For about 20 years after 1996 the S&P 500 P/E ratio stayed high by 1935-1995 standards.

Use standard S&P 500 P/E data (Shiller trailing or CAPE series). True: from 1996 to about 2016, the P/E stayed above its 1935–1995 average for nearly all of the period, with at most brief excursions to or below that average (a few months in total, e.g. around March 2009). Mostly true: it was above that average for the large majority of the period (about 85% or more of months), with one or two short dips. Mixed: it was above average for roughly half to most of the period (about 50–85%). Mostly false: it was above average for less than half of the period, but with some long stretches of high readings. False: it was mostly at or below the 1935–1995 average over 1996–2016.

Set before the research began.

1. E1 CAPE Fear: Why CAPE Naysayers Are Wrong Financial Analysts Journal (fa-mag.com), by Research Affiliates authors · 5 January 2018

   “Since 1996, the US CAPE ratio has been above its long-term simple average (16.6) 96 percent of the time, and above 24, roughly one standard deviation above its historical norm, more than two-thirds of the time.”

2. E2 CAPE Fear: Why CAPE Naysayers Are Wrong Financial Analysts Journal (fa-mag.com), by Research Affiliates authors · 5 January 2018

   “before the mid-1990s, the normal valuation was in the mid-teens, and in the mid-twenties thereafter.”

3. E3 CAPE Fear: Why CAPE Naysayers Are Wrong Financial Analysts Journal (fa-mag.com), by Research Affiliates authors · 5 January 2018

   “its always exceeding the estimated fair-value trend line over both the first quarter-century and the latest quarter-century, except briefly in 2008–09.”

E1, an independent Research Affiliates analysis, reports that since 1996 the US CAPE ratio was above its long-term simple average of 16.6 about 96% of the time, and above 24 for more than two-thirds of the time. E3 says the CAPE stayed above its estimated fair-value trend over the latest quarter-century except briefly in 2008–09. E2 says normal valuations were in the mid-teens before the mid-1990s and in the mid-twenties afterward. E1 uses the long-term average rather than the exact 1935–1995 average, but E2 puts pre-1990s norms in the mid-teens. Together they indicate the P/E stayed high for nearly all of 1996–2016, with a brief dip around 2008–09 [E1][E3]. None of the sources measures against the exact 1935–1995 average, but E2 puts the pre-mid-1990s norm in the mid-teens, at or below the 16.6 average that E1 says was exceeded 96 percent of the time. On that basis the reading is consistent with the criteria for true.