Real Term Premia in Life-Cycle Portfolio Choice

Sammanfattning

This paper explores how assumptions about real term premia and the resulting time- varying expected excess returns in bonds alter optimal life-cycle portfolio decisions. I construct 10-year real-yield specifications that progressively scale the real term premium from a zero-premium expectations-hypothesis lower bound to a Fisher-adjusted upper bound. These yields are derived by deflating nominal rates with an AR(1) inflation forecast and assuming a zero inflation-risk pre- mium. For each yield specification, I estimate an annual VAR(1) over a 1920-2011 U.S. sample for equity and 10-year real bond returns, incorporating yield-curve and valuation-based predictors. I then embed these estimated return dynamics into a life-cycle portfolio-choice model with risky labor income, Social Security, mortality risk, and a bequest motive. This setup provides a struc- tured counterfactual: how much does the introduction of real term premia shift life-cycle portfolio decisions? I find that moving from the expectations-hypothesis baseline to the Fisher-adjusted yield raises optimal long-bond demand from 8% to 75% at age 45. Rather than substituting for equities, this larger real bond demand is financed by taking short positions in real bills. Decomposing this increase reveals that the majority of it is driven not by the unconditionally higher bond premium, but by intertemporal hedging against fluctuations in future real bond risk premia. Finally, life-cycle forces actively shape the overall level of this bond exposure, causing mid-life and late-retirement households to hold more 10-year real bonds than an infinite-horizon investor