Crisis-Proof Investments: What Defensive Sectors Saved in Four Crashes, and What They Cost

Every list of crisis-proof investments names the same three sectors: utilities, consumer staples and healthcare. The list is not wrong. It answers the wrong question. Whether a sector falls less in a crash, whether it gets back to its high sooner, and what it costs you in the rally that follows are three different questions, and the data gives three different answers. Most guides stop after the first one.

We measured all three for the US Select Sector SPDR ETFs against SPY, with dividends reinvested, across the four bear markets since the funds launched in December 1998. Staples and healthcare fell less than the market in all four. Utilities fell more than the market in 2020. In the twelve months after each low, an equal-weight basket of the three defensive sectors trailed the market by between 14.7 and 30.2 percentage points. And over the full period the basket compounded at 7.75% a year against 8.68% for SPY. Resilience is real, and it has a price.

How each sector held up in four crashes

Each column is the return of the ETF over the same window: from SPY’s total-return peak to its total-return low. Dividends are reinvested and fund fees are already deducted. Negative numbers smaller than SPY’s mean the sector fell less than the market.

ETF (sector)	2000–02 24 Mar 2000 – 9 Oct 2002	2007–09 9 Oct 2007 – 9 Mar 2009	2020 19 Feb – 23 Mar 2020	2022 3 Jan – 12 Oct 2022
SPY (S&P 500)	−47.5%	−55.2%	−33.7%	−24.5%
XLU (utilities)	−35.8%	−42.5%	−35.4%	−11.3%
XLP (consumer staples)	+1.2%	−28.5%	−24.2%	−10.6%
XLV (healthcare)	−17.2%	−38.3%	−27.9%	−11.6%
XLE (energy)	−23.7%	−47.9%	−56.1%	+44.5%
XLF (financials)	−22.2%	−81.6%	−42.8%	−22.3%
XLK (technology)	−81.9%	−51.4%	−31.2%	−33.1%
XLI (industrials)	−36.3%	−62.3%	−41.6%	−18.2%
XLY (consumer discretionary)	−23.2%	−56.6%	−33.5%	−33.5%
XLB (materials)	−20.6%	−56.7%	−36.2%	−22.1%
XLRE (real estate)	—	—	−37.8%	−32.2%
XLC (communication services)	—	—	−29.8%	−38.7%
Defensive trio (XLU, XLP, XLV equal weight)	−16.8%	−36.3%	−29.2%	−10.8%

Three things stand out. Only two of the 40 sector results are positive: staples in 2000–02 and energy in 2022. Nothing was crisis-proof in the literal sense. Staples and healthcare were the only sectors that fell less than SPY every time. And the sector that defended best changed with the kind of crisis: staples when a stock bubble burst, energy when inflation and rising rates drove the fall.

The practical consequence: “defensive” describes a tendency across crises, not a property you can count on in the next one. What a sector defends against depends on what causes the fall.

What “defensive” means in the data

The usual explanation for why some investments are resilient is demand: people keep paying electricity bills, buying toothpaste and filling prescriptions in a recession, so earnings in these sectors move less with the economy. The measurable version of that idea is beta, how much a sector moves when the market moves, and the split between how much of the market’s gains and losses it captures. Over 27.7 years of monthly data:

ETF	Beta to SPY	Share of SPY’s up months captured	Share of SPY’s down months captured	Daily volatility (annualised)
SPY	1.00	100%	100%	19.2%
XLU (utilities)	0.46	53%	33%	19.2%
XLP (staples)	0.50	55%	43%	15.2%
XLV (healthcare)	0.71	81%	70%	17.8%
Defensive trio	0.56	63%	49%	14.7%
XLE (energy)	0.96	102%	90%	28.6%
XLF (financials)	1.13	102%	113%	28.2%
XLK (technology)	1.31	134%	133%	26.0%

Capture is the sector’s average return in months when SPY rose (or fell), divided by SPY’s average in those months. The defensive trio took about half of the market’s monthly losses and about 63% of its gains. That asymmetry is the whole case for these sectors.

