Comment 30: Inflation, CPI, CPI-core and all that (May 29, 2006)

The fear that the accelerating US inflation rate will force The Federal Reserve Board (abbreviated as FED) to continue to raise interest rate has been attributed as the cause of stock and financial market turmoil around the world, especially after the release of April Consumer Price Index (abbreviated as CPI) at May 17, 2006. Month to month variations of CPI and CPI-core, that excludes energy and food from CPI, are quite volatile. It is doubtful that any meaningful information can be derived from such a volatile one-month data. Month-to-month percentage changes (annualized) of both CPI and CPI-core are plotted in the top portion of the graph at the right; blue dots are for CPI and red dots are for CPI-core. Only thing we can say from the data is that the gyrations of CPI and CPI-core data have increased substantially since 1999. Especially the increased volatility in CPI data is quite noticeable. However, the right most data points, the April 2006 data of both CPI and CPI-core do not indicate any sign of a sudden worsening of inflation aspect. To make sense out of those chaotic data, some kind of smoothing out process is necessary. A frequently used method is to compare the data of a month with the data of the same month of the previous year and plot their percentage changes. This year-to-year comparison is plotted in the lower half of the graph; blue dots are for CPI and red dots are for CPI-core.

Many Wall Street analysts concentrate their attention on CPI-core only. Their logic is that energy and food prices are often volatile, so it is better to exclude them from consideration. This kind of logic only makes sense if energy and food prices average out to a constant value for a long period but are gyrating around this constant average value violently month-to-month. However, in a period that energy and food prices are persistently rising, the ignorance of CPI and the concentration of attention on CPI-core will severely underestimate the real inflation pressure. In order to understand the actual inflation picture, we must consider both CPI and CPI-core simultaneously. The year-to-year comparison curve of CPI has a bottom around 1998, a peak at 2000 to the early part of 2001 and another bottom at 2002. Since the middle of 2002, year-to-year comparison curve of CPI is moving up in a zigzag way. The data of April 2006 falls into this up trend. The year-to-year comparison curve of CPI-core tends to bottom and peak roughly 6 months to one year behind the curve of CPI, indicating the time lag for rising and falling energy and food prices to penetrate into non-energy and non-food items. Considering the sharp rise of year-to-year comparison curve of CPI well into the end of 2005, we should anticipate that the year-to-year comparison curve of CPI-core would continue to rise through most part of 2006. The temporary lull of CPI-core curve from early 2005 to early 2006 is only reflecting the stagnation of CPI curve from June 2004 to July 2005. Thus the rise of year-to-year comparison curve of CPI-core in April 2006 should be fully anticipated and it is rather ridiculous for many observers to attribute the worldwide turmoil of financial and stock markets to this highly predictable rise of year-to-year curve of CPI-core. We should emphasize that the most recent CPI data does not tell us where CPI is going from here. Even if crude oil price drops sharply from here (NOTE: This is a pure hypothetical assumption thrown in to make a point. It does not mean that we are predicting the drop of oil price soon), year-to-year curve of CPI will start to drop and inflation fear will be eased, but the year-to-year curve of CPI-core will continue to rise through 2006 and then fall sometime in 2007.

Year-to-year comparison curve of CPI is only telling us the inflation picture of roughly 6 months ago. As discussed in the previous paragraph, year-to-year comparison curve of CPI-core is telling us the inflation situation of one year or more ago. Apparently we need some other methods to study the up-to-date inflation situation. For that purpose we draw in the second graph at the right the CPI and CPI-core indices themselves, but in conventional logarithmic scale. In the lower right-hand corner, data of recent years are redrawn with enlarged logarithmic scales to make the features of the curves easier to observe. In logarithmic scales, if data points line up in a straight line, say a rising line toward the right, the data points are rising with a constant rate. Both CPI and CPI-core indices are calibrated by taking the average value of 1982 to 1984 as 100. That means that if we plot the data back to 1982 to 1984, CPI and CPI-core would have overlapped. The fact that the curve of CPI-core lies above CPI curve in the graph means that after the 1982 to 1984 period there was a prolonged period of weak crude oil prices so that CPI-core grew faster than CPI. In the whole history of CPI-core from 1957, it mostly lies above CPI curve except in the period from 1974 to 1982, the years of energy crises; during that period CPI curve lies above CPI-core curve most of the time. In the graph a trend line A is drawn for the CPI curve to guide the eyes. Trend line A corresponds to an annual growth rate of +2.5%. We can see that until year 2004 CPI was growing about 2.5% a year in average. From 2004, a steeper trend line labeled as line B (in the lower right-hand corner) can be drawn; line B represents a growth rate of 3.2% a year. From June 2005, another trend line C can be drawn, representing the annual growth rate of 4.0%. Thus we can see the steady acceleration of CPI in recent years due to rapidly rising crude oil prices. However, the April 2006 CPI data still treads along the trend line C and does not show any sign of worsening inflation aspect. CPI-core curve is rising along the trend line D (representing annual growth rate of 2.2%) since the beginning of 2004, but there is a strong chance that a newer and steeper trend line may ensure from August 2005 as anticipated in the discussion of the previous paragraph.

