The St. Louis Fed's Regime-Based Approach

St. Louis Federal Reserve President James Bullard today presented “The St. Louis Fed’s New Characterization of the Outlook for the U.S. Economy.” This is a change in how the St. Louis Fed thinks about medium- and longer-term macroeconomic outcomes and makes recommendations for the policy path.

“The hallmark of the new narrative is to think of medium- and longer-term macroeconomic outcomes in terms of regimes. The concept of a single, long-run steady state to which the economy is converging is abandoned, and is replaced by a set of possible regimes that the economy may visit. Regimes are generally viewed as persistent, and optimal monetary policy is viewed as regime dependent. Switches between regimes are viewed as not forecastable.”

Bullard describes three “fundamentals” that characterize which regime the economy is in: productivity growth (high or low), the real return on short-term government debt (high or low), and state of the business cycle (recession or not). We are currently in a low-productivity growth, low rate, no-recession regime. The St. Louis Fed’s forecasts for the next 2.5 years are made with the assumption that we will stay in such a regime over the forecast horizon. They forecast real output growth of 2%, 4.7% unemployment, and 2% trimmed-mean PCE inflation.

As an example of why using this regime-based forecasting approach matters, imagine that the economy is in a low productivity growth regime in which the long-run growth rate is 2%, and that under a high productivity growth regime, the long-run growth rate would be 4%. Suppose you are trying to forecast the growth rate a year from now. One approach would be to come up with an estimate of the probability P that the economy will have switched to the high productivity regime, then estimate a growth rate of G1=(1-P)*2%+P*4%. An alternative is to assume that the economy will stay in its current regime, in which case your estimate is G2=2%<G1, and the chance that the economy switches regime is an “upside risk” to your forecast. This second approach is more like what the St. Louis Fed is doing. Think of it as taking the mode instead of the expected value (mean) of the probability distribution over future growth. They are making their forecasts based on the most likely outcome, not the weighted average of the different outcomes. 

Bullard claims that P is quite small, i.e. regimes are persistent, and that regime changes are unforecastable. He therefore argues that "the best that we can do today is to forecast that the current regime will persist and set policy appropriately for this regime."  The policy takeaway is that “In light of this new approach and the associated forecast, the appropriate regime-dependent policy rate path is 63 basis points over the forecast horizon.”

The approach of forecasting that the current regime will persist and setting policy appropriately for the current regime is an interesting contrast to the "robust control" literature. As Richard Dennis summarizes, "rather than focusing on the 'most likely' outcome or on the average outcome, robust control argues that policymakers should focus on and defend against the worst-case outcome." He adds:

"In an interesting application of robust control methods, Sargent (1999) studies a simple macro-policy model and shows that robustness, in the “robust control” sense, does not necessarily lead to policy attenuation. Instead, the robust policy rule may respond more aggressively to shocks. The intuition for this result is that, by pursuing a more aggressive policy, the central bank can prevent the economy from encountering situations where model misspecification might be especially damaging."

 In Bullard's Figure 1 (below), we see the baseline forecast corresponding to continuation in the no recession, low real rate of return, low productivity growth regime. We also see some of the upside risks to the policy rate path, corresponding to switches to high real rate of return and/or high productivity growth regimes. We also see the arrow pointing to recession, but the four possible outcomes associated with that switch are omitted from the diagram. Bullard writes that "We are currently in a no recession state, but it is possible that we could switch to a recession state. If such a switch occurred, all variables would be affected but most notably, the unemployment rate would rise substantially. Again, the possibility of such a switch does not enter directly into the forecast because we have no reason to forecast a recession given the data available today. The possibility of recession is instead a risk to the forecast."

 In a robust-control inspired approach, the possibility of switching into a recession state (and hitting the zero lower bound again) would get weighted pretty heavily in determination of the policy path, because even if it is unlikely, it would be really bad. What would that look like? This gets a little tricky because the "fundamentals" characterizing these regimes are not all fundamental in the sense of being exogenous to monetary policy. In particular, whether and when the economy enters another recession depends on the path of the policy rate. So too, indirectly, does the real rate of return on short-term government debt, both through liquidity premia and through expected inflation if we get back to the ZLB.

