in expected return or volatility. If the Sharpe ratio for asset classes is closer, then there will be more advantage with holding the risk parity portfolio. This makes sense. It is worth differentiating asset classes if there is greater differentiation in the asset classes. 

In some sense the idea of holding a 60/40 stock-bond portfolio is one way to avoid the errors associated with return forecasting. Unfortunately, if you have a 60/40 portfolio you place the majority of your risk in equities which means that you are still dependent on stock return streams. The idea that you would look at risk as the determining allocation factors is a further refinement of avoiding traps based on return forecasts. 

There is too much error in forecasting returns so a good approach is to change the focus of analysis. Assuming that Sharpe ratios are the same across asset classes in the long-run is not really a bad  assumption. Equities could return more, but the return per unit of risk is the same across asset classes. Risk parity can be thought of as assuming that correlations are the same and Sharpe ratios are equalized; however, there can be further work to adjust these Sharp ratios and tilt the allocations of return to risk. If you move to differences in Sharpe as opposed to equal Sharpe ratios, some would argue that this is more akin to risk budgeting as opposed risk parity. Risk parity would be a special case of risk budgeting. Looking at differences in Sharpe is the same as looking at differences or accounting for differences in risk premia. 

Implementing risk parity is harder than what you may think. One of the issues is that the distribution of returns is not normal so hat using some simple meansure like volatility alone may cause a manager to choose more risk than he would like. Still, risk parity is an important addition to any discussion on risk management.