Last week, we highlighted a MarketWatch column that referenced a new study showing that stocks, against conventional wisdom, have actually served as poor hedges during periods of marked inflation. But a closer look at the study shows that that is far from the whole story.

The study was performed by Elroy Dimson, Paul Marsh, and Mike Staunton of the London Business School, who over the years have performed some of the most in-depth research ever on asset prices. Their latest effort looks at 19 different markets since 1900, examining how different assets performed during periods of “marked” inflation. They found that bonds — not surprisingly — get crushed in such periods, averaging a real return of -23.2%. Stocks, however, also struggle, they found, losing an average of 12% in real terms.

That sounds troubling, especially given how stocks are generally considered to be good inflation fighters. But if you look at Dimson, Marsh, and Staunton’s analysis (which is part of a Credit Suisse report), there’s a lot more to the story. For one thing, in their study, “marked” inflation signified inflation of at least 18% annually — a very steep figure that the U.S. didn’t even reach during the sky-high inflation of the late ’70s/early ’80s.

So how have stocks fared when inflation has been high, but not at that 18% mark? Well, according to the professors, when inflation has been between 8% and 18%, stocks haven’t fared great, but they’ve remained ahead of inflation, with 1.8% average annual real returns. Bonds, on the other hand, have averaged real annual losses of 4.6% in those periods. And, when inflation has run between 4.5% and 8%, stocks have produced annual real returns of 5.2%, while bonds have barely eked out a positive return.

Gold, meanwhile, has been a good hedge in those periods when inflation has been at least 18%, with real returns that were just in positive territory. In the 8% to 18% inflation range, gold has also outperformed stocks, with real returns of 4.4%. In that 4.5% to 8% inflation range, however, gold has produced real returns of just 2.8%.

But Dimson, Marsh, and Staunton warn against using inflation data as a timing signal, particularly since investors don’t know what the inflation rate is for a particular period until the period is over, and asset prices have already moved. “This is not a market timing tool,” they write. “High inflation may look like a sell signal, but our model is derived with hindsight and could not be known in advance; there is clustering of observations, so many of the signals may occur at some past date (e.g. the 1920s); and it is not clear where sales proceeds should be parked. In particular, real interest rates tend to be lower in inflationary times, the expected real return on Treasury bills will be smaller after an inflation hit, and other safe-haven assets like inflation-linked bonds are likely to provide a reduced expected return in real terms.”

Click here for a PDF copy of the professors’ report, which includes in-depth analysis of how inflation has impacted a number of asset classes over the past 112 years.