Look at the utilities row, though. XLU’s beta is 0.46, but its daily volatility is exactly the market’s. Low beta is not low risk. Utilities move a lot. They just move for their own reasons, interest rates among them, and those moves line up only loosely with the stock market. A holding can be a poor hedge and a volatile position at the same time. That is why the risk-reduction rules in how to manage investment risk start with position size and time horizon, not sector labels.

Where the defensive label failed

An honest version of the table has to include the misses. The window returns above are measured from SPY’s peak. The ETF’s own drawdown, measured from its highest close inside that window to its lowest close, is sometimes much worse:

• Utilities, 2000–02. XLU held up through 2000 and peaked in December 2000. It then fell 52.3% into October 2002, a deeper fall from its own high than SPY’s 47.5%. Over SPY’s window it looked defensive. From its own peak it was not.
• Utilities, 2020. XLU fell 35.4% against SPY’s 33.7%. In a five-week liquidity crash, the rate-sensitive “bond proxy” sector was sold with everything else.
• Staples, 2000–02. Over SPY’s window, XLP was up 1.2%. But from its December 2000 high it fell 32.9% by July 2002 and kept falling, reaching −35.9% in March 2003, after the market had already bottomed. It got back to that high in October 2006, 5.8 years later, on the same day SPY regained its March 2000 peak.
• Financials, 2007–09. Banks were a dividend-investor favourite before 2008. XLF fell 81.8% from its October 2007 high and needed 9.3 years to recover, until February 2017.
• Energy, 2020. XLE fell 56.1% while SPY fell 33.7%. On its worst day, 18 March 2020, it was 71.3% below its 2014 high, the deepest drawdown of any full-history sector ETF except financials and technology.
• Technology, 2000–02. XLK fell 82.0% and took 16.9 years to get back to its March 2000 high, in March 2017.
• Real estate, 2022. REITs are often sold as income-plus-resilience. XLRE fell 32.2% in 2022, more than SPY, and did not regain its January 2022 high until April 2026.

The failures follow a pattern. REITs failed in 2022, when rates rose, because their valuations depend on borrowing costs and on how bond yields compare with their dividends. That link is explained in why bond yields can rise even when central banks cut rates. Utilities failed in the 2020 dash for cash, when everything was sold at once, and in 2001–02, after rallying through 2000 while the rest of the market fell. Financials failed in the crisis that started in their own balance sheets. Energy failed when the crash was a collapse in demand for oil. A sector protects you from the crisis it is not at the centre of.

Falling less is not the same as recovering sooner

The second question is how long each sector stayed below its previous high. Each cell shows the ETF’s own maximum drawdown inside the crisis window and the years from that high until it was regained.

ETF	2000–02	2007–09	2020	2022
SPY	−47.5% · 6.6 yrs	−55.2% · 4.9 yrs	−33.7% · 0.5 yrs	−24.5% · 1.9 yrs
XLU	−52.3% · 4.2 yrs	−46.5% · 4.5 yrs	−35.6% · 1.2 yrs	−20.7% · 1.9 yrs
XLP	−32.9% · 5.8 yrs	−32.4% · 1.5 yrs	−24.3% · 0.5 yrs	−16.3% · 1.9 yrs
XLV	−29.8% · 3.2 yrs	−39.2% · 3.4 yrs	−27.9% · 0.3 yrs	−16.1% · 1.7 yrs
XLE	−41.5% · 3.2 yrs	−57.4% · 5.0 yrs	−56.3% · 1.0 yrs	−26.0% · 0.4 yrs
XLF	−36.6% · 3.0 yrs	−81.8% · 9.3 yrs	−42.8% · 0.9 yrs	−25.8% · 2.1 yrs
XLK	−82.0% · 16.9 yrs	−53.0% · 4.2 yrs	−31.2% · 0.3 yrs	−33.1% · 1.4 yrs
XLI	−43.5% · 4.0 yrs	−62.3% · 3.6 yrs	−41.6% · 0.7 yrs	−21.6% · 1.4 yrs
XLY	−31.5% · 1.9 yrs	−56.6% · 3.1 yrs	−33.9% · 0.3 yrs	−35.9% · 2.8 yrs
XLB	−31.1% · 1.4 yrs	−59.8% · 5.0 yrs	−36.3% · 0.4 yrs	−24.7% · 1.9 yrs
XLRE	—	—	−38.8% · 1.1 yrs	−32.2% · 4.3 yrs

