Overall this is related to the concept of the term premium and why on average we've seen higher returns from longer-term bonds than shorter-term bonds in the 20th century. It's a driving force behind why one would use bonds rather than cash to derisk portfolios.
As Ben Bernanke put it in a 2015 article here,
The term premia of Treasuries across the yield curve cannot be directly observed, but can be estimated. The higher the value, the lower the price (higher yield) for longer-term bonds relative to what you'd get investing in short-term bonds and rolling those over.To explain the behavior of longer-term rates, it helps to decompose the yield on any particular bond, such as a Treasury bond issued by the US government, into three components: expected inflation, expectations about the future path of real short-term interest rates, and a term premium.
Briefly, a term premium is the extra return that lenders demand to hold a longer-term bond instead of investing in a series of short-term securities (a new one-year security each year, for example). Typically, long-term yields are higher than short-term yields, implying that term premiums are usually positive (investors require extra compensation to hold longer-term bonds instead of short-term securities).
One of the more-cited models for the yield curve term structure these days, building on decades of work since the expectations hypothesis and other earlier ideas that are widely regarded as empirically untrue now, is from Adrian, Crump, and Moench (ACM) of the New York Fed. They have a post here with some details, a link to updated data, a link to a paper explaining the estimation methodology, and a comparison to some other models and estimates.
For example, one alternative way to derive a term premium estimate would be to take expert surveys of expected short-term interest rate paths and compare those to current longer-term rates. Not all of the methods and models agree, and any one model may be substantially wrong.
All that said, I thought it would be interesting to go back and check what the ACM model is saying these days. I haven't checked it in a while. It does more than just estimate the term premia, but that's one of the things that falls out of it (as the difference between fitted Treasury yields and the risk neutral yield).
So I downloaded the data and graphed for the 2-year, 5-year, and 10-year Treasury bonds:
(click for larger, updated 2017-12-12)
Turns out we're back to negative term premia according to this particular model, which effectively posits that currently the market is pricing Treasury bond yields so high that holding short-term T-bills would be expected to return more over time (averaged over potential futures). I can't tell you if this is actually correct, and I have no comments on the methodology for now, but I thought it was worth at least a look and possibly some discussion. No, I am not suggesting wholesale market timing.