Or, some people continue to defend the view that rapid inflation is just around the corner
An Econbrowser reader writes, in defense of Governor Perry’s assertion that the Fed is debasing the currency: “The CPI is not a valid indicator of ‘debasement.’” I think this comment provides a wonderful example of the Alice in Wonderland world in which some people reside — if the data do not cooperate, redefine the terms!
The reader continues:
“Even putting aside quite legitimate quibbles with CPI calculation, the CPI remains contained because we are in a Depression and demand is anemic. This does not imply that the Fed is not debasing the dollar. It clearly is, as illustrated by M2, MZMN the Fed balance sheet, and gold.”
I have my criticisms of the CPI , but this indictment is broader, applying to the entire concept of an index. So let’s dissect this all too common view.
Debasement Defined, and Some Data
Now, in the dictionaries I read, I see the definition for debasement sound something like this (here, from Merriam Webster):
To reduce the exchange value (of a monetary unit)
So I just simply don’t understand the statement that the rate of change in the CPI is not the way to measure the pace of debasement, in the absence of an alternative bundle of goods to use as a reference.
Actually, it is true that the exchange value of money has been decreasing against a bundle of goods purchased by households, namely the basket in the consumer price index. It’s also doing so against a bundle of consumer goods and services. (For a discussion of the distinction between the CPI and PCE, see this post.) Of course, it’s been doing that for most of the past century (with a break in the Great Depression).
Figure 1: Headline CPI (blue) and core CPI (red), year-on-year, seasonally adjusted. Growth rates calculated as log differences. NBER defined recession dates shaded gray. Source: BLS via St. Louis Fed FRED, NBER, and author’s calculations.
Figure 2: Personal consumption expenditure deflator, chained index (teal) and core PCE deflator (dark red), year-on-year, seasonally adjusted. Growth rates calculated as log differences. NBER defined recession dates shaded gray. Source: BEA via St. Louis Fed FRED, NBER, and author’s calculations.
Hence, a little context is helpful here — the rate of inflation was faster under Ronald Reagan than it is now. It was also faster under George W. Bush than it is now. Debasement, defined this way, has been happening for quite a long while, and at faster rates than currently.
Perhaps Incipient Hyperinflation?
I think the reference to money aggregates and balance sheets and oil and gold refers to inflation about to happen, but hasn’t shown up in the data. Well, individuals like the aforementioned reader have been predicting inflation — actually hyperinflation — for a couple of years, and it hasn’t shown up yet, so I decided to look at the evidence in favor of the hyperinflation-in-the-near-future thesis (I call this the “Apocalypse Soon” scenario).
First, what do survey based based indicators of expected inflation? They fail to exhibit hyperinflation hysteria (or accelerated debasement anxiety for that matter), for the next ten years. Second, what do market based indicators suggest? Using the spread between TIPS and Treasurys adjusted for potential bias, there is little evidence as well. I reprise these two graphs from this post on Governor Perry’s comments on monetary policy.
Figure 3: Ten year expected CPI inflation. Source: Philadelphia Fed’s August 2011 Survey of Professional Forecasters.
Figure 2, TIPS-Treasury spread estimates of ten year inflation, corrected for liquidity effects (Christensen, Lopez, and Rudebusch 2010 and Christensen 2008), from Christensen and Gillan (June 2011).
If anything, inflationary expectations extracted from market prices are declining!
Monetary Aggregates as Canaries in the Coal Mine
Well what about monetary aggregates? Are they signalling a burst of inflation missed by professional forecasters and market participants? Most in the profession long ago gave up paying much attention to monetary aggregates, but they seem to have a renewed popularity as the economy has hit the zero interest bound. We know money base has little correlation with future inflation, according to past data (see this post). But what about these other aggregates mentioned, including M1, M2, and MZM (not sure what MZMN cited by the reader is, but I’m guessing it’s MZM — “money zero maturity” –, the alternative definition of money excluding small time deposits but including money market mutual funds Teles and Zhou (2005), Box 1)?
Figure 5: M1 to nominal GDP (blue), M2 to nominal GDP (red), and MZM to nominal GDP (green), all quarterly averages of monthly money data, and seasonally adjusted. NBER defined recession dates shaded gray. Source: Federal Reserve Board and BEA via St. Louis Fed FRED, NBER, and author’s calculations.
M2 and MZM both seem to be rising in a noticeable fashion. Could there be any reason for this, aside from a allegedly treasonable action by the Fed Chairman? (Of course, the interpretation of recent accelerated debasement is problematic, given the trend in MZM/GDP; it’s been rising since at least 1997, or even back to 1980. So watch out Greenspan and Volcker — don’t go to Texas!) Well, these series are nominal money divided by nominal GDP. Those familiar with the Quantity equation (MV=PY) will recognize these series as the inverse of velocity. Velocity should be expected to be constant only in certain cases (the Quantity Theory for instance). In fact, one can see there’s a reason why MZM has been trending up, as suggested by Teles and Zhou (2005), VMZM should vary with the opportunity cost of holding money. Figure 4 presents MZM to GDP ratio against the difference between the 3 month t-bill rate and the MZM own rate. And indeed it does.