Presidential Candidates and Fed Accountability

In an interview with Fortune, Donald Trump gave his views on  Federal Reserve Chair Janet Yellen, who will come up for reappointment in 2018. "I don’t want to comment on reappointment, but I would be more inclined to put other people in," he remarked, despite his opinion that Yellen "has done a serviceable job."

A change in the political party in power does not always result in a new Fed chair. Yellen's predecessor, Ben Bernanke, was first appointed by President George W. Bush and later reappointed by President Obama. Obama remarked, upon reappointing Bernanke in 2009, that "Ben approached a financial system on the verge of collapse with calm and wisdom; with bold action and out-of-the-box thinking that has helped put the brakes on our economic freefall."

Time reported in 2009 that "The Fed chairman is often described as the second most powerful U.S. official; the main check on him is the first most powerful official's power not to reappoint him. That power won't be used this year, and it's easy to see why. But someday, a President may have to use it..." I have written before that Fed accountability is a two-way street requiring diligence on the part of both the Fed and Congress. But the President also plays a role in checking the Fed's power. Just how far should a (prospective) President go?

Recently, Narayana Kocherlakota, who was President of the Federal Reserve Bank of Minneapolis from 2009 through 2015, has been urging Presidential candidates to address their views on the Fed. He proposes five questions we should ask the candidates, including whether they would seek a chair that would want to change the Fed's 2% inflation target, whether they would want the next chair to change the Fed's approach to its full employment mandate, whether they would want the chair to agree with using a Taylor-type rule for monetary policy, and whether they would want the chair to take an interventionist approach in a future crisis.

Good to see Mr. Trump talking about mon. pol. - more Pres. cands need to talk about this issue: https://t.co/1kMKThHmAS

— NRKocherlakota (@kocherlakota009) April 20, 2016

Kocherlakota tweeted, "Good to see Mr. Trump talking about mon. pol. - more Pres. cands need to talk about this issue." This was not Trump's first discussion of the Fed. Trump previously claimed that "Janet Yellen for political reasons is keeping interest rates so low that the next guy or person who takes over as president could have a real problem."

In Trump's Fortune interview, he continued to express some qualms with low interest rates, namely: "the problem with low interest rates is that it’s unfair that people who’ve saved every penny, paid off mortgages, and everything they were supposed to do and they were going to retire with their beautiful nest egg and now they’re getting one-eighth of 1%." However, he also pointed to an upside of low rates, noting that he would like to take advantage of low interest rates to refinance the debt and increase infrastructure and military spending.

Interestingly, neither of Trump's takes on the Fed's interest rate policy are directly related to the Fed's Congressional mandate. He does not evaluate the Fed's success in achieving either price stability or full employment. Rather, he is concerned with the distributional and fiscal implications of low interest rates--areas in which the Fed chair is traditionally reluctant to tread.

The other candidate who has said most about the Fed is Bernie Sanders, who wrote an op-ed about the Fed in the New York Times in December. Sanders' remarks focus mainly on Fed governance and financial regulation, though he also comments on the Fed's interest rate policy:

The recent decision by the Fed to raise interest rates is the latest example of the rigged economic system. Big bankers and their supporters in Congress have been telling us for years that runaway inflation is just around the corner. They have been dead wrong each time. Raising interest rates now is a disaster for small business owners who need loans to hire more workers and Americans who need more jobs and higher wages. As a rule, the Fed should not raise interest rates until unemployment is lower than 4 percent. Raising rates must be done only as a last resort — not to fight phantom inflation.

On Friday, I took my students in my Federal Reserve class at Haverford on a field trip to DC, where we got to meet with Ben Bernanke at the Brookings Institute. I asked Bernanke whether he thought that the presidential candidates should talk about monetary policy and the (re)appointment of the Fed Chair. He agreed with Kocherlakota that candidates should talk about what they would like to see in a Fed Chair, but said that he does not think it's a good idea to politicize individual interest rate decisions, emphasizing that the Fed does not have goal independence, but does have instrument independence. In other words, Congress has given the Fed a monetary policy mandate—full employment and price stability—but does not specify what the Fed needs to do to try to achieve those goals.

