Tuesday, November 1, 2016

Yellen Questions

Fed chair Janet Yellen gave a remarkable speech at a Fed conference in Boston. I have long wanted to ask her, "what are the questions most on your mind that you would like academics to answer?" That's pretty much the speech.

Some commenters characterized this speech as searching for reasons to keep interest rates low forever. One can see the logic of this charge. However, the arguments are thoughtful and honest. If she's right, she's right.

The last, and I think most important and revealing point, first:

1. Inflation
"My fourth question goes to the heart of monetary policy: What determines inflation?"
"Inflation is characterized by an underlying trend that has been essentially constant since the mid-1990s; .... Theory and evidence suggest that this trend is strongly influenced by inflation expectations that, in turn, depend on monetary policy....The anchoring of inflation expectations...does not, however, prevent actual inflation from fluctuating from year to year in response to the temporary influence of movements in energy prices and other disturbances. In addition, inflation will tend to run above or below its underlying trend to the extent that resource utilization--which may serve as an indicator of firms' marginal costs--is persistently high or low."
I think this paragraph nicely and clearly summarizes the current Fed view of inflation. Inflation comes from expectations of inflation. Those expectations are "anchored" somehow, so small bursts of or disinflation will melt away. On top of that the Phillips cure -- the correlation between inflation and unemployment or output -- is causal, from output to inflation, and pushes inflation up or down, but again only temporarily.

What a remarkable view this is. There is no nominal anchor. Compare it, say, to Milton Friedman's MV=PY, the fiscal theory's view that inflation depends on the balance of government debt to taxes that soak up the debt, the gold standard, or John Taylor's rule. In the Yellen-Fed view, "expectations" are the only nominal anchor.

Even Fisher’s interest rates have vanished from the economics of inflation. Nominal interest rate = real interest rate plus expected inflation suggests something linking nominal interest rates and inflation, but that's gone too.

You can see also the implication: don’t worry about energy prices and other “disturbances” to inflation. Don’t even worry about "overheated" real economies, temporary Phillips-curve induced bouts of inflation. With "anchored" expectations, the inflation will melt away.

To be sure, “inflation expectations..in turn, depend on monetary policy.” But just how?
“…we need to know more about the manner in which inflation expectations are formed and how monetary policy influences them. Ultimately, both actual and expected inflation are tied to the central bank's inflation target, whether that target is explicit or implicit. But how does this anchoring process occur? Does a central bank have to keep actual inflation near the target rate for many years before inflation expectations completely conform? Can policymakers instead materially influence inflation expectations directly and quickly by simply announcing their intention to pursue a particular inflation goal in the future?”
The two paragraphs together are an interesting melange of old and new Keynesian economics. In full-on new-Keynesian economics, the answer to my question is straightforward. The Fed announces an inflation target, and a rule following the Taylor principle: for each 1% that inflation exceeds or undershoots the target, the interest rate will rise more than 1%. In new-Keynesian models, this leads to inflation or deflation that spirals away from the target. So, this threat of hyperinflation or deflation “coordinates expectations” around the Fed’s target. It’s a Dr. Strangelove sort of target — do what we want or we blow up the world.

What if inflation is so low that interest rates hit zero, as they have for the past 8 years? Don’t worry, sooner or later a shock will come and inflation will rise all on its own, the Fed can start manipulating inflation again. So, expectations that the Fed will in the future go about this Dr. Strangelove business can still anchor expectations of future inflation around the Fed’s target, and, working back, inflation today.

You can tell that this is a step too far for Mrs. Yellen, and most policy people trained in the 1970s (and me too, but for other reasons). Though “marginal costs” enter her Phillips curve, and though expectations of future inflation clearly anchor that Phillips curve, she clearly does not buy the idea that monetary policy affects those expectations by threatening explosive interest rates. Here she clearly has in mind the old-Keynesian view that higher interest rates lower and stabilize subsequent inflation, not the other way around.

Half her heart goes with adaptive expectations — “Does a central bank have to keep actual inflation near the target rate for many years before inflation expectations completely conform?” Anchored expectations come from the Fed’s painful success in the 1980s, and belief that it will do that again. But half her heart goes with the promise of new-Keynesian models: “Can policymakers instead materially influence inflation expectations directly and quickly by simply announcing their intention to pursue a particular inflation goal in the future?”