Utilities in 2020 and 2022 took as long as the market to recover, or longer, despite falling about the same or less. Staples after 2000 took 5.8 years against SPY’s 6.6, which is barely an improvement on a much smaller fall. A small drawdown recovers quickly only if the sector then rises at a normal pace, and defensive sectors rise slowly: the trio captured 63% of the market’s up months. The clearest win was healthcare: shallower falls than SPY and quicker recoveries in all four episodes.

For the market-level version of this question, including how long recoveries take once inflation is counted, see bear markets since 1950: depth, duration and recovery.

The cost arrives in the rally

Defensive sectors capture about half of the market’s losses and about 63% of its gains. After a crash, that second number is the one that matters. Measured from each SPY low:

Crisis	12 months after the low: SPY	12 months after the low: defensive trio	Peak to low + 12 months: SPY	Peak to low + 12 months: trio
2000–02	+35.7%	+21.0%	−28.8%	+0.6%
2007–09	+71.9%	+44.1%	−23.0%	−8.2%
2020	+77.5%	+47.3%	+17.7%	+4.3%
2022	+23.5%	+2.2%	−6.7%	−8.8%

The last two columns show the answer. Take the whole episode, from the market’s peak through the first year of recovery. The defensive basket was well ahead in the two long bear markets, which lasted 30.5 and 17 months from peak to low. It was behind in the two short ones. In 2020 the crash lasted five weeks and the rebound erased the advantage within months. In 2022 the basket’s 13.7-point advantage at the low became a 2.1-point deficit a year later.

The calendar-year record shows the same trade-off. SPY lost money in six calendar years between 1999 and 2025: 2000, 2001, 2002, 2008, 2018 and 2022. The defensive trio did better than SPY in all six. In the 21 full years when SPY rose, the trio did better in only six, and in the median up year it trailed by 6.3 percentage points.

The practical consequence: defensive sectors pay off in long, grinding bear markets and cost you in short, sharp ones. Nobody knows in advance which kind is coming. So the choice is not whether defensives “work”. It is whether you would rather give up return in most years to lose less in a few.

What 27 years of resilience cost

From the close of 31 December 1998 to 24 September 2026, with dividends reinvested:

ETF	Annual return (CAGR)	$10,000 became	Worst drawdown
XLK (technology)	10.53%	$160,787	−82.0% (Oct 2002)
XLY (consumer discretionary)	9.21%	$115,146	−59.0% (Mar 2009)
XLI (industrials)	9.08%	$111,452	−62.3% (Mar 2009)
XLE (energy)	9.04%	$110,194	−71.3% (Mar 2020)
SPY (S&P 500)	8.68%	$100,604	−55.2% (Mar 2009)
XLV (healthcare)	8.59%	$98,208	−39.2% (Mar 2009)
XLB (materials)	7.94%	$83,238	−59.8% (Mar 2009)
Defensive trio (monthly rebalanced)	7.75%	$79,347	−39.2% (Mar 2009)
XLU (utilities)	7.13%	$67,497	−52.3% (Oct 2002)
XLP (consumer staples)	6.53%	$57,777	−35.9% (Mar 2003)
XLF (financials)	5.89%	$48,929	−82.7% (Mar 2009)

The trio’s worst fall was 16 points shallower than SPY’s. It cost 0.93 points a year, or about a fifth of the final sum. Healthcare is the outlier. It came within 0.09 points a year of SPY with a beta of 0.71 and a worst drawdown of 39.2%. That fits the low-beta effect documented by Frazzini and Pedersen: across many markets, lower-beta assets have earned more per unit of risk than the simplest models predict. One sector over one period does not prove it.