Anyone who wants to is welcome to evaluate the Fed on how successfully they are achieving that mandate. Anyone who wants to is also welcome to evaluate the merits of the mandate itself. Different people will come to different evaluations depending on their own beliefs and preferences. But neither of these two evaluations requires an audit of monetary policy by the Government Accountability Office, as both Sanders and Trump have advocated.

Anyone who is dissatisfied with the mandate itself can go through the usual channels of political change in a democracy and pressure Congress to change the mandate. Congress, by design, is susceptible to such pressure: they need votes. Presidential candidates are in a good position to draw public attention to the Fed's mandate and urge change if they believe it is necessary. Sanders, for example, could propose redefining the Fed's full employment mandate to mean unemployment below 4 percent. I'm not quite sure what kind of mandate Trump would support. It is also fair game for any member of the public to evaluate the Fed on how successfully they are achieving their mandate. But Congress does not (or at least, should not) tell the Fed how to set interest rates to achieve its mandate, and Presidential candidates shouldn't either.

Did Main Street Expect the Rate Hike?

Over a year ago, I looked at data from the Michigan Survey of Consumers to see whether most households were expecting interest rates to rise. I saw that, as of May 2014, about 63% of consumers expected interest rates to rise within the year (i.e. by May 2015). This was considerably higher than the approximately 40% of consumers who expected rates to rise within the year in 2012.

Of course, the Federal Reserve did not end up raising rates until December 2015. Did a greater fraction of consumers anticipate a rise in rates leading up to the hike? Based on the updated Michigan Survey data, it appears not. As Figure 1 below shows, the share of consumers expecting higher rates actually dropped slightly, to just above half, in late 2014 and early 2015. By the most recent available survey date, November 2015, 61% expected rates to rise within the year.

Figure 1: Data from Michigan Survey of Consumers. Analysis by Binder.

Figure 2 zooms in on just the last three years. You can see that there does not appear to be any real resolution in uncertainty leading up to the rate hike. Consistently between half and two thirds of consumers have expected rates to rise within the year every month since late 2013.

Figure 2: Data from Michigan Survey of Consumers. Analysis by Binder.

Did the Natural Rate Fall***?

Paul Krugman describes the natural rate of interest as "a standard economic concept dating back a century; it’s the rate of interest at which the economy is neither depressed and deflating nor overheated and inflating. And it’s therefore the rate monetary policy is supposed to achieve."

The reason he brings it up-- aside from obvious interest in what the Fed should do about interest rates-- is because a recent paper by Thomas Laubach of Federal Reserve and San Francisco Fed President John Williams has just provided updated estimates of the natural rate for the U.S. Laubach and Williams estimate that the natural rate has fallen to around 0% in the past few years.

The authors' estimates come from a methodology they developed in 2001 (published 2003). The earlier paper noted the imprecision of estimates of the natural rate. The solid line in the figure below presents their estimates of the natural real interest rate, while the dashed line is the real federal funds rate. The green shaded region is the 70% confidence interval around the estimates of the natural rate. (Technical aside: Since the estimation procedure uses the Kalman filter, they compute these confidence intervals using Monte Carlo methods from Hamilton (1986) that account for both filter and parameter uncertainty.) The more commonly reported 90% or 95% confidence interval would of course be even wider, and would certainly include both 0% and 6% in 2000.

Source: Laubach and Williams 2001

The newer paper does not appear to provide confidence intervals or standard errors for the estimates of the natural rate. As the figure below shows, the decline in the point estimate is pretty steep, and this decline is robust to alternative assumptions made in the computation, but robustness and precision are not equivalent.

Source: Laubach and Williams 2015

Note the difference in y-axes on the two preceding figures. If you were to draw those green confidence bands from the older paper on the updated figure from the newer paper, they would basically cover the whole figure. In a "statistical significance" sense (three stars***!), we might not be able to say that the natural rate has fallen. (I can't be sure without knowing the standard errors of the updated estimates, but that's my guess given the width of the 70% confidence intervals on the earlier estimates, and my hunch that the confidence intervals for the newer estimates are even wider, because lots of confidence intervals got wider around 2008.)

