Global Tactical Asset Allocation

A pension fund's investment committee sets a policy mix once a year or so: say, 50% global equities, 35% bonds, 10% real assets, 5% cash. That mix is chosen to match the fund's liabilities and risk tolerance over a decade, and it deliberately ignores this month's headlines. Global tactical asset allocation (GTAA) sits on top of that policy mix and asks a narrower question: given what valuations and trends look like right now, should the fund be a few percentage points overweight equities and underweight bonds this quarter — not abandon the policy mix, just lean against it slightly, using the same signals (value, momentum, carry) that work within a single asset class, applied instead to the choice between asset classes and countries.

How a tilt actually gets sized

A GTAA overlay typically starts from the policy weights and applies bounded tilts, often ±5 to ±10 percentage points per asset class, driven by a blend of signals: relative valuation (is this equity market cheap versus its own history and versus bonds?), trend (has this market been rising or falling?), and macro indicators (yield curve shape, credit spreads, PMI surprises). The tilts are combined into a score, the score is mapped to a position size, and the position is capped so that even a maximally confident signal can't turn a 50/35/10/5 policy portfolio into something unrecognizable — the overlay is meant to add a percentage point or two of return a year, not to replace the policy decision.

Worked example. Suppose the policy mix is 50% equities / 35% bonds / 15% cash, and a GTAA model observes: equity valuations (cyclically-adjusted earnings yield) sit one standard deviation cheap relative to bonds, and 12-month equity momentum is positive. Both signals point the same direction, so the model tilts equities to 57% (+7pp, near the cap), bonds to 30% (-5pp), cash to 13% (-2pp). If equities then outperform bonds by 6% over the following year, the tilt contributes roughly $0.07 \times 6\% = 0.42$ percentage points of extra portfolio return relative to holding the static policy mix — a modest but real number, consistent with the "lean, don't bet the fund" design of the strategy.

What this means in practice

GTAA overlays are run by macro hedge funds, some pension funds' internal teams, and multi-asset mutual funds, usually as a distinct, risk-budgeted sleeve of an otherwise passive or strategically-allocated portfolio — the tilt's tracking error against the policy benchmark is itself a monitored, capped number, often 1-3% annualized, so the overlay can be judged on its own risk-adjusted return separate from the underlying portfolio's.

What erodes it

The signals GTAA relies on — cross-asset value and momentum — are the same crowded factors discussed elsewhere (see Style Premia Across Asset Classes), so GTAA inherits their crowding risk. It also faces a harder problem than single-asset factor investing: valuation comparisons across asset classes (is equity's earnings yield "cheap" relative to a bond's yield?) depend on assumptions about the equity risk premium that are themselves unstable, so the tactical signal can be confidently wrong for extended periods, as many GTAA overlays were in the trending, low-rate 2010s when a value-tilted overlay repeatedly underweighted equities that kept rising.