# 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]], [[inflation-linked-bonds]], [[gold-in-retirement]], [[managed-futures]]). So do the more exotic defenses ([[long-volatility]], [[global-macro]], [[return-stacking]]). This one gives the map and the verdicts. ::: cle The spec for a real defensive asset Four requirements, in order. (1) Hold, or **rise**, when the equity engine suffers. Not only through six-month crashes, but through hostile **regimes** that last for years ([[market-regimes]]). (2) A decorrelation that **survives** a crisis. Plenty of average "decorrelations" converge to 1 exactly when you need them ([[fat-tails]]). (3) Liquidity: a defensive asset you cannot sell at the bottom to fund a withdrawal defends nothing. (4) A carry cost you know and have budgeted. Real insurance carries a premium, and a "defensive" asset that also claims to pay is hiding its premium somewhere. ::: The candidates' crisis report card, before the detailed review, fits in one grid. ::: figure defenses-bulletin 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 ([[cash-buffer]]), 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]]). **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]]). **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-withdrawal-rates]]). 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]], [[bond-ladders]]). 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-retirement]]). **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]]). **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 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-retirement]]). 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]]). 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]]. 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]]. ::: attention The gallery of false defensives Four products **sell** themselves as defensive without meeting the spec. - **Dividend stocks** are stocks, with a beta of 0.8 to 0.9. They fall with the market, and a dividend is not a coupon. It gets cut in a crisis (2008-2009, S&P dividends down 25%). This is the yield illusion, taken apart by ERN (Parts 29 to 31 and 40, [[the-ern-series]]). - **High yield** is a stock-bond hybrid whose defaults land exactly in the crashes, with a correlation to stocks around 0.7 in a crisis. It cushions nothing. - "Principal protected" **structured products** buy their protection by selling the upside and paying opaque margins. Almost always, the plain bonds-plus-stocks mix you assemble yourself does better. - **Min-vol** or "low volatility" funds are real stocks that wobble a little less. A legitimate style choice inside the equity sleeve, not a defense. The infallible test: ask "what did this product do in 2008 and in 2022?". A real defensive asset shows a good report card in **at least** one of the two, and no disaster in the other. The full case, covered calls, dividend aristocrats, REITs, private equity and crypto included, is in [[false-defensive-assets]]. ::: ## Putting it together: diversify the defense itself The summary table, to be read against the regime grid: | Enemy of the plan | Starter | Backup | |---|---|---| | Short growth crash | Long bonds, cash | Intermediate duration, safe-haven currency | | Long bear regime | Managed futures | Gold, cash (optionality) | | Persistent inflation | Linkers, gold | Managed futures, commodities | | Deflation / bust | Long bonds | Cash | | Crisis of confidence / political tail | Gold | Safe-haven currencies, geographic diversification | | Funding withdrawals at the bottom | Cash, money market | Short 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]]). 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 ([[the-three-phases]], [[glidepaths]]). 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 ([[stock-bond-allocation]]). 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]]). **Testing the defensive sleeve in simulation.** A defense is judged A/B, like the all-weather portfolios ([[all-weather-portfolios]]). Build the corrected version from blocks you can actually buy ([[building-it-with-us-etfs]] 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 ([[expected-returns]]). The balance **shows up** in the distribution of outcomes. It cannot be declared. ::: exemple Filling two wide-open rows The starting portfolio: 65% global stocks, 35% in a stable value fund. Audit against the table: short crash covered, persistent inflation **wide open** (a stable value fund follows rates years late), confidence **wide open**, long regime half covered. The fix: stocks unchanged at 65%, the stable value sleeve cut to 15% (buffer and liquidity), plus 8% short linkers, 7% gold and 5% long bonds, the deflation insurance the stable value fund does not supply either. Verdict: central-case failure probability barely moves (3.9 to 3.7%), the sequence stress goes from 6.8 to 5.6%, and above all the 1966 and 1973 vintages, replayed, go from "borderline" to "ridden out". Median wealth gives up 6%. Two rows filled, a known price, and a plan that no longer depends on a regime being kind. ::: ## The essentials - "Defensive" does not exist in the abstract. Each asset defends against one enemy (crash, long regime, inflation, deflation, confidence) at a given carry cost. The spec: holding up through a hostile **regime**, a decorrelation that survives a crisis, liquidity, and a premium you know. - The starters, enemy by enemy: long bonds (deflation), linkers and gold (inflation and confidence), managed futures (long regimes), cash (liquidity and optionality), short duration (the generalist bedrock). And for anyone whose home currency is not the dollar, the currency exposure of an unhedged global equity fund is a free shock absorber. - The false defensives to leave alone: dividend stocks, high yield, structured products, min-vol. The test is one question: "what did they do in 2008 and in 2022?". - Defense gets diversified like everything else: every row of the table covered, no starter massive and alone, a total dose of 25 to 40% during the fragile window. And always the reverse check in simulation, because the median must not pay more than the tails bring in. - The major blocks each have their article ([[bonds-in-retirement]], [[inflation-linked-bonds]], [[gold-in-retirement]], [[managed-futures]]), and so do the exotic defenses ([[long-volatility]], [[global-macro]], [[return-stacking]]). The full assembly job is in [[all-weather-portfolios]]. --- ## Going further - Ilmanen, *Expected Returns* (2011) and *Investing Amid Low Expected Returns* (2022): the premia and carry costs of every candidate, with numbers. - Early Retirement Now, Parts 29 to 31 and 40 (the yield illusion), Part 34 (gold as a sequence hedge) ([[the-ern-series]]). - [portfoliocharts.com](https://portfoliocharts.com) (the blocks tested in every mix); AQR's research library for managed futures and the carry cost of option hedges. - In this book: the founding grid [[market-regimes]], and each block in its own article.