The Lost Decade for Stocks: What 2000-2009 Did to Investors

What is the lost decade in investing

The phrase is best known from Japan, where "lost decade" describes the persistent slow growth of the 1990s (St. Louis Fed), the years after its stock and property bubble burst. The label later stuck to the US stock market of the 2000s, because the S&P 500 closed the decade below where it had opened.

Two crashes did the damage. The first was the dot-com bust: from its closing high of 1,527.46 on March 24, 2000, the S&P 500 fell to 776.76 on October 9, 2002, a 49% drop. The index then climbed for five years and set a new record of 1,565.15 on October 9, 2007, only for the financial crisis to drag it down 57% to 676.53 on March 9, 2009. That final low sat 56% below the 2000 peak. The stories behind each crash are in our pieces on the dot-com bubble and the 2008 financial crisis; this article sticks to what the full decade added up to.

The chart explains why the decade felt worse than a single crash. The index did get back above its 2000 peak, on May 30, 2007, at 1,530.23. Anyone who held on for seven years and finally saw their money back then watched it fall by more than half again within two years. The 2007 high was not beaten until March 28, 2013, when the S&P 500 closed at 1,569.19.

S&P 500 lost decade returns: price vs. total return

Which number you quote depends on whether dividends count. In our data the S&P 500 price index went from 1,455.22 on the first trading day of 2000 (January 3) to 1,115.10 on December 31, 2009. That is -23.4%, or about -2.6% a year. Measured from the last close of 1999 (1,469.25) the loss is 24.1%.

Dividends softened the blow but could not reverse it. Compounding the annual S&P 500 total returns published by NYU Stern's Aswath Damodaran gives -9.1% for 2000-2009, or -0.95% a year (Damodaran, NYU Stern). Dividend yields were low at the start, around 1.1-1.2% in 1999-2000, so they offered little cushion (Damodaran, implied premiums).

Inflation made it worse. Consumer prices rose 28.8% between December 1999 and December 2009 (CPI-U, FRED). Adjusted for that, the price index lost about 41% of its purchasing power over the ten years.

What still worked: gold, bonds, cash and a few stocks

A decade that was terrible for the index was fine, or better, for several things sitting right next to it. The table uses the same start and end dates for every series in our dataset, plus two external benchmarks for comparison.

Gold was the standout. In our spot data it started the decade at $289 an ounce, first closed above $1,000 in March 2008 and ended at $1,096.60. Treasuries gained in nine of the ten years, helped by falling interest rates: the monthly average Fed funds rate peaked at 6.54% in 2000, sat around 1% from mid-2003 to mid-2004 and was 0.12% in December 2009. Falling yields push up the price of existing bonds, and by Damodaran's figures the 10-year Treasury returned 20.1% in 2008 alone while the S&P 500, dividends included, lost 36.6%. Cash held up too. Our model deposit at the Fed funds rate earned about 3% a year, only a little ahead of 2.6% annual inflation, yet it ended the decade 34% larger while the S&P 500 price index ended 23% smaller.

Among individual stocks the spread was enormous. Exxon Mobil rose while oil climbed from about $26 to $79 a barrel (WTI, FRED). Altria's figure includes two large spin-offs, Kraft Foods and Philip Morris International. Walmart's shares ended lower. The tech leaders of 1999 fared far worse: in our data, which includes reinvested dividends, Cisco lost more than half its value, Intel almost half and Microsoft more than a third, while Citigroup ended the decade down almost 90% after the 2008 crisis. Apple's share price rose more than sevenfold. Picking it in advance was another matter, which is the point of our piece on survivorship bias.

Dollar-cost averaging vs. lump sum, computed

To see how regular saving fared, we ran a simple test on the S&P 500 price index. Assumptions: $100 invested at the close of the first trading day of every month from January 2000 to December 2009 (120 purchases, $12,000 in total), against $12,000 invested in one go on January 3, 2000. No dividends, no fees, no taxes. Both portfolios are valued at the December 31, 2009 close.

