One of my pet peeves is when people try to compare different eras, whether they be sports-related or markets-related. For example:

In 1982, the Lakers made 13 three-pointers all season and won the championship. Steph Curry has made 392 this season with one game left.

— Eddy Elfenbein (@EddyElfenbein) April 12, 2016

The fact of the matter is today’s NBA is dramatically different than it was 30 years ago when the Lakers prevailed or 20 years ago when the Chicago Bulls were setting the NBA record for 72 wins in the regular season. In short, comparing Steph Curry’s Golden State Warriors to Michael Jordan’s Chicago Bulls is an exercise in apple and orange comparisons.

The same is often true in the financial markets as well. Analysts always want more data. That is why having stock, or bond market, data that stretches not just decades or centuries is so attractive. Unfortunately that data is not only suspect but it is also highly questionable whether it is even relevant. In my book I wrote:

There are two issues with going back this far to generate long-term estimates of the equity risk premium. The first is that there are real methodological issues with trying to extract data from back then. There are no computer files one can consults to generate these data time series. Jason Zweig at the Wall Street Journal looked at many of these issues and found any data before 1870 to be suspect. In short, we have a lot less reliable data than commonly thought.

Second and more importantly what relevance do equity returns from these early periods really have to us today? The process of equity investing was very different back then. It took a much more intrepid investor to invest in equities back then. Information was scarcer and the costs were much higher. Equities must have seemed like a much more speculative proposition back then than they do today. Today we have nearly unlimited data, nearly free commissions and the ability to trade stocks with a mouse click. Today’s equity investor is no longer facing the hurdles an investor a century ago would have faced. We should therefore take these long run estimates of the equity risk premium with a big grain of salt.

To avoid the dreaded apple vs. orange problem, what period is relevant when making time-series comparisons?

I’m skeptical of any market study involving data before WW2. I’m naming this the Ben Carlson Rule @awealthofcs.

— Eddy Elfenbein (@EddyElfenbein) April 12, 2016

Ben Carlson at A Wealth of Common Sense took up this topic this week in a post looking at the secular decline in costs for investors. The argument being that over time the cost of being an equity investor has declined. This could very well show in up declining nominal equity returns. Carlson writes:

When people look back at the historical stock market returns I think they often forget that there were no discount brokerages or Vanguard index funds to make things easier for investors. It’s quite possible that investors actually required a much higher gross rate of return in the past because it was far costlier to own stocks. This is one of the reasons that dividend yields were so high historically. Companies had to convince people that it was worth it to own equities over fixed income.

Portfolio construction decades ago was also more difficult. We now take for granted nearly free index ETFs. People often forget that until the late 1950s that common stock dividend yields were higher than 10-year Treasury yields. In short, things change. Time marches on and makes comparisons between different market eras problematic at best.

It may be to argue which team from different eras is better. It often depends on your age and perspective. At least the hosts (and callers) into sports radio shows recognize that these arguments are theoretical and for fun. Don’t let yourself get sucked into those same arguments when it comes to the markets. Because in the markets your assumptions actually matter.