The title of this post borrows from Ed Leamer´s text with the same title. After reading the following posts:
Joseph Politano writes “The Fault in R-Stars: The Problems in Policymaking Based on "Natural" Rates” and says that
Policymakers have put too much faith in unobservable “Natural” rate estimates, and the results have been catastrophic.
Taylor Shiroff writes that forecasting inflation is hard:
One place not to look for a good inflation forecast: consumer surveys. But if you do, listen to the younger respondents! They’ve been much more accurate over time.
Gabriel Mathy, Skanda Amarnath and Alex Williams write about bad application of models in “Expecting Inflation: The Case of the 1950s”. The Executive Summary reads:
Whenever inflation becomes a part of political or economic discourse, policymakers and commentators instinctively reach for narratives and models drawn from the experience of the 1970s inflation. However, these models offer little explanation for even adjacent experiences of inflation. Case in point: the combination of macroeconomic data-points and policy responses observed in the 1950s would suggest — under models of the 1970s — that runaway inflation was imminent. Instead, inflationary pressures quickly resolved themselves as the economic situation changed and new capacity was built out.
The theoretical inadequacy that this fact demonstrates should lead us to view explanations of present inflationary pressures that rely heavily on models developed to explain the 1970s — like the Phillips Curve, inflation expectations, or Fed credibility — with a heavy dose of skepticism. Instead, then and now, a sectorally disaggregated view of capacity bottlenecks and required adjustments provides a far superior picture of the source of, and remedies for, inflationary pressures.
I thought it would be interesting to find if a pattern emerged that was backed by a consistent story.
My strategy is to follow the considerations made by Gabriel Mathy et al and compare different periods to find a consistent story that explains the events observed.
Like them, I start with the 1950s. The panel below, shows the “patterns” I will elaborate upon.
The NGDP growth chart provides the “fundamental” information, in that it defines the stance of monetary policy. From the equation of exchange in growth form: M+V=P+y, where M=money supply growth, V=change in velocity, P=inflation and y=growth of real output (RGDP) and P+y defines the growth of nominal GDP (NGDP). So, if NGDP growth is rising, monetary policy is deemed expansionary, and vice versa.
You see immediately that oscillations in NGDP growth are closely negatively correlated with oscillations in unemployment. Also, when NGDP growth oscillations decrease in early 1953, variations in inflation are also reduced.
Before looking at the 1970s (“Great Inflation”) it is enlightening to look at the 1990s/2000s period. This period is characterized by a stable growth of NGDP, much like the 1953-1960 years, so it will be a good test of the story behind the pattern.
There are, however, different degrees of stability, some more, and others, less, volatile.
This is the case of a comparison between the 1953-1960 and 1992-2005 periods as illustrated in the chart below.
While for 1953/60, mean NGDP growth was 5%, for 1992/05 it was a bit higher at 5.4%. Volatility (measured by the standard deviation) of NGDP growth, however, was almost 3 times higher in 1953/60.
The panel for 1992/05 follows.
Note that, with much more stable NGDP growth, unemployment falls continuously, until a significant NGDP growth oscillation takes place. We will see that pattern repeated in the 2010-19 period of even less volatile NGDP growth. Inflation also remains low & stable.
Moving to the 1970s, the “Great Inflation” decade, the stories behind the patterns remain consistent. I noted earlier that the NGDP growth chart provides the “fundamental” information as it defines the stance of monetary policy. As the panel for the 1970s show, by showing an up trending NGDP growth, monetary policy was consistently expansionary.
The close negative correlation between swings in NGDP growth and unemployment remain, enhanced by shocks like price & wage controls and the price of oil. Inflation remains mostly on an upward trend, also modified and enhanced by shocks.
The difference in monetary policy in the 1950s and 1970s is sufficient to explain the difference in the behavior of unemployment & inflation in the two periods (obviously considering the different shocks that intervened).
The panel below depicts the 2010-2019 period. The similarities with the 90s/00s is strong, but we´ll see there is one major difference.
NGDP growth is a little more stable (with standard deviation of 0.9 against 1.2 for the 90s/00s), but the average growth is 4% (versus 5.4% for the earlier period). As expected, highly stable NGDP growth is consistent with continuously falling unemployment (and stable & low inflation).
Note that the unemployment rate is sensitive to NGDP growth. Even very low oscillations in NGDP growth is capable of “braking” the fall in unemployment, as noted in the shaded area.
The major difference in the behavior of NGDP in the two periods is not so much the lower growth rate of the 2010s, but the fact that the LEVEL path of NGDP was significantly lower than in the 90s/00s as indicated in the chart below.