You see here some of the debate between the traditional ISLM Keynesians and monetarists at the Fed, on one side, and the promise of these new-Keynesian elements on the other. Both sets of traditional models took their adaptive expectations seriously, and worried that any increase in inflation would raise expectations of inflation and off to 1970s we go. Now both sets of traditionalists are the doves.

How much easier it could be to simply announce an inflation target, everyone believes it, and inflation or lack of inflation follows! It’s the ultimate in speak loudly enough and you don’t even need a stick.

I' skeptical. I think people have heard a lot of promises from public officials, and believe nearly none of them. Every year for the last about half-century the secretary of the Treasury has issued a forecast that the deficit will be eliminated one year after the President’s term ends. How many people in the US know the difference between Janet Yellen and Judge Judy? You have to spend a lot of time inside the walls of the Fed to think that Fed announcements of what their inflation target will be 10 years from now makes a difference to anyone but about 100 bond traders.

Our thesis topic for the week: Is it possible to write down this melange of new and old Keynesian models? You are looking for some model in which higher interest rates lower future inflation, which usually takes adaptive expectations, yet an announcement of a target can anchor expectations, which usually takes rational, forward-looking ones. I guess it’s possible to write down any model, so I should qualify, in a simple and vaguely believable way?

(I should put my horse in the race. I think the “anchor” is fiscal policy. Expected inflation is stable so long as people think fiscal policy is in control. That makes Mrs. Yellen right in a lot of ways. However, higher interest rates might make people quickly realize fiscal policy is not under control, which makes her critics’ nervousness also right. But today is not about my answer or the right answer, it's about Mrs. Yellen's and the Fed's views. I say this mostly so that I don’t get counted as one or the other side of the current hawk v dove debate.)

2. The Phillips curve
"While this general framework for thinking about the inflation process remains useful, questions about some of its quantitative features have arisen in the wake of the Great Recession and the subsequent slow recovery. For example, the influence of labor market conditions on inflation in recent years seems to be weaker than had been commonly thought prior to the financial crisis..."
Translation. Inflation just sat there and did nothing in the face of the hugest unemployment we've seen since the great depression. The Phillips curve, relating inflation to unemployment or output, has completely fallen apart.  This being the central piece of economics in Fed story for how it affects inflation — higher rates lead to less output and employment, lower marginal costs, lower prices — we’re a bit befuddled.

The implications of a vanishing Phillips curve are fun to debate. At a recent meeting at the Fed, I opined it was falling apart because huge variation in unemployment correlated with tiny changes in inflation. No, my counterpart said with a wry smile! It means that we can cure unemployment with only half a percentage point more inflation!

Putting the last two observations together, I think we see where the Fed is going. If inflation is just a trend, battered around by commodity prices, anchored by speeches, and immune to anything the Fed actually does; then that frees the Fed from what used to be its main job -- worrying about inflation -- to just worry about real stimulus with no worry about inflation. Moreover, if unemployment can skyrocket with no huge deflation, as it did, then the Fed can push unemployment way down without worry about more inflation, even in the short run. Instead of the Fed mainly determining inflation, with recessions an unfortunate byproduct, we now have a vision of the Fed mainly worrying about real stimulus, and not needing to worry about inflation. The fact that my first point, inflation, was Mrs. Yellen's last, encourages this reading.

3. Hysteresis. Does demand create its own supply? And vice versa.  (Yes, the Say's law echo is intentional.)
."..one study estimates that the level of potential output is now 7 percent below what would have been expected based on its pre-crisis trajectory, and it argues that much of this supply-side damage is attributable to... the deep recession and slow recovery..... a marked slowdown in the estimated trend growth rate of labor productivity. The latter likely reflects an unusually slow pace of business capital accumulation since the crisis and, more conjecturally, the sharp decline in spending on research and development and the very slow pace of new firm formation in recent years." 
It is easy to read this as the latest excuse for dovishness, a new instance of the answer in search of a question. But take the argument seriously. Surely "demand" and "supply" -- poor concepts in the first place -- do leak to each other. If "demand" causes a long depression of investment in human or physical capital, then "supply" will be lower.
"...the natural next question is to ask whether it might be possible to reverse these adverse supply-side effects by temporarily running a "high-pressure economy," with robust aggregate demand and a tight labor market. ...More research is needed, however,.."
I hope the Fed will do that more research before jumping on this theory. Casual investigation of past episodes are not promising. The late 1970s are the textbook case of a "high pressure economy" stimulated by monetary policy and "demand." They did not produce wonders of "supply," either of greater capital or more economic efficiency.
"More generally, the benefits and potential costs of pursuing such a strategy remain hard to quantify, and other policies might be better suited to address damage to the supply side of the economy."
I have noticed a tendency for Fed economists to work hard on issues that have an ear at the top. I  wish Mrs. Yellen had mentioned one or two such "other policies" to reduce the chance that this is interpreted as a throw-away line, not an invitation to write papers proving hysteresis.

