The Phillips Curve Today: Beware the White Swan

The theory, at its core, is pretty straightforward: businesses compete with one another, and they’re constantly looking for ways to cut costs so they can increase – or maintain – their market share. The bulk of their costs are labor costs – wages and benefits. When the unemployment rate is low, it’s hard to reduce labor costs. Businesses are constantly finding ways to make workers more productive, but during good times, those increases in productivity are eaten up by increases in wages and benefits, which are necessary to retain workers who face relatively abundant alternative opportunities and relatively little competition. When the unemployment rate is high, businesses continue to compete by increasing productivity, but they can also compete by keeping wages down. Under those circumstances, they undercut one another’s prices, and the general price level tends downward.

It gets more complicated, of course. The largest complication is that businesses have relationships – and often contracts – with both customers and employees, and unanticipated changes in prices and wages can disturb those relationships. Consequently, businesses anticipate changes in prices, wages, and market conditions, and they set their own prices accordingly, in the hope of minimizing future surprises. As a result, inflation tends to have momentum. If prices have been rising by 2 percent per year, businesses anticipate that price growth, and the actual inflation rate – under idealized “normal” conditions – comes out close to 2 percent per year. But if the unemployment rate is very low, competition for workers forces businesses to raise prices more quickly, and the inflation rate rises. And if the unemployment rate is very high, competition for customers forces business to raise prices more slowly, and the inflation rate falls.

So much for the theory. I could add a lot more complications – the supply and demand for money, the difference between flexible and sticky prices, the impact of different degrees of competition, the various ways businesses might form expectations about prices, the relationship between unemployment and job vacancies, the possibility of structural changes in the economy over time, and so on – but let’s just stop here and take a look at the evidence in its simplest form. (To produce the chart below, I first took the rate of change in the core CPI from December to December for each year. Then I subtracted the previous year’s rate of change from the current year’s rate of change, for each year in the sample, and I plotted the result against the average unemployment rate for the current year. The core CPI series starts in 1957, so the first observation for which I could compute the change in the inflation rate is 1959. All the underlying data are from the Bureau of Labor Statistics.)



The correlation isn’t perfect – and we wouldn’t expect it to be, since there are other factors that affect the inflation rate in the short run. But it’s strong enough to be quite statistically significant.

And under today’s circumstances, it’s strong enough to be disturbing. The core inflation rate for 2009 was 1.8 percent. If you take the regression line at face value and plug in an average unemployment rate of 9.6 percent – a little toward the low end of what most economists expect for the year – it implies a 1.8 percentage point decline in the core inflation rate. And if you look at the actual data for January through April 2010, we are right on target for a zero percent core inflation rate. I probably don’t have to point out that the unemployment rate will almost certainly still be quite high in 2011, and it won’t be low in 2012.

The May CPI comes out tomorrow morning. It’s expected to show a very slight increase in core consumer prices. If it does show only a very slight increase, or no increase at all, how many will report that “inflation is still under control” and describe it as good news? If you’re worried about the black swan of inflation, I guess it is good news each month that the black swan doesn’t appear. But under today’s circumstances, the white swan – the common species that past experience would lead us to expect – is deflation.



DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.

Second Dip?

As you might surmise from the conclusion of my previous post, I have been worried about the possibility of a second dip, a new recession beginning sometime in the next year or so, before the current recovery has had a chance to produce much improvement. I was surprised to read (hat tip to Mark Thoma’s twitter feed) that Macroeconomic Advisors is suggesting that there is no chance of a second dip. (I was particularly surprised because MA’s own estimates of the growth impact of the waning fiscal stimulus were one of the reasons I was worried.) After reading their case for zero chance, I have to say that I am still worried. Verbally-intuitively, the case for a second dip still seems pretty overwhelming to me. I take comfort in the knowledge that I tend to have a pessimistic bias, and in the fact that sophisticated quantitative models are generally putting the odds of a second dip quite low. On the other hand, successfully forecasting recessions has not been a strong point of quantitative models.

