“We should continue the MBS purchases into next year and continue the Treasury purchases,” Williams said to reporters today after a speech in Salt Lake City.

The purchases will be warranted until there’s evidence of “a sustained, significant improvement in the labor market,” such as a pace of payrolls growth more than 200,000 jobs a month, Williams told reporters.

“If you talked about 200,000 or 190,000 or that kind of thing, that’s not enough of an improvement,” Williams said. “I would want to see a measurable decline in the unemployment rate and a significant, sustained, well above 200,000 jobs a month on payrolls, and coherence across lots of indicators. Not one that’s strong, but all of them pointing in the same direction.”

Given that the current inflation rate is quite low and is expected to stay low for several years, we have the flexibility to push for more improvement in labor markets than if inflation were not so subdued. My own personal assessment is that as long as inflation and inflation expectations are expected to remain well-behaved in the medium term, we should continue to forcefully pursue asset purchases at least until the national unemployment rate falls below 7.25 percent and then assess the situation.

I think of this number as a threshold, not as a trigger – and the distinction is important. I think of a trigger as a set of conditions that necessarily imply a change in policy. A threshold, unlike a trigger, does not necessarily precipitate a change in policy. Instead, I think of my proposed threshold as follows. Once the unemployment rate declines to this level, we would undertake a full assessment of labor market conditions and inflationary pressures to determine whether further asset purchases are consistent with the desired trajectory for reaching our inflation and unemployment mandates in the medium term. Thus, a threshold precipitates a discussion and a more thorough assessment of appropriate policy, versus a trigger which starts a change in policy.  As an example, suppose we reach one’s threshold unemployment rate but at that time the economy is slowing, and no further improvement in the unemployment rate is expected in the short to medium term. This hypothetical situation would not necessarily imply a change in policy stance, especially if inflation was projected to remain below target.

Let me say also that an unemployment rate of 7.25 sounds high, but achieving an unemployment rate of 7.25 percent would require real GDP growth of roughly 3 percent for a year. That would be growth that is a full percentage point faster than the economy’s so-called “potential” rate of growth, making this a challenge to achieve.

...

Despite the challenge to communicating in simple terms the likely exit strategy, I will say that my own preference would be to continue asset purchases until we had at least reached an unemployment rate of 7.25 percent, given the low rates of inflation we have been experiencing and are likely to be experiencing over the medium term – however, I will say again that this is a threshold, not a trigger. Assuming that inflation and inflation expectations remain well-behaved, at that point the discussion should center on whether overall labor market conditions are consistent with “substantial improvement” – for example, whether the lower unemployment rate reflects job creation rather than reductions in the labor force as discouraged workers stop seeking jobs; whether we are seeing sustained, robust payroll employment growth; and whether we envision continued substantial improvement in labor markets for some quarters to come. Under those conditions, I would stop asset purchases.

Remember that at the beginning of my remarks I characterized my thinking on the meaning of "substantial improvement" as a work in progress. In that spirit, let me share a qualitative framework for defining "substantial improvement."

The starting point certainly should be the headline unemployment rate and the payroll jobs number. The interpretation of movements in these two statistics would be enriched and reinforced by a review of additional data elements.

Here are examples of what I would look for:

First, I would look for lower unemployment rates that are driven by increased flows of job seekers into employment. I would not interpret discouraged workers dropping out of the labor force as a sign of improvement, even if the unemployment rate falls as a consequence.

Conversely, I'd like to see growing public confidence in the labor market as measured by increased movement of people from out-of-the labor-force status into the labor force—that is, growing labor force participation. I would interpret a reduction in the number of marginally attached workers as a sign of improvement, even if the unemployment rate goes temporarily higher.

Third, I'd look for employment gains that are associated with reductions in underemployment. I would interpret a pickup in job growth less positively if it is associated with increases in part-time jobs for people who seek full-time work.

Finally, I'd like to see signs that improvements in all these indicators are gaining momentum and are sustainable. A framework for assessing labor market conditions needs to include forward indicators of labor market health, such as falling claims for unemployment insurance.

We should always judge the appropriateness of the Fed’s policies in terms of how the economy is doing relative to the two Main Street goals that Congress has set for the FOMC. Such a comparison does not suggest that monetary policy is currently too easy.

Another reason why monetary policy has become less effective in stimulating the economy is because the impetus from a given level of monetary accommodation likely has become attenuated—that is, less powerful—over time. Historically, attenuation has not been important because monetary policy typically has not stayed exceptionally easy for long periods of time. But this time is different and that difference may be important.

Let me be clear. This was the time to act. With the problems we face and the potential dangers lying ahead, it is essential to do as much as we can now to bolster the resiliency and vibrancy of the economy. We cannot be complacent and assume that the economy is not being damaged if no action is taken.

If we continue to take only modest, cautious, safe policy actions, we risk suffering a lost decade similar to that which Japan experienced in the 1990s. Underestimating the enormity of our problems and the negative forces holding back growth itself exposes the economy to other potentially more serious unintended consequences. That type of passivity is a gamble that is not worth taking.

My assessment is that the appropriate policy is likely to be tighter going forward than anticipated by the Committee at this point. Thus, I do see some risks to inflation in the longer run given the current stance of monetary policy.

In my view, we are unlikely to see much benefit to growth or to employment from further asset purchases. If I am right, then conveying the idea that such action will have a substantive impact on labor markets and the speed of the recovery risks the Fed’s credibility. This is quite costly: If the public loses confidence in the central bank, our ability to set effective monetary policy in the future will be harmed and households and businesses will feel the consequences.

While these risks are very hard to quantify, it is clear that the larger the Fed’s portfolio becomes, the higher the risk and the potential costs when it comes time to exit. And based on my economic outlook, that time may come well before mid-2015.

I have been persuaded that the problem [of persistently high unemployment] is, to a significant enough extent, one of weak growth that can be ameliorated by prudent monetary policy actions. A stronger overall pace of recovery is central to improvement in the labor market.

Lockhart told reporters after his speech he was skeptical about a proposal made yesterday by Minneapolis Fed President Narayana Kocherlakota that the central bank should hold interest rates at around zero until unemployment drops below 5.5 percent.

“My comment on that is 5.5 percent is much closer to current measures of full employment,” Lockhart said. “I am not so sure that it will not be appropriate to begin the process of tightening before we would get to what we would consider full employment. I don’t want to rule out that could be the case.”

It is possible that these purchases will yield somewhat smaller interest rate declines and may not stimulate economic activity as much as past large-scale asset purchase programs.

I am very pleased that monetary policymakers in the U.S. are proving willing to take difficult actions like these rather than accept the possibility of a long, slow recovery turning into a stagnation that someday earns the dubious title of “Great.”

I felt an urge at the meeting last week to tie the chairman to the mast, Odyssean-style, and to stuff wax in the ears of my fellow committee members, in order to resist the Siren call of further large-scale asset purchases.

But I have no such powers.

 “I question the efficacy of these large-scale asset purchases,” Fisher said. “What we are doing is not having the impact on employment.”