I point this out not to say that these findings are insignificant. Quite the opposite, in fact. The economic significance of a decline in the natural rate is so large, in terms of policy implications and what it says about the underlying growth potential of the economy, that this result merits a lot of attention even if it lacks p<0.05 statistical significance. I think it is more common in the profession to overemphasize statistical significance over economic significance.

Most Households Expect Interest Rates to Increase by May

Two new posts on the New York Federal Reserve's Liberty Street Economics Blog describe methods of inferring interest rate expectations from interest rate futures and forwards and from surveys conducted by the Trading Desk of the New York Fed. In a post at the Atlanta Fed's macroblog, "Does Forward Guidance Reach Main Street?," economists Mike Bryan, Brent Meyer, and Nicholas Parker, ask, "But what do we know about Main Street’s perspective on the fed funds rate? Do they even have an opinion on the subject?"

To broach this question, they use a special question on the Business Inflation Expectations (BIE) Survey. A panel of businesses in the Sixth District were asked to assign probabilities that the federal funds rate at the end of 2015 would fall into various ranges. The figure below compares the business survey responses to the FOMC's June projection. The similarity between businesspeoples' expectations and FOMC members' expectations for the fed funds rate is taken as an indication that forward guidance on the funds rate has reached Main Street.

What about the rest of Main Street-- the non-business-owners? We don't know too much about forward guidance and the average household. I looked at the Michigan Survey of Consumers for some indication of households' interest rate expectations. One year ago, in December 2013, 61% of respondents on the Michigan Survey said they expected interest rates to rise in the next twelve months. Only a third of consumers expected rates to stay approximately the same. According to the most recently-available edition of the survey, from May 2014, 63% of consumers expect rates to rise by May 2015.

The figure below shows the percent of consumers expecting interest rates to increase in the next twelve months in each survey since 2008. I use vertical lines to indicate several key dates. In December 2008, the federal funds rate target was reduced to 0 to 0.25%, marking the start of the zero lower bound period. Nearly half of consumers in 2009 and 2010 expected rates to rise over the next year. In August 2011, Fed officials began using calendar-based forward guidance when they announced that they would keep rates near zero until at least mid-2013. Date-based forward guidance continued until December 2012. Over this period, less than 40% of consumers expected rate increases.

In December 2012, the Fed adopted the Evans Rule, announcing that the fed funds rate would remain near zero until the unemployment rate fell to 6.5%. In December 2013, the Fed announced a modest reduction in the pace of its asset purchases, emphasizing that this "tapering" did not indicate imminent rate increases. The share of consumers expecting rate increases made a large jump from 55% in June 2013 to 68% in July 2013, and has remained in the high-50s to mid-60s since then.

But since 1978, the percent of consumers expecting an increase in interest rates has tracked reasonably closely with the realized change in the federal funds rate over the next twelve months (fed funds rate in month t+12 minus fed funds rate in month t). In the figure below, the correlation coefficient is 0.26. As a back-of-the-envelop calculation, if we regress the change in the federal funds rate in twelve months on the percent of consumers expecting a rate increase, the regression coefficients indicate that when 63% of consumers expect a rate increase, that predicts a 25 basis points rise in rates in the next year.

This survey data does not tell us for sure that forward guidance has reached Main Street. The survey does not specifically refer to the federal funds rate, just to interest rates in general. And households could simply have noticed that rates have been low for a long time and expect them to increase, even without hearing the Fed's forward guidance.  In an average month, 51% of consumers expect rates to rise over the next year, with a standard deviation of 15%. So the values we're seeing lately are about a standard deviation above the historical average, but they have been higher historically. In the third and fourth quarters of 1994, after the Fed had already begun tightening interest rates, 75-80% of consumers expected further rate increases. At the start of 1994, however, only half of consumers anticipated the rate increases that would come.

In May 2004, the FOMC noted that accommodation could “be removed at a pace that is likely to be measured.” That month, 85% of consumers (a historical maximum) correctly expected rates to increase.