Guest post: A former Fed insider explains the internal debate over QE3

In this guest post, Bill Nelson, formerly a deputy director of the Federal Reserve Board’s Division of Monetary Affairs and now the chief economist of The Clearing House, explains how Fed officials thought about the benefits and risks to an open-ended asset purchase programme.

Last month, the Federal Reserve released transcripts, presentations, and Tealbooks for the 2012 Federal Open Market Committee meetings. While the Committee made a number of momentous decisions over the course of the year, the discussion around the decision to start the flow-based asset purchase programme (aka “QE3” or “QE infinity”) is required reading for understanding the Fed’s actions over subsequent years.

Perhaps most importantly, QE3 forced the Fed to change its plans for normalising its balance sheet.

In September 2012, the Fed began buying $40bn of mortgage bonds issued by Fannie Mae and Freddie Mac each month. This was in addition to the $45bn in longer-term Treasury securities the Fed had already been buying every month as part of the maturity extension programme (MEP), which had started in September 2011. The longer-term Treasury purchases were being offset by sales of shorter-term Treasury securities.

In December 2012, the Fed announced that it would keep buying $45bn of longer-term Treasury securities every month even after the MEP expired, but would no longer sell an equivalent amount of shorter-term Treasury securities. The Committee also indicated that it would continue its bond-buying until there was a substantial improvement in the outlook for the labor market, subject to periodic reviews of the efficacy and costs of the program. This made it an open-ended programme, unlike the first two rounds of large-scale asset purchases.

The Committee entered into “QE infinity” for the same reason it conducted the earlier bond-buying programs: by taking longer-duration assets out of private hands, the purchases were intended to reduce longer-term interest rates and stimulate economic growth. The complications of the program stemmed from the uncertainty about its eventual size, which depended on how long the asset purchases continued.

The flow-based asset purchase program was first introduced as a possibility as Monetary Policy Alternative A in the Tealbook B provided to the FOMC for the June meeting. Broadly, Tealbook B discusses monetary policy while Tealbook A provides the economic forecast.

As background material for each FOMC meeting the staff provides the Committee policy alternatives in Tealbook B, almost always three alternatives labelled A, B, and C. The alternatives are developed in consultation with the Chairman. B is the preferred alternative, with A the dovish alternative and C the hawkish alternative. Usually A and C are chosen to cover the range of alternatives desired by the Committee members, but sometimes they serve as staging areas for potential future policy.

The Tealbooks are distributed several days in advance of the meeting; over the subsequent days the alternatives can evolve in response to further input from FOMC participants, although they often don’t. Tealbook B includes forecasts for the Fed’s balance sheet for each alternative. The staff projections for the flow-based purchases depended on their size and on when they ended.

When the flow-based programme was discussed as Alternative A in the July/August meeting, staff projected that the programme would continue through the third quarter of 2013 and the purchases would total $1tn. But when the program was proposed as Alternative B in September, staff projected that it would only continue through June 2013 and total $750bn. The staff projection for the October and December FOMC meetings also included a June 2013 end date for the purchases. (The programme eventually lasted until October 2014.)

The big complication of the programme was that it was projected to generate losses. In June 2011, the FOMC had published its plans for normalising the balance sheet once stimulus was no longer needed. This bit was the source of the problem:

Sales of agency securities from the SOMA will likely commence sometime after the first increase in the target for the federal funds rate…Once sales begin, the pace of sales is expected to be aimed at eliminating the SOMA’s holdings of agency securities over a period of three to five years, thereby minimizing the extent to which the SOMA portfolio might affect the allocation of credit across sectors of the economy. Sales at this pace would be expected to normalize the size of the SOMA securities portfolio over a period of two to three years.

As the Fed expanded its holdings of mortgage-backed securities under QE3, the projected sales became substantial. The problem was that the Fed expected to be selling its MBS when interest rates were higher than when it had bought them, which would generate capital losses. In some periods, especially under scenarios where interest rates rose sharply, those losses would surpass the net interest income the Fed was expecting to make. In other words, the Fed’s net income was projected to be negative.

The staff briefed the Committee on the potential for losses in both September and December. Under the Fed’s accounting rules, such losses would not reduce the central bank’s capital, but rather would result in the accumulation of a negative asset that reflected the Fed’s plans to retain profits in the future.

The longer QE3 lasted, the bigger the losses from bond sales under the original normalisation plan. If the purchases continued past June 2013, the size of these projected losses would make normalisation infeasible. As the Fed’s holdings of MBS grew, the volume of sales necessary to get rid of all the securities within five years also increased. The pace of sales under longer-term QE3 scenarios became so large that it had the potential to disrupt MBS markets.

A number of FOMC participants, including Governors Powell, Stein, and Duke and Presidents Pianalto, Fisher, and Lockhart expressed concern about the program. Several indicated that while they were willing to support the program, they were not sure the benefits would exceed the cost if the programme grew substantially larger than forecast by the staff.

Moreover, they were sceptical the programme would really end by mid-2013 because the staff forecast for the unemployment rate and the projected monthly gains in employment in mid-2013 both showed no improvement. For instance, Powell asked “How is this not a $1 to $2 trillion LSAP [large scale asset purchase program]? Where is the improvement in labor markets?” Stein indicated a concern that “a muddled mix of small disappointments” could result in a Groundhog Day situation where the purchases continue long past mid-2013.