Here is what I see as the case for and against a second dip. As you will see, I am more skeptical about the case against. Maybe someone can tell me what I have overlooked or how I am being too pessimistic.

The Case for a Second Dip

1. The Fed’s policy of quantitative easing, which was temporarily buttressing demand, is over, and its impact will likely decline over time, imparting a downward bias to growth in the coming quarters.
   
   
2. This fiscal stimulus, which was temporarily buttressing demand, has been largely exhausted and has likely reached its point of peak impact (even if additional fiscal measures are taken), so that its impact will be declining in the coming quarters, imparting a downward bias to growth.
   
   
3. Pent-up demand from consumers (many of whom were worried about the losing their jobs last year but no longer are) has been largely exhausted, and its impact will likely decline over time, imparting a downward bias to growth in the coming quarters.
   
   
4. The process of inventory adjustment has run its course, and firms have been able to increase production again to maintain inventories at the new, lower level and to begin slightly increasing inventories in anticipation of a recovery. Significant increases in production are no longer necessary to maintain inventories, so that an upward bias that has been imparted to growth in recent quarters will no longer be present in future quarters.
   
   
5. With the dollar relatively strong again and the pace of world recovery expected to slow, export growth, which had offered the possibility of a robust recovery, no longer seems to offer that possibility.
   
   
6. Normally, the surge in productivity at the beginning of a recovery is followed by a surge in employment. They typical lag is about two quarters. Last year’s surge in productivity took place over the last three quarters of the year, which suggests that a surge in employment should have taken place beginning in the last quarter of last year and continuing through the current quarter. Aside from temporary census employment, the anticipated surge does not appear to be taking place. Meanwhile, productivity growth has settled back into the normal range, which dampens hope for a future surge in employment.
   
   
7. The Bush tax cuts expire at the end of 2010, creating an incentive for high-income individuals (and their corporate agents) to shift income out of 2011 into 2010. To the extent that they are successful in doing so, and to the extent that the shifted income is associated with actual economic activity taking place during the period in which it is declared, we should expect a downward bias to growth between 2010 and 2011. (This point comes from a recent Wall Street Journal op-ed by Arthur Laffer, to which a colleague referred me. People who know my work well know that I have had my quarrels with Arthur Laffer in the past, but in this case, I don’t see any fundamental flaw in his argument.)
   
   
8. Given all these negatives, there is no evidence of any positive stimulus to growth that would offset them. The financial panic of late 2008 subsided long ago, and the residual financial weakness is lifting very slowly, with no suggestion that the pace of improvement will accelerate, especially in the light of potential fallout from financial difficulties in Europe. With capital ratios still an issue, the current regulatory environment is not conducive to rapid increases in bank lending.
   

The Case Against a Second Dip

1. In the years since the Great Depression, there is no precedent for a long recession (longer than 8 months, in this case about 18 months) followed by a short recovery (shorter than 35 months). The two closest “double dip” examples (both with first dips lasting 8 months or less) are 1980 – when the second dip was essentially intentional on the part of the Fed – and 1960 – when the economy had already made nearly a full recovery by the time the new dip happened. On the other hand, double dips appear to have been fairly common in the years before the Great Depression, so the validity of this piece of evidence depends on the premise that something (the fixed gold standard?) fundamentally changed in the 1930’s and has not since reverted.
   
   
2. Recessions seldom begin when the unemployment rate is already high. In particular, since the end of the Great Depression, we have not seen a recession begin with an unemployment rate greater than 7.5 percent. (Today it is 9.7 percent.) Having said that, though, I should note that the second dip of the Great Depression began with an unemployment rate of over 14 percent. (Presumably the reason this happened in the 1930’s is that fiscal and monetary policy were tightened, whereas in subsequent cycles fiscal and monetary policy have generally been loosened when the unemployment rate remained very high. Unfortunately, in the light of the first two points adduced in favor of a second dip, this contrast doesn’t bode well for the immediate future.)
   
   
3. Recessions are normally preceded by stock market declines of greater severity than what we have seen recently. (Of course, if your concern is whether to own stock, the fact that the stock market has not yet had a large decline isn’t much of a comfort.)
   