Two caveats make these figures less clean than they look. First, the answer depends heavily on when you start. Of the 10-year windows that began between January 1999 and December 2007, the defensive trio beat SPY in 100%. Of those that began from 2008 onward, it beat SPY in 9%. The first group includes the 2000–02 crash near the start. The second includes the long rally since 2009. Any “defensives beat the market” or “defensives always lag” claim is mostly a statement about its start date. Second, the sectors change. Real estate was part of financials until GICS gave it its own sector in August 2016. Alphabet and Facebook moved from technology into the new communication services sector in September 2018. XLRE (from October 2015) and XLC (from June 2018) have short histories. Since launch, both trailed SPY by a wide margin: 6.59% a year against 14.85% for XLRE, and 11.59% against 14.90% for XLC.

Buy and hold vs rotating into defensive sectors

Most readers do not need to choose between owning defensives and not owning them. They need to decide whether to switch into them once a crash is under way. We tested the common versions of that decision with monthly data from January 1999, using the same equal-weight trio of utilities, staples and healthcare, and no trading costs or taxes.

Strategy	Annual return	$10,000 became	Worst month-end drawdown	Switches
Buy and hold SPY	8.68%	$100,604	−50.8%	0
70% SPY / 30% defensive trio, rebalanced monthly	8.52%	$96,626	−45.8%	0
Buy and hold the defensive trio	7.75%	$79,347	−33.6%	0
Switch to the trio after SPY falls 20%, back at a new SPY high	7.82%	$80,833	−41.5%	6
Switch to the trio after SPY falls 20%, back when within 10% of its high	7.56%	$75,494	−41.5%	8
10-month moving average: SPY above it, trio below it	8.59%	$98,467	−36.1%	40

Month-end drawdowns understate daily ones. On daily closes SPY’s worst was −55.2%. The rule “switch after a 20% fall, return once the market is back at its high” is how rotation usually happens in practice: you sell after the damage is visible and buy back once it feels safe. It ended with about a fifth less money than simply holding SPY. The damage was not spread evenly. The rule moved into defensives three times. From April 2001 to November 2006 the trio and SPY returned about the same (+31.0% against +31.9%). From October 2008 to March 2012 the trio did slightly better (+31.8% against +30.9%). From October 2022 to December 2023 the trio returned +9.1% while SPY returned +35.7%, and that single episode accounts for almost the whole shortfall. The rule never triggered in 2020: at the end of March that year SPY was 19.4% below its high, and a month later the fall was half undone. Switching after the fall is a bet that the bear market will be long. When it is short, the switch costs you most of the recovery.

The two alternatives that came closest to buy and hold share a feature: they did not depend on how the investor felt during the crash. The static 70/30 split cost 0.16 points a year and reduced the worst month-end drawdown by 5 points. That is cheap, but it also shows that a 30% defensive sleeve barely changes the worst case. The moving-average rule uses the 10-month signal from Mebane Faber’s 2007 paper, switching into the defensive trio here rather than into cash as the paper does. It came close to buy-and-hold returns with a much shallower drawdown. But it traded 40 times. In a taxable account every exit from SPY realises gains, and this result comes from one market over one period. It is not a guarantee.

The practical consequence: for most investors, buy and hold beats trying to time a move into defensives. If you want defensive exposure, hold it all the time at a weight you have chosen in advance, and rebalance on a rule. The rotation decision is where the money was lost. And if the aim is to change the worst case materially, a sector tilt inside an all-equity portfolio is a weak tool. The bigger decisions are how much you hold outside equities, which is the question in whether 60/40 still works, and whether to buy explicit insurance, which is what portfolio hedging strategies price out.