4. Heterogeneity.

For nearly a century, the main simplification of macroeconomics has been to gloss over differences between people. "Consumption" and "employment" may be too high or low, but the fact that gains and losses are not spread evenly does not matter, to first order, when understanding the movements of the same aggregates. Note, I do not say they don't matter -- they matter a lot. If one in 10 loses their job, it matters a lot to the person losing the job. The issue is, if you want to know how monetary policy affects average employment or consumption, does it matter that 1 in 10 loses a job and the rest keep their jobs, vs. each of us working 10 percent fewer hours?

Of course it matters, the layman says. But simplification is the key to progress in any science. Chemistry did not get going by working out quantum mechanics. Furthermore, you can see quickly that heterogeneity matters only if economic decisions are nonlinear -- and we know how to model that nonlinearity. Linear decisions add up and behave just like a single household with the average response. Again, the layperson says of course economics is nonlinear. But unless you know just how it's nonlinear, that complication doesn't help. Wrong nonlinearity and state dependence is worse than none at all.

There is a huge new literature on heterogeneity. When a shock hits, some people don't have any savings and have to stop spending, now. Others can dip in to savings. People who lose their jobs are different from people who lose some hours, in big ways.

Watching from afar, I, like Mrs. Yellen, am impressed by this effort, but still scratching my head to understand what it all means for the economy as a whole.
" the various linkages between heterogeneity and aggregate demand [and supply! and equilibrium! Please, Mrs. Yellen, there is more to life than "demand"] are not yet well understood, either empirically or theoretically." 
She continues
"More broadly, even though the tools of monetary policy are generally not well suited to achieve distributional objectives, it is important for policymakers to understand and monitor the effects of macroeconomic developments on different groups within society."
That's another sentence that deserves careful study. It is easy to cross the line from "understand and monitor" to target. It's not just easy, it's inevitable. So, the mandate of monetary policy has stretched from price stability to add low interest rates and maximum employment (by statute), to "financial stability," which now means understanding and monitoring, and inevitably trying to control, asset prices, housing prices, debt levels, bank profits (not yet at the Fed, but clearly on the minds of ECB and BOJ policy), and now will be targeting inequality too. I wish for another throwaway line on "other policies." Yes, Fed policy does affect some people more than others. But if the Fed tries to counter the ill effects of other policies -- bad public schools, say -- by monetary policy, it will first do a terrible job of its main objective, and second it can no longer stay independent and a-political.

5. Finance
In light of the housing bubble and subsequent events, policymakers clearly need to better understand what kinds of developments contribute to financial crises. ...
Research on this topic has, of course, been ongoing for some time, and it has expanded greatly in the wake of the financial crisis. But I believe we have a lot more to learn about the ways in which changes in underwriting standards and other determinants of credit availability interact with interest rates to affect such things as consumer spending, housing demand and home prices, business investment (especially for small firms), and the formation of new firms.
I can hardly object to the idea that we need to better understand how finance links to macroeconomics. Since 2008, about 3/4 of the papers at every conference or job market talk are about putting various financial constraints into economic models. I'm interested that Mrs. Yellen is also still looking for something solid to come out of all this.

Again though, one can be quite uncomfortable with the implicit message that the Fed needs to understand everything and try to control -- or at least monitor -- everything. If 8 years of nonstop research have led to so little, is not the program of understand heterogeneity, nonlinearity, inequality, financial "frictions" and linkages, crises and bubbles, and then masterfully address them all, just a little hopeless?

Would the Fed not do better both economically and as an institution to say, "look, our job is the price level. We take care of inflation, we do it independently and a-politically. We are not the master planner of the economy."