   
4. Credit spreads do not suggest a high risk of recession. (Again, if your concern is whether to own bonds, this is not much comfort. But perhaps the stock and bond markets should find each other’s lack of severe concern reassuring.)
   
   
5. The price of oil has been reasonably stable, not exhibiting the sort of spike that has helped induce most of the post-WWII recessions. (However, since the second dip, if it happens, is likely to have deflationary characteristics, we need to be concerned that any lack of strength in commodities such as oil could be in anticipation of a second dip.)
   
   
6. The yield curve (difference between long-term and short-term interest rates) is unusually steep. Recessions normally begin with a flat yield curve. Short-term interest rates normally fall during a recession, whereas a steep yield curve suggests rather that short-term rates are expected to rise. However, as Paul Krugman points out, this usual interpretation doesn’t apply now. If there is a second dip, short-term rates will not fall, because there is nowhere down for them to go. Under these circumstances, the steep yield curve likely only indicates the possibility of a rise in short-term rates (without the offsetting possibility of a fall), not the likelihood of a rise. In fact, it could be argued that the steep yield curve is reason to worry more about a second dip: in linear models, a false signal from the unusually steep yield curve could easily outweigh other indicators that are showing valid, but less intense, signs of trouble. (For example, the stock market hasn’t declined dramatically, but it has declined. Should we be worried? Ordinarily, with such a steep yield curve, the answer would be an unambiguous “no.” Today, we’re likely to hear that “no” from linear models, but it could well be based on a single indicator giving a flawed signal.)
   

It’s possible that the case for a second dip is basically right but that we still don’t technically get one. With normal productivity growth and population growth, we could have a severe slowdown, involving maybe one quarter of negative growth, or two quarters of very slightly negative growth, or three quarters of very slightly positive growth, and it might not qualify as a recession. Obviously, it would still suck.

What worries me particularly is that, even if the case for a second dip is completely wrong, the employment picture going forward is still dismal, and there is still a case for deflation. Am I wrong in understanding that this is standard textbook macroeconomics? There is a non-accelerating inflation rate of unemployment (NAIRU). When the actual unemployment rate is above the NAIRU, the inflation rate declines. The further the unemployment rate is above the NAIRU, the more quickly the inflation rate declines. The unemployment rate is currently 9.7% and is not expected to fall rapidly, even under optimistic scenarios. Recent estimates put the NAIRU at about 5%. The current core CPI inflation rate is about 1%. You do the math.



FOOTNOTE: Well, OK, technically you can’t do the math, since I didn’t give you a Phillips curve coefficient. From what I can tell, Phillips curve coefficients are all over the map these days, with some people arguing that the coefficient is zero as long as monetary policy is credible. (But is monetary policy really credible?) At the other end of the spectrum, coefficients with magnitude as high as 0.5 (implying a half percentage point decline in the inflation rate each year for every percentage point that the unemployment rate is above the NAIRU) seem to be well within the mainstream. I recommend against doing the math with that coefficient if you have a heart condition.


DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.

The Mankiw Rule with Quantitative Easing: Why is the Fed So Tight?

In my last post, I suggested that the Fed – at least if it behaves in a reasonable manner consistent with its past practices – is not likely to raise its federal funds rate target any time soon. I argued that the Mankiw Rule (a.k.a. Greg Mankiw’s version of the Taylor Rule) has done a good job of tracking Fed policy in the Greenspan-Bernanke era and that it has now fallen well into negative territory, out of which it will take some time to climb. Some commenters pointed out that, while the Fed obviously can’t make interest rates go negative, it did continue to loosen during the period of zero interest rates, by means of “quantitative easing” or “credit easing” – attempting to pull down the level of riskier or higher maturity interest rates by acquiring unconventional assets. I don’t think this observation really affects the main point of my previous post, but it it’s interesting to take a closer look.