Do resource stocks outperform after downturns?

A common claim is that energy and materials lead the market out of a crash because an economic recovery lifts commodity demand. Here are the twelve months after each SPY low:

• Energy (XLE) beat SPY in one of four recoveries: 2020, +114.0% against +77.5%. It lagged after 2002 (+23.3% against +35.7%), after 2009 (+52.4% against +71.9%) and after 2022 (+13.5% against +23.5%).
• Materials (XLB) beat SPY in three of four: +40.8% against +35.7% after 2002, +85.2% against +71.9% after 2009 and +102.3% against +77.5% after 2020. It lagged after 2022: +16.0% against +23.5%.

The stronger pattern has nothing to do with resources. The biggest rebounds came from the sectors that had fallen furthest. After March 2009, financials gained 149.8% in a year, consumer discretionary 101.2% and industrials 100.9%. After October 2002, technology gained 67.8% and utilities 50.7%, both coming off the worst falls in their ETFs’ histories. Energy’s +114.0% in 2020 followed a 56.1% crash, bigger than any other sector’s in that window. Rebounds track how far a sector fell. The “resource” label adds little. And a sector that has fallen furthest has usually fallen for a reason. Financials gained 149.8% in those 12 months, and still did not regain their 2007 high until 2017.

Energy’s real defensive record is different. It was the only sector that rose in 2022, up 44.5% while SPY fell 24.5%, in the one bear market in this sample driven by inflation and rising rates. In 2008 and 2020, when the crash came from collapsing demand, it fell harder than the market from its own high. It behaves more like a hedge against inflation and rising rates than like a defensive sector. Any single-year number, including its +42.8% so far in 2026, says more about oil than about resilience.

A resilience strategy that survives contact with the data

Decide which crisis you are insuring against. Staples and healthcare fell less than the market in all four crashes. Utilities did in three, but not in the 2020 dash for cash. REITs failed when rates rose in 2022. Energy held up in the inflation-driven fall and failed in the demand collapses. Holding one “defensive” sector is a bet on one kind of crisis.

Hold the allocation, don’t time it. Every rule that switched after the damage was visible ended up with less money than buy and hold. A permanent weight, rebalanced mechanically, sells defensives after rallies and buys them after crashes without needing a forecast.

Count dividends, but don’t confuse them with safety. Since 1999, the gap between price return and total return was 3.6 points a year for utilities, about half of their 7.13%, and 2.5 points for staples. That is why every figure here uses total return. Dividends do not stop a fall: financials, a dividend favourite before 2008, fell 81.8%. The arithmetic of dividend return is covered in dividend investing: the arithmetic most guides skip.

Size the equity risk before choosing sectors. Even the best-behaved basket here fell 39.2% at its worst. If the money is needed inside a few years, the answer is less equity, not better equity.

What this does not say

This covers four crashes in one country over 27 years. That is too few to estimate how likely a sector is to defend next time. Sector ETFs are weighted by market value and can be dominated by a few companies. Sector definitions changed during the period, so XLK in 2000 and XLK in 2026 do not hold the same businesses. The returns ignore taxes and trading costs, which would hurt the rotation strategies more than buy and hold because they trade more. The crisis windows use SPY’s peak and low, not each sector’s. That is why the second table, which uses each sector’s own high, is included. Nothing here predicts which sector will hold up in the next crash, or when it will come.

Frequently asked questions

Which sectors are most resilient in a stock market crash?
Across the four US bear markets since 1999, consumer staples (XLP) and healthcare (XLV) were the only sectors that fell less than the S&P 500 every time. Utilities fell less in three of four but fell more than the market in 2020. Energy was the only sector that rose in 2022, and it fell far more than the market in 2020.

