As we approach the rest of the trading day, keep the info from this month/quarter-end bond buying primer in mind.  Remember that the bond market unofficially closes at 3pm Eastern time (due to the 3pm close of CME pit trading as well as simple market convention) even though trading visibly continues through the 5pm close of CME electronic trading.  Bottom line: there can be a gigantic flurry of volume and volatility centered on 3pm, and there won't be any overt correlation to news or data.

So how exactly does month-end affect bond markets?

Money managers are one of the major contingents of bond market participants.  That means that what they do impacts trading levels, and they operate with a slightly different set of motivations at times.  This is most easily seen in the context of 'month-end' buying (also referred to as 'month-end index extensions' or simply 'month-end trading/tradeflows').

Many portfolio managers use industry benchmark indices (like Barclay's Aggregate Bond Index, which also includes specific indices for MBS and Treasuries) to determine the right mix of bonds in their portfolio.  In other words, if a fund purports to hold 40% Treasuries, they're not just holding any random Treasuries, but a specific mix that averages out to the index value.

Index value is all about "duration exposure."  For instance, let's say Barclay's Treasury Index was at 4.0 to start the month, meaning there was an average duration of 4.0 years.  And let's say a money manager holds the right mix of bonds so that the average duration is 4.0 years.  Over the course of a month--all other things being equal--that average life-span would go down to around 3.98 if the investor didn't buy or sell anything in that portfolio. 

Because of this natural decay of duration in bonds, the Index would be drifting down slightly each month and thus need to be 'extended' to bring it back in line with market reality.  Again, if no one was buying or selling bonds, it would simply move lower by 1 month at a time until it reached zero, but because market participants are constantly buying and selling bonds and because the index attempts to capture the entire bond market landscape, it adjusts once a month to bring itself back in line with that entire landscape. 

The bottom line is that Barclays publishes "index extensions" once a month that let investors know how much they have to increase their bond market holdings in order to keep pace with the rest of the market.  That creates net-buying needs in MBS and Treasuries at or near the end of the month (provided the index durations are actually extending.  They don't always, depending on market performance).

Keep in mind that this only affects certain market participants, so it doesn't guarantee positive movement in MBS simply based on the calendar, but it does mean there is usually some extra measure of demand among a portion of the trading community.

A Farmers' Market Analogy for Month-End

Think about month-end buying like a swap meet among subsistence farmers.  Each farmer has plenty of one crop and needs to trade with other farmers to get some diversity in their diet.  The swap meet closes at a certain time of day and if the farmer doesn't get the produce they came for, their family won't have the required nutrition on its table.  As such, the level of activity and deal-making at this swap meet increases as the closing bell draws closer.

In such a swap meet, each farmer would know that other farmers are there with the same goals, but they wouldn't necessarily know how those other farmers are pursuing their goals unless they were standing right next to them when a deal was being made.  For instance, did Old Jimbo trade all of his apples for Oranges earlier in the swap meet because he figured oranges would be in higher demand toward the end and allow him the opportunity to leverage the higher price of oranges toward better deals on the rest of the produce he needs?  

The analogy is meant to illustrate the point that traders know what other traders are doing around month-end, but only in a general sense.  They do NOT know exactly how they are doing it, and that mystery can add to the frenzy.  If we suddenly see Old Jimbo running around frantically within 10 minutes of the closing bell because all he has are oranges, we may well wait to buy our oranges until he's really hard-up, and we may also try to buy up some of the produce we think he needs in order to sell it at a higher price.  

Month-end momentum in bonds can be quite similar.  If we see yields moving in a certain direction with conviction, traders who are more tactical and opportunistic may try to take advantage of that momentum by trading in the same direction of the momentum.  This is one of the factors that typically fuels snowball trades.