What has implications for the behavior of the unemployment rate is the Stability of NGDP growth. The Level path has implications for the Fundamentals of the labor market, things such as the Employment Ratio & Labor Force Participation.
If the Fed had tried to put NGDP back on the 90s level path in mid 08, instead of allowing it to plunge, bringing about the “Great Recession”, NGDP would have recovered back to the previous trend level path. Unemployment would have increased much less, and would have started falling sooner and faster. The fundamentals of the labor market would not have deteriorated significantly and the ensuing low unemployment would have been associated with a much more robust economy. (Trump might have even been avoided!).
But, no matter the initial state of the economy (higher or lower level path) it was on when the Covod19 shock hit, the shape of things would be much the same. Given the small amount of data (19 data points) for the variables concerned, examine the “patterns” from the level of NGDP instead of its growth rate and I look at RGDP instead of the unemployment rate.
The panel for the period is illustrated below.
Where the top right hand side chart provides a stylized growth version of the other charts.
We have to understand the Covid19 shock as both a demand and supply shock. The demand “leg” of the shock was not the more usual drop in money supply but was caused by a massive increase in money demand (velocity tanked). The shift up and to the left of aggregate supply was much less intense. The combination of a strong negative demand shock and a weaker negative supply shock results in a drop of inflation (point b).
The Fed reacted quickly, rapidly increasing money supply to offset the fall in velocity. That´s why the recession was deep but short, with RGDP soon climbing (and unemployment falling) back to the trend path it was on before the shock hit. (Just imagine the depression that would have taken place if those that worried about the 30% YoY growth in money supply had been heard!).
At point c, aggregate demand (AD or NGDP) is back on the trend path. However, supply bottlenecks linger, so RGDP remains below the trend path. With supply constrained and demand growing, the result is higher inflation.
The stability of the trimmed PCE inflation indicates that inflation is not a “generalized” phenomenon. While some sectors experience supply constraints, others don´t. Also, overall inflation seems to be “tapering”. Keeping NGDP growing stably along the trend path, is the best the Fed can do, since it should not react to supply shocks.
When bottlenecks are resolved, the Fed can strive to place the economy on a higher level path. As we have confirmed from different experiences, a stable rate of growth of NGDP will be accompanied by stable & low inflation and a falling/low rate of unemployment. If the Fed is successful in placing the economy on a higher trend path, labor market conditions will also improve.
Some notes
Throughout, I haven´t said anything about Fed Funds rate, “tapering”, dot plots, slope of Phillips Curves, natural unemployment, output or interest rates, or fiscal policy. The “wide spectrum antibiotic” for keeping the economy “healthy” is a stable rate of NGDP growth, obtained by the Fed appropriately offsetting changes in velocity (“appropriately” means managing to attain the desired stable growth of NGDP).
I have also argued that the level of the trend path is important, so, over time, the Fed should try to place the economy at the highest feasible path. When (if) that happens, it will also attain “maximum employment”.
PS Best cartoon of FOMC Members on meetings to decide on policy
"Macroeconomic Patterns & Stories"
NGDP targeting makes sense. But then you go an say the Fed may target a new path. So now you're introducing a new variable - which path to shoot for. That presumably would be based on future productive capacity - and we know productivity forecasts are basically useless.
For NGDP targeting to be credible, the Fed would need to pick a path and stick to it. With only perhaps very rare "resets" for changes in demographics, which are observable.
By making the path itself an additional variable you introduce all the bad analysis and political maneuvering that was responsible for the historical periods where monetary policy went wrong in the first place.
A 4% NGDP growth path strikes me as very reasonable and achievable.
Hey, found this through your comment on Claudia's sub. It's good! I have a question:
Is this one of the arguments for a nominal gdp targeting framework that I hear beckworth go on about constantly in his macro musings podcast?
I'm a total newbie, but am wondering whether that would be a potentially limiting framework if our potential NGDP growth were higher than the target, but we undershot because our target was too low.
If I'm reading this right, might an answer be: well, yes that's a possible worry, but every time we increase nominal GDP really quickly, we get bad results, so best to limit its growth in the short run, and then if we think we're consistently overshooting, maybe revise the framework to have a higher target?
The kind of cases I have in mind are ones where some big innovation comes along and massively increases potential output right away. A NGDP target may not allow an economy to quickly realize that potential, but perhaps that's for the best if quick expansions lead to bad results.
Then again, maybe quick expansions only led to bad results _under the conditions of the old framework_ in which case, I'm not sure what to make of those historical data.