Ever since Repo 105 (and long before that), it has been well-known that Primary Dealers enjoy padding their books before the end of every quarter, typically collapsing their asset holdings in the week just before the quarter end in order to have cash on the books, and to make their capitalization ratios appear better than they really are. Well, the "book padding" that just occurred in Q4 2010 was a doozy, with total PD asset holdings plunging by a stunning $126 billion in the past month, the bulk of which was due to a drop in PD holdings of Treasurys. Was this huge selling by the Primary Dealer community, either for window dressing reasons, or due to expectations of future increases in Treasury yields, one of the main reasons for the drop in bond prices? It is unclear, but the massive selling certainly has not helped. And now that window dressing is again over for at least three more months, PD holdings can only go up (or so the myth goes). So with PDs now back with fresh books for 2011, and once again lifting offers, is the sell off in bonds about to be replaced with a major buying spree?
Using New York Fed data, we have compiled weekly Primary Dealer holdings for all of 2010 (and going much further back). The window dressing observations are stunning, and shows that regardless of the short-term scandal over Repo 105, banks continue to pursue short-term book padding strategies. And in Q4 the difference from the peak in asset holdings, attained on November 24, when PDs accounted for $368 billion in total asset, to the drop, which as always just so happens was on the last day of the quarter, was a massive $127 billion, to a low of $241 billion.
The chart below shows total Primary Dealer holdings on a weekly basis, with the EOQ/EOY holdings highlighted in red:
The selloff was particularly evident in Treasury holdings, both Bill and Coupon. The chart below, which shows seggregated and total UST holdings by PDs speaks for itself:
From a weekly fund flow perspectve, the sell off was pretty much across the curve:
For those who want tabular detail, here are all the weekly PD holdings across all disclosed asset classes (equities, of course, are those that get goosed the most courtesy of POMO; these however are never disclosed as they are irrelevant for Fedcollateral purposes, until, of course we have another Lehman, and PDs pledge bankrupt stocks at 100 cents on the dollar with Bill Dudley).
What is notable in the chart above is that PDs year end comnbined holding position is only $22 billion higher than its was at the beginning of 2010, mostly due to an increase in MBS ($16 billion) and Coupons ($15 billion). What is troubling is that PDs actually closed 2010 with a $6 billion lower total position in corporate bonds compared to the start of the year. And this occured in a year in which corporate bonds closed at nosebleed levels, and saw tons of buying by hedge funds and retail. At least we now know who was dumping the hot grenades to the greater corporate bond foolds.
But probably the most dramatic chart is the one showing quarterly window dressing, together with quarter minimum and maximum holdings by PDs. That we closed at the lows as now traditional is not surprising. After all there is window dressing to be done. What is more troubling is that the closing Q4 level was the lowest close in all of 2010, and in fact the closing print of $241 billion was last seen back on December 30, 2009, when total holdings were $218 billion (only to increase by $34 billion in the next week).
Bottom line, now that the PD books are clean, look for PDs to start buying up all the bonds that retail and all other greater fools are finally selling after accumulating for one year, as the primary dealers reestablish much lower cost-bases. It also means that PDs will now be a bullish impact on bond prices. We are curious how this will be reconciled with the Fed's endless POMO machine, which demands that PDs flip at least half of their holdings in the Fed's endless quest for monetization.