Are defensive sector ETFs worth holding during market volatility?
They reduce losses. They do not avoid them. An equal-weight basket of utilities, staples and healthcare ETFs captured about 49% of SPY’s monthly losses and 63% of its gains, and its worst drawdown was 39.2% against 55.2% for SPY. The cost was 7.75% a year against 8.68% from 1999 to September 2026.

Is buy and hold better than rotating into defensive stocks?
In our test, yes. Switching into the defensive trio after a 20% fall and returning at a new market high turned $10,000 into $80,833, against $100,604 for simply holding SPY. Almost all of the shortfall came from the 2022 rotation, which missed most of the rebound that started in October 2022.

Why do resource stocks outperform after downturns?
Mostly they don’t. Energy beat the S&P 500 in the year after the low in one of four recoveries (2020) and materials in three of four. The more reliable pattern is that the sectors that fell furthest rebounded most, such as financials after 2009 and technology after 2002, whatever they produce.

What is a resilience strategy for investors?
Hold a fixed weight in sectors whose earnings move less with the economy, rebalance to it on a rule, and decide the total amount of equity risk before choosing sectors. What the data argues against is switching into defensives after a crash has started.

Are utilities safe in a recession?
Not reliably. XLU fell 42.5% in 2007–09 and 35.4% in the 2020 crash, slightly more than the market, and from its December 2000 peak it fell 52.3%. Its beta to the market is low (0.46), but its daily volatility over the period was the same as the S&P 500’s.

Sources

• Select Sector SPDR ETFs (XLU, XLP, XLV, XLE, XLF, XLK, XLI, XLY, XLB from 22 December 1998; XLRE from 8 October 2015; XLC from 19 June 2018) and SPDR S&P 500 ETF (SPY), daily closes adjusted for splits and distributions (Yahoo Finance via the yfinance library, auto_adjust=True), through 24 September 2026, retrieved 25 September 2026. Adjusted closes approximate total return after fund expenses and before taxes. Source of every table in this article.
• S&P Dow Jones Indices and MSCI, Further revisions to the GICS structure in 2016, press release, 2 November 2015. Real Estate moved out of Financials into its own sector after the close on 31 August 2016.
• S&P Dow Jones Indices and MSCI, Select list of companies changing due to revisions to the GICS structure in 2018, press release, 11 January 2018. Alphabet and Facebook moved from Information Technology to Communication Services, and Netflix and Comcast from Consumer Discretionary, after the close on 28 September 2018.
• Andrea Frazzini and Lasse Heje Pedersen, Betting Against Beta, Journal of Financial Economics 111(1), 2014, pp. 1–25.
• Mebane T. Faber, A Quantitative Approach to Tactical Asset Allocation, The Journal of Wealth Management 9(4), Spring 2007, pp. 69–79. Source of the 10-month moving-average rule tested above.
• Crisis windows, drawdowns, recovery times, capture ratios, rolling windows and strategy results are our own calculations from the series above. Crisis windows run from SPY’s highest total-return close before each bear market to its lowest close. Sector drawdowns are measured from each ETF’s highest close inside that window. Strategy signals use month-end closes and take effect the following month.

Related reading

• Bear markets since 1950: depth, duration and recovery: the market-level record these sector numbers sit inside.
• Portfolio hedging strategies: what each one costs: explicit insurance, as an alternative to a sector tilt.
• Is 60/40 still alive?: the allocation decision that changes the worst case more than any sector does.
• Why bond yields rise when central banks cut rates: the rate mechanism behind the utilities and REIT failures.
• How to manage investment risk: why position size comes before sector choice.
• How often should you rebalance: the rule that makes a permanent defensive weight work.
• Dividend investing: the arithmetic most guides skip: why yield is not the same as resilience.

This article is general information, not personalised investment advice. It does not take into account the financial situation, objectives or risk tolerance of any individual reader. Sector ETFs are named as measurable proxies for their sectors, not as recommendations to buy or sell. All figures describe specific historical periods and do not predict future outcomes. Capital is at risk and past performance does not indicate future results.