When the Federal Reserve was founded in 1913, there was no role for it in the marketing and selling of government debt. This wasn’t an oversight on the part of Congress. For more than two years before the Fed, the Treasury Department hadn’t issued any marketable instruments at all. In those days it just didn’t seem a necessary function.

World War I changed that, and it changed the way the government would be financed in this modern era. The Federal Reserve banks themselves, the twelve individual branches of the reserve system, were expected to own operating assets so as to generate enough income (and throw off sufficient liquidity as a byproduct) to cover costs. The original idea was local government support as well as bankers’ acceptances. By the 1920’s, however, the branches had been given a heavy taste of Treasuries for this role.

In 1922, the twelve branches formed the Committee on Centralized Executions of Purchases and Sales of Government Securities by Federal Reserve Banks to do what the committee’s long name specified. The original aim was to combine the order function of the various banks into a central space, to execute individual branch trades rather than conduct policy.

It was replaced rather quickly once the Executive Board in Washington realized the twelve were doing these things outside of its purview. Though it wasn’t a part of policy, it very well could have been used that way. The Open Market Investment Committee subsequently created was identical to the previous version with but the one exception, that it operated under the “general supervision” of the Board.

Marketable debt outstanding had grown throughout the twenties, but it was, obviously, in the thirties where these roles really mattered. Whereas in 1930 there were approximately $14.4 billion of UST’s in that category, by 1940 the sector had more than doubled to $33.8 billion. That required not just the central bank acting a part, but also the banking system in general.

There had been no formal network of institutions providing similar and corresponding resources. Nonetheless, as FRBNY stated in a 1940 report, “it has always been the policy of the Federal Reserve System not to deal directly with ‘investment’ holders of Government securities but only with … dealers and other merchandisers.” The New York branch had been given the role of System Account in the Great Depression reforms of the 1930’s, specifically the Banking Act of 1935, under the new Federal Open Market Committee (FOMC) policy structure.

To the FOMC and the Open Market Desk, what counted for a primary dealer was: 1. A firm with an excellent reputation; 2. Capital; 3. Ability to make markets on both the long and short sides; 4. Volume.

The dealer network has become more formalized in the decades since its early days. Its role remains largely the same at least in terms of what it does in context of monetary policy. The dealers themselves, however, are not the same.

By that I don’t mean merely the names on the first informal list submitted for review in 1939. Most of them are but footnotes in long ago history (the first eight to be officially recognized were: Bankers Trust, CF Childs, CJ Devine, Discount Corp, First Boston, Guaranty Trust, New York Hanseatic, and Salomon Brothers & Hultzer), but distinct for what they were. These were purely securities firms whose business was to buy paper, warehouse that paper, and then sell it over time to customers.

They were not banks for a reason, and not just because of Glass-Steagall prohibitions. It used to be common knowledge, wisdom even, that it wasn’t wise to mix banking with non-banking finance.

All that began to change in the 1980’s and 1990’s. I’ve written before about Solly, the colloquial name of Salomon Brothers, a firm that survived a lot of things but really lost everything in 1990 and 1991 (bailed out by Warren Buffet) because of Treasury auctions. Nobody could figure out why the firm’s traders, one in particular, would risk so much for what appeared to be so little.

It is easier today to see this with much greater clarity, as the wholesale banking system is now fully revealed (to those that want to make even slight inquiry), but the contemporary haze should not excuse lack of appreciation then. There is great significance of government and agency debt at auction and issuance, as it is on-the-run securities that control the repo environment. A bond, note or bill just auctioned is the most liquid because it contains the most direct and quantifiable characteristics; once a security is replaced by the next auction in the series, that security becomes highly liquid OTR and the previous fades into trading obscurity (off-the-run). In short, the frenzy over OTR is repo at a time when collateral wasn’t as widely available and the limited OTR’s were quite limited (a shortage the bubbles, greater sovereign issuance and securitizations would eventually but temporarily overcome).

It wasn’t just Solly, though, nor was it limited in scope to just UST collateral. The US government report on the affair examined the books of MBS auctions and found shocking results.

Some traders added random amounts to their actual customer orders. Others increased the number and amount of customer orders reported to the GSEs to include “anticipated” or “historic” sales, i.e., an amount that the trader believed, based on past experience, the selling group member would be able to sell after the GSE announced the price. Even in those instances where a selling group member had identifiable customers for the number and amount of the customer orders reported to the GSEs, the trader would not indicate to the GSEs that many of the orders were subject to significant conditions.

Everybody, or almost everybody, it seemed had in the early 1990’s resorted to cheating at various auctions to claim as much of this pristine OTR collateral as they could. Many of these MBS dealers had resorted to keeping two sets of books, one for the regulators and the other for internal securities management purposes.