The forward guidance about the purchases included in the FOMC’s post-meeting statement included an important escape hatch: the purchases could be stopped if they were found to be ineffective or too costly. A number of participants noted it would be difficult to explain why a key monetary policy tool was not working properly.

Fisher worried that once the purchases started they would be hard to stop without disappointing markets. Duke was sceptical the FOMC would be able to adjust market expectations for the purchases to continue without significant disruption. She thought it was more likely that the FOMC would move toward market expectations for the program than the market move toward FOMC expectations.

(The so-called “taper tantrum” of mid-2013 suggests Duke’s scepticism was well-founded. Chair Yellen has managed to avoid a second tantrum, however, as maturing bonds have been allowed to gradually roll off the Fed’s balance sheet without incident.)

FOMC participants’ concerns about the program are reflected in their comments in their inputs for the Summary of Economic Projections, the quarterly report on FOMC participants’ forecasts for policy and the economy. While a summary of the forecasts is provided with the minutes of the meeting (and at the post-meeting press conference), individual forecasts and comments (on an anonymous basis) are released along with the FOMC transcripts five years later. The identities of the FOMC participants who made each forecast and comment are provided only after 10 years.

In September and December, participants were asked if their assumptions about the Fed balance sheet differed from the staff’s. The instructions provided the participants have not yet been released to the public, but the comments suggest some participants were indicating what they considered to be the path for the balance sheet under appropriate policy, while others seem to be reporting a simple projection.

In both September and December, the staff assumed the Fed would purchase $85bn of bonds each month through June 2013. In September, 11 of 19 FOMC participants disagreed with the staff projection and in December, 12 of 19 participants disagreed. Of the 12, three thought additional purchases were a mistake and assumed they would end immediately or very soon. Nearly all of the remainder indicated they expected the asset purchases to continue for longer than assumed by the staff, with most expecting the purchases to continue through the end of 2013.

Chairman Bernanke and President Dudley spoke to participant concerns largely by pointing to the “efficacy and costs” escape hatch. If the program didn’t appear to be helping or appeared to be generating market functioning or financial stability problems, it could be stopped even if labor market improvements hadn’t yet been achieved. Bernanke indicated he would emphasise the escape hatch at his post-meeting press conference (which he did). Dudley said any disparity between market and FOMC expectations could be “very easily managed.”

Both Dudley and Bernanke argued the Fed could avoid losses by altering the normalisation plan. Instead of selling its assets, the central bank could simply let them roll off as they matured, or even reinvest the principal from maturing bonds into new debt indefinitely. Dudley noted this could be consistent with the Committee’s need to eventually raise interest rates. The Fed could pay additional interest on excess reserves to conduct monetary policy rather than try to return to the pre-crisis method of limiting bank reserves and targeting the Fed funds rate.

When the FOMC announced the first leg of the new asset purchase program at the end of the September meeting, the market read the statement as consistent with purchases well past mid-2013. The FOMC measures market expectations about policy largely through the New York Fed’s survey of primary dealers before each meeting.

After the September meeting, which announced the new mortgage bond purchases, the New York Fed conducted a snap survey on views about the likely impact of the new flow-based purchase program. In reaction to the announcement, broker-dealers raised their expectations that the Fed would expand its balance sheet $1tn. After the December 2012 FOMC decision, which added in additional Treasury purchases, traders told the NY Fed the open-ended bond-buying programme would last until the beginning of 2014.

As it turned out, purchases ultimately lasted until the end of October 2014. That was 15 months later than originally than projected by the staff. In total, the Fed bought $1.75tn in mortgage bonds and Treasury debt — more than double the amount originally projected.

The massive size of the programme and the fear of realised losses from selling bonds early encouraged the Committee to change its normalisation principles in September 2014. Rather than selling its MBS within five years, the Committee now indicated it would simply allow the portfolio to shrink as assets matured. In the minutes to the meeting (the transcript is not available), concern about the possibility of making losses is cited as a reason for the change.

The experience with QE3 also helped convince the Committee to keep the balance sheet larger than it was before the crisis, as Dudley had foreshadowed in 2012.

Back in 2008, staff discussions of the post-crisis monetary policy framework had focused on how to return to the pre-crisis model as quickly as possible. But when the Committee discussed staff work on frameworks in 2015, it indicated that it preferred to work with a larger balance sheet using interest on excess reserves and other tools, rather than the pre-crisis framework.

Relatedly, staff projections for the balance sheet before QE3 began assumed that banks would only hold $25bn in reserves on deposit at the Fed. The latest official projections of the Fed’s balance sheet now assume banks will hold $500bn in reserves.

The Fed is continuing to assess and debate its monetary policy framework. It is worth keeping in mind that the Fed didn’t make an explicit decision to keep its balance sheet so large for so long because doing so would support efficient monetary policy. Instead, the Fed fell into its current situation because the original plan to drain excess reserves and sell assets became untenable once people realised selling such a large portfolio so quickly would generate large losses.

Bill Nelson is Executive Managing Director, Chief Economist, and Head of Research, The Clearing House Association, and Chief Economist of The Clearing House Payments Company. Prior to joining The Clearing House in 2016, Mr. Nelson was a deputy director of the Division of Monetary Affairs at the Federal Reserve Board where his responsibilities included monetary policy analysis, discount window policy analysis, and financial institution supervision. Mr. Nelson attended Federal Open Market Committee meetings and regularly briefed the Board and FOMC.