6. General comments

As some of the commenters point out, the speech is pretty remarkable for its implicit admission of the fact that the Fed really has very little idea of how its policies work. The jump from cocktail party speculation about, say "hysteresis" or "secular stagnation" or "anchored expectations" to serious consideration of such effects at high levels is pretty short, by scientific standards.

But in defense of Mrs. Yellen and the Fed, Mrs. Yellen is remarkably in touch with the best there is (such as it is) on these questions. I cannot imagine the top-level administrator of any other government agency, from cabinet secretaries on down, anywhere near as conversant with the state and limitations of current research.

Indeed, there is not great academic research answering these questions. And not for want of effort. Saying snarky things about the state of macroeconomics is easy. Coming up with serious answers to Mrs. Yellen’s questions is a lot harder. She is remarkably honest about her, and the Fed’s, limited understanding of the system they are trying to manage.

Compare the Fed here to the rest of the economic policy world -- the FTCs regulation of mergers (about 50 years behind anti-trust economics and legal scholarship), the FCC's regulation of the internet, the SEC's regulation of financial markets, the FSOC (indulging the Fed) regulation of "financial stability,” the CFPB “protection” efforts and so on. Anchoring, hysteresis and heterogeneity are scientific bedrock compared to "contagion," “liquidity," "abuse" and all the other mumbo-jumbo these agencies think they understand and control. And their pretense of knowledge is off the charts greater than Ms. Yellen's humility.

(Thanks to commenters on a previous post who brought up the speech and have active and thoughtful commentary going on.)


  1. Thanks professor Cochrane, I prefer the Dallas Fed trimmed mean. It’s a more modern method of trying to get the underlying inflation rate, rather than just throwing out components, which is what the core measure does. The core was really invented in the ’70s as an expedient way to do this. There are better ways to do it. And one frustrating thing about the core is that, even though you throw energy out, energy then still impacts some of the other components that are still in there.

    1. A good topic for another blog post... CPI is actual inflation. So why throw things out? I think the point of core is to come up with a simple measure that forecasts future inflation. The idea is that cpi = core + noise, and core is close to a random walk, so core_t is the best estimate of cpi_t+1. But if you state the question that way, there are obviously better answers.

    2. John,

      "So why throw things out?"


      "The report was important because inflation, as calculated by the Bureau of Labor Statistics, is used to index the annual payment increases in Social Security and other retirement and compensation programs."

      That was before Robert "Trader Bob" Rubin introduced TIPs and other inflation "protected" securities in 1997.

    3. The concept of core is useful as a projection tool. Your economic models may have a decent shot at predicting core inflation, while you can let analysts, who follow the weather and Middle East politics, predict food and energy.

    4. Thanks Prof, - That’s right. So you have a whole spectrum of goods, and there’s inflation in every one of those goods, and then you throw out the highs and the lows and you get something that’s more in the middle, and that’s kind of their idea. And this prevents big movements in anything, but often energy, also many other things, from sort of giving you a distorted picture of how inflation is moving day to day. ATB

    5. How do we differentiate between changes in real value and inflation? A good beoming rarer means an increased price but not increased inflation (and vice versa), yet we seem to be happy to conflate the two

  2. my question is are 5000 years of economic theory and results wrong and Yellen and the other group think central bankers right or will we see the mother of all pay backs for the rescue of the crime bankers and various billionaire asset owners. Me think it is the later.... It is so obvious what they are doing is suicide(debt for rescue)...but alas bonuses are wanted at the mega banks and they can buy control... and the richest have bought the best self serving policy makers... You know this won't end well... have you any recollection of a bubble being created to rescue from a bubble burst where long term things end well? What percent of the world's central bankers....hearken from Goldman Sachs...which coincidentally is part owned by the richest man in the world Warren Buffet? What percent of Wall Street have bribed Hillary Clinton with speaking fees? You think you are grumpy?

    bankers and hedgefunds now take unlimited Zero Interest money and gamble and gamble....shear foolishness with no adults or risk in sight! Always a central bank willing to print trillions for gamblers! Maybe more chapters of gamblers anonymous should opened on Wall Street, in London, etc

  3. I don't think that we have a good definition of "inflation" for the purposes of monetary policy. If the price of oil or land or copper goes up as a result of resource depletion, or a growing population, is that "inflation" that should be addressed by the Fed increasing interest rates?

    In the opposite direction, measures of inflation have a lot of trouble capturing or allowing for improvements in quality. If the price of basic health care goes up 10% but it is 20% more effective - is that "inflation" for the purposes of monetary policy?

    Nor have I ever read a plausible explanation about how going to 3% or 4% inflation would somehow provide a free lunch stimulus to the economy.

    Ultimately we seem to have angry arguments about plus or minus 2% on the "inflation" rate when we cannot define or measure it to within that accuracy.

  4. Why isn't inflation measured in a cap-weighted way? Would spending on gasoline, food, healthcare and housing then dominate? Well, yeah. Just like the real world.

    1. Ron,

      See GDP Deflator.


      The interesting part of the GDP deflator is that:
      1. It does not include imported goods (for instance imported oil)
      2. It includes services

      Recognize that an increase in the CPI may not be accommodated by an increase in the GDP deflator.

  5. We are about to hit the biggest hyperinflation in the history. Its very possible that the world of the finances will be forced to give up the current monetary order.

  6. Wow. This is what happens when you dismiss the equation of exchange. It's not an equation, it's an identity and a constraint. It's the ideal gas law of economics. And it says that inflation and gdp growth are negatively related if the LHS is constant, not positively. No such thing as "overheating" unless LHS is moving the chains. If the Fed controls the LHS, the RHS has to fall in line. No inflation. Figure a way to control the LHS. Period. Expectations have to be wrapped up in there somewhere.
    I miss Milty.

    1. This was probably mostly true when most transactions didn't involve and exchange of bank deposits but rather an exchange of bank notes, because deposits interest rates are a function of inflation and will cancel out the opportunity cost of holding wealth vs spending it, doing very little to change behaviour in real terms. Nowadays I think the most important factor is to provide stable long term expectations, because of long term interest rate markets are the only capital market where wealth is negatively correlated to interest rates.

  7. You sound frustrated by the knowledge created by those "8 years of nonstop research" on how finance links to macro. What is the reference point when we try to quantitatively evaluate the macroeconomic research production function? How much time, policy experimentation, research effort and luck did it take since Wicksell to generate knowledge solid enough to convince you that price stability should be the primary concern of the Fed?
    Going forward, if you think that research since 2008 has taken a path that turned out disappointing, do you have thoughts about what the main corrections should be? Macroeconomists interested in finance like Olivier Blanchard have recently expressed opinions on the subject (https://piie.com/system/files/documents/pb16-11.pdf), but I am curious to hear the presumably different perspective of one of the rare researchers who came instead from finance to macro.

  8. Regarding Fiscal policy and linking it with inflation and (more importantly) inflation expectations, an ever increasing debt/GDP ratio will start to increase the growth adjusted real interest rate and will this not raise peoples doubts, eventually causing them to adjust expectations.
    Also, whether or not the phillips curve has fallen apart depends on whether rational or adaptive expectations control the dynamics of it. Surely if we presume rational expectations, then only unexpected changes can alter unemployment?

  9. God is this all so devoid of practicality. Mosler is a great hedge fund billionaire and better economist than you all, because he worked in banks and understands how basic accounting works.

    What are the 2 primary human causes of inflation? (assume monetary / banking function is just basic competent and non fraudulent). We then see these two easily: 1) population growth, and 2) standing armies. Both peaked globally in 1969. Thirteen years later women went back to work, just when Volcker was trying to make America into Brazil.

    And ever since ... there has been zero inflation. Duh. This should be obvious.

    Bonus question - Who does the quality adjustment for the invention of Viagra? Or blood-thinning statins (Lipitor, now off-patent) judged to add more life extension than all heart surgery expenditure, with zero side effects (comparatively). Or a quick #3, free-worldwide video calling. Who did that quality adjustment in the inflation statistics said to be "real"? What a fantasy.

  10. Thank you Professor Cochrane for this amazing analysis. If the Fed is doing an excellent job controlling inflation by anchoring people's beliefs or expectations, would this also mean the Fed can no longer use inflation to affect unemployment rate? Since the omitted variable between inflation and unemployment rate is now fixed. I'm assuming some kind of money illusion is affecting both inflation and unemployment.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.