The mechanics of being a dealer was at the start truly simple. The Federal Reserve would aid in the process by funding this warehousing of paper. Dealers would buy securities from the government or whatever agency (or private bonds later on) and then repo them back to the Fed or other dealers (with spare cash) as a way to finance them while they were being sold through brokerage. Since the funding arrangements were fully collateralized and often very short term, it meant some of the cheapest funding available.

By the 1980’s, it was significantly less than deposit funding (after the demise of Regulation Q) such that banks wanted in on repo, too. That meant a robust underground of sorts, where dealers could, for a fee, meaning spread, lend out their warehouse inventory to banks seeking repo financing. The 1990’s were a good time to be in securities lending, so long as, unlike Solly, you didn’t get caught being too aggressive.

This is how the dealer network was positioned throughout the late eurodollar period. Eurodollar is in my use an imprecise term, as the system overall became far more than just overseas deposits of dollars. It has become the emblem of wholesale or shadow functions that similar to the original eurodollars are freed from geographical constraints. So long as you are in that business, a dealer might conduct securities lending with a bank anywhere in the world.

As spreads on securities lending thinned out in the 2000’s, partly due to the interest rate environment, perhaps more so because of the entry of other institutions with vast pools of otherwise idle securities (insurance companies like AIG), it became standard practice to lend collateral to more than one counterparty simultaneously. This rehypothecation would create ownership entanglements when presented with the prospect of bankruptcy, before 2007 thought to be a very remote possibility.

Dealers who had been freely lending UST paper, including bills, by the week of August 9, 2007, shifted – catastrophically. The level of their net long or short did not change based on interest rate considerations but collateral concerns that included counterparty determinations (if your collateral would get stuck at a firm about to declare bankruptcy and what that might do to other firms in the rehypothecation chain) as well as internal deliberations about a liquidity safety net. Better to hoard your own warehouse paper for the looming storm than risk lending it out several times and losing it to “bad banks” and bankruptcy petitions.

For a time in the wake of QE3, like the economy it appeared as if the dealer network might actually get back to normal. It didn’t last, however, and for whatever proximate reason(s) in June 2014 dealers stopped lending more warehoused paper (on net) while repo fails suddenly spiked (the two are very much related). The “rising dollar” had begun.

From this point of view, it is difficult to say that it had ever ended. There was a decline (less net long) from early 2016 through the end of last year, but nowhere near as clear cut as the “reflation” in 2013 that brought dealer positions net short once again. Befitting the smaller stature of the 2016 “reflation”, there really wasn’t much collateral enthusiasm to de-hoard.

Beginning in late February 2017, oil prices suddenly shifted lower. That followed six months or so of little additional price gains beyond the first part of the rebound from the ultimate low in February 2016. Ever since that time, there has been noticeable magnetism between T-bill (equivalent) yields and oil prices. So it isn’t exactly surprising to find that during this time of closer correlation between what might seem unrelated financial inputs dealer hoarding of UST’s, including bills, has gone back to more extremes again.

Sure enough, oil prices that turned lower last week again have brought down with them bill yields. The 4-week bill was on Friday 5 bps below the RRP as the so-called floor despite those things being otherwise equivalent monetary alternatives. Why would anyone lend to the federal government in a 4-week bill at 95 bps annual yield when they could alternately lend to the federal government through the Federal Reserve at 100 bps fully collateralized by Treasury paper?

The 3-month bill is similarly threatening the floor, 102 bps Friday and 104 bps today. Dealer statistics are several weeks in arrears, so we don’t know if repo fails have risen that much more, but we do note that they have been moving upward unsurprisingly since March during this dealer re-hoarding trend.

Is it oil and thus collateral swaps that threaten repo collateral streams, or is it repo tightness through collateral that makes funding oil futures and positions that much more difficult (forcing some crude to be liquidated during these times)? Or is WTI merely a signal for more general propositions (if oil is rising, perhaps dealer risk budgets would rise, too, pushing them to be more aggressive for what $100 oil might mean beyond just inventories in Cushing, OK)?

We are left to ask but “why?” There is no 2a7 bogeyman in 2017 to offer benign comfort. Regulatory changes (capital and efficiency numbers that make dealer capacities more expensive) are now long since passed and digested, with only whispers and dreams of their possible removal on the horizon now.

Some will surely claim that the Fed’s exit is to blame, because many still believe (are taught to believe) the Fed actually matters like it once did. The simple truth will never be acceptable to these people, even though as far back as the days of Solly the Fed was but an afterthought. Alan Greenspan could tinker with the federal funds rate, but everyone inside knew that it was elsewhere money in money was being made (and cheated).

For all that the “rising dollar” was supposed to have ended last year, in this very important place (these places) it doesn’t seem to have received the memo. Dealers are clearly reluctant again about systemic liquidity factors, and that matters – quite a bit, actually. It really doesn’t look like much has changed this year at all.