So I tried to come up with a simple measure of monetary policy stance that incorporates both the federal funds rate and quantitative easing. My first thought was to look at the growth of unconventional assets on the Fed’s balance sheet, but as it turns out, it’s not really necessary to specify “unconventional” assets, since the Fed had already reduced holdings of its conventional asset – Treasury bills – to near zero by the time Lehman Brothers failed. We can therefore measure the subsequent quantitative easing as an unusually rapid growth in the Fed’s total assets – or equivalently total liabilities, which is to say, the monetary base. To get a composite measure, we need to somehow graft a measure of this monetary base growth onto the federal funds rate. The simplest way is to subtract the monetary base growth rate from the federal funds rate. Here I have chosen to use the average monthly growth rate over 12 months, because it was a simple specification that gave vaguely reasonable results. Those results are summarized in the chart below.



If you take this chart at face value, it looks like the Fed initially far overshot the level of easing prescribed by the Mankiw Rule, but remember that my choice of equivalence between interest rate percentage points and monthly growth rate percentage points was arbitrary. I could have used a weekly growth rate, and the picture would look quite different. Moreover, I could have used a 3-month or 6-month average instead of 12 months, though I think such choices would only have made the picture look even more strange. The one conclusion that is robust to reasonable changes in specification is that the composite measure is now moving close to zero again, even as the Mankiw Rule interest rate remains well below negative 3 percent. (It will likely rise slightly above negative 4 percent based on the May data, but I’m waiting for the CPI report before I update.) Quantitative easing is over, but the economic conditions that justified it are still with us – at least if we measure retrospectively.

Basically, once we recognize that quantitative easing is an option – and one that is no longer being pursued – we can draw the conclusion that the Fed is much tighter today than what the Mankiw Rule would suggest. Indeed, relative to the Mankiw Rule, the Fed is much tighter than at any time during the Greenspan-Bernanke years. Since 1957, when the core CPI data series begins, there have only been two times when the Fed was as tight as it is today relative to the Mankiw Rule. One was in 1973, when the effect of Nixon’s price controls was artificially reducing the retrospective inflation rate used in the Mankiw Rule. The other was during the early 1980’s, when the Fed was targeting monetary aggregates rather than interest rates and attempting (with great success) to reduce the inflation rate dramatically.

So why is the Fed so tight? Here are some possibilities:

1. Fed policy is better described by a rule that is non-linear in unemployment. With the unemployment rate so tremendously high, perhaps marginal increases in the unemployment rate affect the Fed less than they would if the rate were closer to normal. But given the Fed’s mandate to pursue high employment, wouldn’t the need for more aggressive monetary policy in response to higher unemployment rates be even more acute when the employment situation is already so obviously out of whack? And wouldn’t the unusually high unemployment rate, in and of itself, tend to eliminate the risk of pushing the unemployment rate too low and thereby free the Fed to pursue more aggressive policies than it otherwise would?
   
   
2. The Fed is anticipating dramatic declines in the unemployment rate and/or increases in the inflation rate. Except that we don’t see those in the Fed’s forecasts.
   
   
3. The Fed is correcting for its earlier overshoot, for being too loose in 2009. Except we’re not seeing much evidence that the overshoot (if there was one) needs to be corrected. There is no economic boom. The inflation rate has continued to fall. If the Fed did overshoot on the ease side, recent economic data suggest that, in retrospect, the overshoot was a good idea and not one that should be corrected by a reversal in subsequent policy.
   
   
4. The Fed is passing the buck to fiscal policy. But fiscal policy is tightening too now, in relative terms. It doesn’t seem likely that the Fed is irresponsible enough to base its policy on hypothetical fiscal policies that aren’t actually happening.
   
   
5. The Fed has “abandoned the Mankiw rule” and is now setting its policy stance according to very different criteria than it has used over the past 23 years. But is there any evidence that Ben Bernanke has had some sort of conversion experience? And is there any reason why the Fed would be interpreting its mandate differently than it has in the past?
   