Figure 6: MZM to nominal GDP (blue, left scale) and 3 month T-bill rate minus MZM own rate (red), all quarterly averages of monthly money data, and for MZM, seasonally adjusted. NBER defined recession dates shaded gray. Source: Federal Reserve Board and BEA via St. Louis Fed FRED, NBER, and author’s calculations.
As interest rates have fallen, the opportunity cost of holding the balances in the MZM aggregate have fallen, and holdings of the constituent balances in MZM have risen…
Anyway, returning to inflation — the belief that MZM would provide a good indicator for future inflation is predicated on a stable relationship between the monetary aggregate and other macro variables, including GDP. Is that assumption reasonable? At this juncture, it’s useful to consult the literature. There was a flurry of interest in MZM back in the early to mid-1990s, when it appeared to be cointegrated with GDP (e.g., Carlson and Keen (1996), and Carlson et al. (1999), among others. However, interest in this variable seemed to evaporate shortly thereafter. I think this is partly because MZM stability was not as enduring as hoped for. I use the MZM/GDP data plotted above from 1975Q2-2011Q2 to check for cointegration using the Johansen maximum likelihood procedure (4 lags of first differences, allowing for constant in cointegrating vector and VAR, deterministic trends in the data). The data fail to reject the no cointegration hypothesis (p-values of 0.35 and 0.29 for trace and maximal eigenvalue statistics, and asymptotic critical values). All the data needed to replicate these regressions are available from FRED.
What does a visual inspection of the correlation between growth rates of MZM to GDP and CPI inflation indicate? I present a scatterplot of the contemporaneous variables, and another with MZM to GDP growth lagged a year.
Figure 7: Scatterplot of 4 quarter CPI inflation rate against 4 quarter change MZM to nominal GDP ratio, 1986Q1-2011Q2. Red line is nearest neighbor (locally weighted) regression, bandwidth = 0.3. Source: BLS, BEA, and author’s calculations.
Figure 8: Scatterplot of 4 quarter CPI inflation rate against 4 quarter change MZM to nominal GDP ratio, lagged one year, 1986Q1-2011Q2. Red line is nearest neighbor (locally weighted) regression, bandwidth = 0.3. Source: BLS, BEA, and author’s calculations.
I will merely state the obvious fact that the slope of the fitted lines is generally downward sloping.
Clearly, one doesn’t want to rely upon bivariate regressions to make conclusions (such as was done in this instance.) Recently, Stock and Watson (2010) have evaluated the usefulness of macro variables, including various money aggregates, in helping predict inflation. They write: “This is not to say that monetary expansions and inflation are unrelated, rather, the evidence here is that the predictive relationship between money and inflation is weak and unstable at short to medium horizons.” Interestingly, what they argue is of use is an “unemployment recession gap [which] is the difference between the current unemployment rate and the minimum unemployment rate over the current and previous eleven quarters.”
So, it could be next year, the relationship that has existed between MZM/GDP growth rates and inflation over the past twenty five years will be completely upended, and we see rapidly accelerating CPI inflation/debasement of the currency. Personally, I’m looking elsewhere for modeling inflation (e.g., expectations augmented Phillips curve allowing for supply shocks   or variants).
What about Oil? And Gold?
One could salvage the inflation argument (once again) by redefining the basket of goods to be composed solely of oil. Or solely of gold. Even then, one would have to argue that the movement in oil prices is due to money debasement. I think a decomposition is useful here. Let poil be the log price of oil, and p be the log general price level. Then:
Δ poil ≡ (Δ poil – Δ p) + Δ p
The term in the parentheses is the relative price of oil. The Fed could conceivably affect that relative price as well as the general price level. However, other things affect the relative price, including demand for oil (think China) and supply of oil (think Libya) . Now, it could be that the Fed controls both (and we just plain don’t know it), but I think a more nuanced (and perhaps less paranoic) view is called for.
As for gold, I could do the same thing, and just say my bundle of goods is 100% gold. But I don’t think that’s why the reader focused on gold; rather I think he believes gold prices are predicting something. And I’m sure gold predicts something. Whether it predicts the future rate of inflation (defined as the change in the exchange value of money for a bundle of goods and services), I’m not so certain of , .
This post originally appeared at Econbrowser and is reproduced here with permission.
One Response to “Fidei Defensor”
Response from the reader in question here: