Defensive assets: the map and the roles

Everything in the portfolio that is not the equity engine has one job: defense. But defense against what?

"Bonds equal safety" skips that question, and it is the first one to ask. A withdrawal plan has several distinct enemies: the short crash, the long bear regime, persistent inflation, deflation, a crisis of confidence in money. No asset defends against all of them. Each defender has a specialty, a carry cost and its own conditions of failure.

This article lays out the whole map. First the spec a real defensive asset has to meet. Then the candidates one by one (cash, short and long bonds, linkers, gold, managed futures, commodities, real estate, safe-haven currencies), each with the regime it serves, what it costs and the verdict. Then the gallery of false defensives, which include some of the best-sold traps on the market (dividend stocks, high yield, structured products). And finally how to put the pieces together, because the last lesson is that the defense itself gets diversified.

The major building blocks each have their own article (Bonds in retirement: types, duration, the exact job, Inflation-linked bonds: the only contract written in real terms, Gold in a retirement portfolio, Managed futures and trend following: the diversification that works in a crisis). So do the more exotic defenses (Long volatility and tail hedging: paying for crashes, Global macro and alternative risk premia: how the big institutions diversify, Return stacking, overlays and portable alpha: stacking the premia). This one gives the map and the verdicts.

The candidates' crisis report card, before the detailed review, fits in one grid.

THE CRISIS REPORT CARDNo asset defends against everythingWhat each candidate returned, in real dollars, through four crises of different natures1973-1974inflation, a longbear regime2008growth crash,deflation scareMarch 2020flash crash,V-shaped rebound2022an inflation andrate shockUS equities (S&P 500)what the defense has to protect−49%−37%−34%−23%Cash (3-month T-bills)−5%+1%0%−4%Government bonds, 7-10 years−11%+18%+7%−20%Government bonds, 20 years and up−15%+34%+14%−35%Gold+135%+6%−2%−6%Trend following (BTOP50)no series+13%−5%+8%Three defenders share four crises, and not one of them holds all four: each collapses in another column.Real total return in dollars over the window, deflated by the US CPI. The frame marks the column's best defender.Windows dated before any measurement: 1973-1974 = December 31, 1972 to December 31, 1974; 2008 = calendar year;March 2020 = February 19 to March 23, 2020 (the S&P 500's record close, then its trough); 2022 = calendar year.Series: S&P 500 total return, 3-month T-bills, Treasuries 7-10 years and 20 years and up, spot gold, BTOP50 index, net.BTOP50 starts at the end of 1986 (empty cell, never extrapolated); a diversified index, tamer than the pure trend of the text.Gold loses 6% in real dollars in 2022, but the dollar's own rise gives 6.2 points back to a holder measuring in euros.
What each candidate returned, in real dollars, through four crises of different natures. The frame marks the best defender of the column: it moves from episode to episode, and every winner collapses somewhere else in its own row. That is the article's whole thesis, before the review that follows.

The candidates, regime by regime§

Cash (money market funds, T-bills, stable value). Regimes served: all of them, weakly. It never rises and almost never falls in nominal terms. It is the zero of the scale. Its real job is not defensive performance but the liquidity of your withdrawals, and the option that comes with it: the ability to buy everything else at the bottom. Its own enemy is inflation. The 1970s and 2022 both cut it down in real terms, interest paid or not. Carry cost: the gap to the portfolio's return, roughly 2 to 4 real points a year. Verdict: essential as a buffer, the cash buffer of 18 to 36 months (The cash buffer: size, cost, and what it is really for), ruinous as a permanent mattress beyond that. Inside a workplace plan, a stable value fund is the upgraded version of the same idea, a smoothed bond yield with the principal protected by an insurance wrapper. It makes a good buffer and a good short sleeve.

Short and intermediate nominal government bonds (1 to 7 years). Regimes served: the disinflationary recession and, more modestly, the growth crash. Neutral elsewhere. They lose in inflation, but not by much, because short duration caps the damage (−5 to −10% in 2022, against −30% for long bonds). This is the generalist defender: never heroic, rarely catastrophic, liquid, with almost no carry cost while real rates are positive. Verdict: the bedrock of the defensive sleeve (Bonds in retirement: types, duration, the exact job).

Long government bonds (15 to 30 years). Regime served: deflation, and growth crashes. There they are the only asset that rises violently (2008, +25 to +30%; 2020, +20%), a power that comes straight from duration. Those are US Treasury numbers, and they do not travel: other government markets paid a fraction of that in the same two years. Killer regime: inflation and rising rates (2022, −30 to −40%, worse than stocks). This is an asset with a thesis, not a "safe investment". It is the most expensive misunderstanding in mainstream wealth advice. Verdict: a dose of deflation insurance held openly (5 to 15%), the way gold plays inflation insurance. Never the core of the sleeve (Bonds in retirement: types, duration, the exact job).

Inflation-linked bonds (linkers, TIPS). Regime served: inflation, by contract. It is the only asset whose cash flow is mechanically indexed to the retiree's enemy number one (Inflation and the withdrawal rate: the exact link). One subtlety caught everyone out in 2022: at equal duration, a linker still moves with real rates. Long linkers fell that year despite the inflation, because real rates were rising faster. The contractual protection pays at maturity, not day by day. Hence the preference for short durations, or for a ladder held to maturity (Inflation-linked bonds: the only contract written in real terms, Bond ladders (and the ladder of linkers)). Verdict: the most under-owned block in retirement portfolios, given what it does. A dose of 5 to 15% of assets, and more when the floor is matched with linkers held to maturity.

Gold. Regimes served: inflation in episodes (spectacular in the 1970s, respectable in 2022), and above all crises of confidence, monetary, geopolitical or banking, where it has no equal. It is the anti-political-tail asset, the only one that is nobody's liability. Killer regime: disinflationary prosperity with positive real rates (1980-2000, −70% real, twenty years of purgatory). Carry cost: a long-run real expectation near zero, volatility of 15 to 18%, and the chance of two decades in the wilderness. Verdict: the defender of the regimes nothing else covers, in a dose of 5 to 10% and never more. And a test of temperament more than an asset (Gold in a retirement portfolio).

Managed futures (trend following). Regimes served: long regimes, up or down. The strategy buys what is rising and sells what is falling, across every asset class. It wins in anything that lasts, and above all in the great slow destructions (2008, +21% for pure trend; 2022, +27%, its best year since 2000, while everything else was losing). Killer regime: trendless markets and V-shaped reversals, like the too-fast crash of 2020. It is the only defensive asset whose long-run expectation is positive (an estimated 2 to 4% real gross, to be marked down for fees). The price you pay: complexity, real fees, and mediocre years in a row. Verdict: the most valuable defensive block for the modern inflationary regimes, in a dose of 5 to 15% for anyone willing to own a managed product (Managed futures and trend following: the diversification that works in a crisis).

The others, faster. Broad commodities offer a genuine inflation hedge. They charge for it in chronic negative carry (the roll, taken apart in Global macro and alternative risk premia: how the big institutions diversify) and in brutal volatility. For an individual they are dominated by gold plus trend.

Direct real estate is a real asset in the literal sense, but illiquid, concentrated and expensive to run. It is supplementary income, not a portfolio defense (Real estate in a FIRE plan (your home, and rentals)). Listed property (REITs) correlates with stocks in a crash (−60% in 2008-2009): diversifying in normal times, not defensive in a crisis.

Safe-haven currencies (USD, CHF, JPY) really do cushion crashes, the dollar rising when everything else falls. Whether that helps you depends on where you live. If your home currency is not the dollar, an unhedged world equity fund already carries a dollar position, free, in every global portfolio: one good reason not to hedge the currency of your stocks (International diversification (and home bias)). If your home currency is the dollar, the cushion is already under you, and buying the others adds little.

Option strategies (puts, tail hedging) offer the only true crash convexity. But their premium eats the expectation (−2 to −5% a year, carried for years on end). They belong to disciplined institutions. For an individual they can mostly be replicated with duration plus gold plus trend. The full case, toxic vehicles included, is in Long volatility and tail hedging: paying for crashes.

Finally, global macro and multi-premia funds share trend's promise, winning in hostile regimes, without its replicable index or its transparency. Their fair trial is in Global macro and alternative risk premia: how the big institutions diversify.

Putting it together: diversify the defense itself§

The summary table, to be read against the regime grid:

Enemy of the planStarterBackup
Short growth crashLong bonds, cashIntermediate duration, safe-haven currency
Long bear regimeManaged futuresGold, cash (optionality)
Persistent inflationLinkers, goldManaged futures, commodities
Deflation / bustLong bondsCash
Crisis of confidence / political tailGoldSafe-haven currencies, geographic diversification
Funding withdrawals at the bottomCash, money marketShort bonds

Three assembly principles follow. First, every row of the table needs at least one starter in your portfolio. That is the regime audit already practiced (Market regimes (growth × inflation, sticky bears) and why they matter). The typical retail portfolio, stocks plus a savings account, leaves the inflation row and the confidence row wide open. Second, no starter should carry its row alone and in a massive dose. Browne's 25% gold is the historical upper limit, and 7% gold plus 7% linkers covers more than 15% gold alone. Defenses get diversified too, because each one has its own failure mode. Third, the total dose of defense follows the phase and the floor (Accumulation, transition, withdrawal: the three lives of a FIRE plan, Glide paths: the bond tent, rising equity, and the fragile window). Count on 25 to 40% of assets during the fragile window, less before and after. Past 40 to 45%, the portfolio leaves the allocation plateau by the wrong side and erosion takes over again (The stock and bond allocation in retirement). If that budget hurts, there is an advanced route: hold the core in a moderately levered fund, so the defense sits on top of the engine instead of in its place (Return stacking, overlays and portable alpha: stacking the premia).

Testing the defensive sleeve in simulation. A defense is judged A/B, like the all-weather portfolios (All-weather portfolios: Browne, All Weather, Golden Butterfly, Dragon). Build the corrected version from blocks you can actually buy (Building it with US-listed ETFs: the retiree's shopping list has the shopping list), then run the current portfolio and the corrected one against each other, same withdrawal rule and same horizon, and compare the outcomes. The readings that decide are the tails, where the defense is supposed to work: a sequence stress, a lost decade, and a replay of the inflationary vintages (1966, 1973). One less obvious effect is worth the detour. A well-defended portfolio needs a shorter buffer, since part of the insurance already sits inside it. Then comes the reverse check, just as important. Central-case failure probability and median wealth must not collapse. A defense that costs 1.5 points of expected return is protecting a plan it made poorer first (Forward-looking expected returns (Morningstar, Vanguard, the investment banks)). The balance shows up in the distribution of outcomes. It cannot be declared.

The essentials§

Going further§