   
6. The Fed has dramatically altered the parameters of its “Taylor Rule.” But why?
   
   
7. The Fed is uncomfortable with quantitative easing and would like to minimize its use and reverse it as soon as possible, irrespective of Taylor Rule considerations. I think we have a winner. The long term effects of quantitative easing are uncertain and could be seen as potentially dangerous. (What will happen if, at some point in the future, the Fed has to choose between liquidating its unconventional assets at a loss, exacerbating an inflationary environment, or raising interest rates high enough to risk a fiscal crisis?) So there is arguably reason for the Fed to be uncomfortable with it. But the implications are disturbing, if you believe in a Philips curve or anything like it. Faced with an excessively high unemployment rate and an excessively low inflation rate, the Fed is choosing to risk exacerbating the situation (i.e., to take the intermediate-term risk of deflation) rather than to risk a very different type of difficult situation in the distant future. Maybe it’s the right decision, but it’s an awfully scary one.
   

Here’s one way to think about the situation. Fed policy typically affects output and employment with a lag of less than a year. Over the past year, the Fed has tightened dramatically. The super-duper-easy aggressive quantitative easing policy of 2009 (especially early 2009) has given the US economy enough monetary fuel to get it almost to an employment growth rate (exclusive of the Census) that could stabilize, but not significantly reduce, the unemployment rate. That policy is gone. In order to believe that the economy is going to strengthen from here, you have to believe either (1) that the lag associated with monetary policy is longer than usual or (2) that the underlying strength of the economy, holding monetary policy constant, has improved dramatically. Maybe one (or both) of those things is true. Or maybe not.






DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.

US Monetary Policy in the 2010’s: The Mankiw Rule Today

To make a short story even shorter, the Mankiw Rule suggests that the Zero Interest Rate Policy will continue for quite some time, barring dramatic changes in the inflation and/or unemployment rates.

“The Mankiw Rule” is what I call Greg Mankiw’s version of the Taylor Rule. “Taylor Rule” is now the general term for a rule that sets a monetary policy interest rate (usually the federal funds rate in the US case) as a linear function of an inflation rate and a measure of economic slack. Such rules provide a simple way of either describing or prescribing monetary policy. Unfortunately, there are now many different versions of the Taylor Rule, which all lead to different conclusions. Not only are there many different measures of both slack and inflation; there are also an infinite number of possible coefficients that could be used to relate them to the policy interest rate. In fact, if you ask John Taylor today, he will advocate a very different set of coefficients than the ones he proposed in his original 1993 paper (pdf).

Parsimony suggests that a good Taylor rule should have 3 characteristics: it should be as simple as possible; it should use robust, easily defined, and well-known measures of slack and inflation; and it should fit reasonably well to past monetary policy. Also, to have credibility, such a rule should have “stood the test of time” to some extent: it should fit reasonably well to some subsequent monetary policy experience after it was first proposed. The Mankiw Rule has all these characteristics. It uses the unemployment rate and the core CPI inflation rate as its measures, and it applies the same coefficient to both. This setup leaves it with only two free parameters, which Greg set in a 2001 paper (pdf) so as to fit the results to actual 1990’s monetary policy. As you can see from the chart below, the rule fits subsequent monetary policy rather well, although policy has tended to be slightly more easy (until 2008) than the rule would imply.



You will notice a substantial divergence, however, after 2008, between the Mankiw Rule and the actual federal funds rate. If the reason for this divergence isn’t immediately clear, you need to take a closer look at the vertical axis. Extrapolating from pre-Lehman experience, this chart suggests that the Fed is still doing the best it can to approximate the Mankiw Rule. When banks lend money to one another in the federal funds market, lenders stubbornly refuse to pay for the privilege of lending, and this perversity does limit the Fed’s options.

If we wanted to make a guess as to when the Fed will (or should) raise its target for the federal funds rate, a reasonable guess would be “when the Mankiw Rule rate rises above zero.” When will that happen? (Will it ever happen?) Nobody knows, of course, but the algebra is straightforward as to what will need to happen to inflation and unemployment. If the core inflation rate remains near 1%, the unemployment rate will have to fall to 7%. If the core inflation rate rises to 2%, the unemployment rate will still have to fall to 8%. Do you expect either of these things to happen soon? I don’t.


DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.