The monthly buyer did better because many purchases happened at low prices in 2002-2003 and 2008-2009. The average index level paid was 1,155.75, close to where the decade ended. The ride was rough: at the end of February 2009 the $11,000 contributed so far was worth $6,840, a 38% paper loss after more than nine years of saving. Those cheap shares mattered later. The units bought through 2009 were worth $16,293 when the S&P 500 set its new high on March 28, 2013, while the January 2000 lump sum was worth $12,940.

This result is specific to a decade that started at a peak. In a rising market, money invested earlier has more time to grow, so a lump sum tends to come out ahead. The lesson here is narrower: steady contributions reduced the damage from a badly timed start.

How rare is a decade like this? Rolling 10-year windows since 1962

Calendar decades are arbitrary, so we checked every 10-year window in our S&P 500 data, measured from one month-end to the same month ten years later. From January 1962 to September 2026 there are 657 such windows. 61 of them, about 9%, ended with the price index lower. All 61 fall into two clusters: windows starting between 1964 and 1972 (ending 1974-1982) and windows starting between 1998 and 2001 (ending 2008-2011).

The worst window in more than six decades of data, February 1999 to February 2009, belongs to the lost decade. In real terms the 1970s were just as harsh: deflated by CPI, the July 1972 to July 1982 window lost 57% of its value and February 1999 to February 2009 lost 54%. Our 1970s inflation article covers how that happened with a flat index and double-digit inflation.

Start-date sensitivity is the uncomfortable part. Calendar decades measured December to December show how much the starting year mattered: 1995-2004 gained 164%, 1998-2007 gained 51%, 1999-2008 lost 27%, 2001-2010 lost 5%, and 2003-2012 gained 62%. A few years' difference in when someone began investing separated a doubling from a loss.

Japan: a 34-year wait to break even

The American version looks mild next to the original. Japan's Nikkei 225 closed at a record 38,915.87 on December 29, 1989. It hit its post-bubble closing low of 7,054.98 on March 10, 2009, about 82% below the peak and the lowest level since 1982 (Nikkei Indexes). The index did not close above its 1989 record until February 22, 2024, when it ended at 39,098.68, 34 years later (The Japan Times).

Both declines began from record highs after long booms. The difference was what came after. The S&P 500 set a new high four years after its 2009 low, and on October 1, 2026, the S&P 500 price index closed at 7,666.49, almost seven times its end-2009 level. Japanese investors who bought at the top waited more than a generation just to get back to even in price terms.

What the lost decade means for long-term investors

None of this predicts the next ten years. It does put some common phrases in context.

• "Stocks always win in the long run" depends on how long the run is. Ten years was not long enough in about 9% of the windows since 1962 on a price basis.
• The starting point mattered. The 2000s began with the S&P 500 near a record and a dividend yield of about 1.2%, well below the roughly 2.7-5.5% range of the 1960s and 1970s.
• Diversification did real work. Gold, Treasuries and cash all rose while the index fell, and the 2008 bond rally offset part of the stock crash for mixed portfolios.
• Regular contributions changed the outcome. In our test the monthly saver lost 3.5% in price terms, the lump-sum investor 23.4%.
• Index returns hide wide dispersion. The same decade produced Citigroup's loss of almost 90% and Altria's more than sixfold gain with dividends reinvested, a reminder of how hard it is to beat the S&P 500 by choosing names in advance.

Living through the 2000s without knowing the ending

Looking at a finished chart, the lost decade seems obvious. In 2003 or 2006 it did not: the market was rising, and no one knew a second crash was coming. Blindfolio, a free browser game, lets you test that feeling. Choose the 2000s as the start decade and you get a 10, 15 or 20-year period that begins at a random point in that decade. You will see the dates, but companies appear only under codenames. Time runs week by week and you can pause at any moment; you start with cash and a monthly salary, rate decisions and inflation arrive as news headlines, and the Fed funds rate is shown on the bank deposit. The catalog includes index funds, Treasuries, gold, a deposit at the Fed funds rate, and companies that did not survive the period. For the fully blind version, play a random round: the years are hidden too, and it may land in this era. The real names, and how you did against the S&P 500, appear only at the end.