Looking at the whole analysis discussed here, we do not see any sudden worsening of inflation aspect in US in the April CPI data. Thus contrary to the opinion of many analysts the recent turmoil of the global markets cannot and should not be attributed to the sudden worry of the worsening inflation but lies in somewhere else. However, the discussion of the market mechanism is outside the scope of this website as has been pointed out in Comment 1 so we will not discuss the reasons of global market turmoil further.

CPI and its related indices as discussed here are not the only price indicators measuring the price levels at the stage of final sales to consumers. Another popular measure among analysts and market participants is called the price deflator of personal consumption expenditure (abbreviated as PCE). Here, we simply call it PCE price indicator. This price indicator has been the favorite measure of former FED Chairman Greenspan and closely resembles the price deflator used in the computation of real GDP. As discussed before, CPI is compiled with the average price level from 1982 to 1984 taken as 100, but PCE price indicator takes the average price level of 2000 as 100. It is not difficult to convert PCE price indicator to the standard of CPI, that is, take the average price level from 1982 to 1984 as 100, so that we can compare those two indicators. The overall CPI at April 2006 is 201.0, whereas the overall PCE price indicator when converted to CPI standard is 182.9. This means that for the past 23 years PCE price indicator is in average below CPI by about 0.79 points a year. Let us look into the major components of those two indicators, i.e., service, non-durable goods and durable goods. PCE service sector is below CPI service sector by 0.64 points a year in average. In the non-durable goods sector, PCE indicator is below CPI by 0.43 points a year in average. In the durable goods sector PCE indicator is below CPI by 0.85 points a year in average. The reason of those differences between two indicators is due to how aggressively apply the highly subjective adjustment of improvement in products and services. Let us take personal computers as an example. Suppose an entry-level desk top PC cost $500 and was running at a clock speed of 8 mega hertz a second 23 years ago, and currently an entry-level PC still costs $500 and is running at a clock speed of 2.0 giga hertz a second, 250 times the clock speed of 23 years ago. The most aggressive advocate of product improvement theory will say that the price of an entry level PC is now $500/250 = $2. The strongest opponent to the theory of product improvement will argue that for a consumer to buy an entry level PC, he needed to pay $500 23 years ago, and he still needs to pay $500 today so the price of PC should be considered stable for the past 23 years. Between those two extremes, inflation compilers can place the price of today's entry-level PC anywhere between $2 to $500, all depending on whether one wants to see a subdued inflation or a heightened inflation.

There are quite a few analysts and many more in the general public who think even CPI is vastly underestimating the actual inflation. Part of the arguments of this group is to claim that in CPI the accounting of product improvement is still too aggressive and thus is artificially suppressing the inflation indicator. Another suspicion about CPI and other indicators is due to the sampling technique used. Let us take the price of bread as an example. Suppose there are two varieties of bread, the higher quality one and the standard one. We assume that the higher quality bread costs $2 a loaf and is bought by 90% of the population. The standard quality bread cost $1 a loaf and is only favored by 10% of the population. In the sampling of the bread price, the higher quality one is naturally chosen as the representative of breads, and the price of a loaf of bread is recorded as $2. Now suppose the price of flour skyrockets, and the price of higher quality bread becomes $4 a loaf and the price of standard quality bread becomes $2 a loaf. At the same time consumers abandon the higher quality and more expansive bread en mass and switch to the standard quality ones; 90% of population now buys the standard quality breads. The compiler of the inflation gauge will take the standard quality bread as the favorite of consumers and will use the price of the standard quality ones as the sample, that is, still $2 a loaf, though common sense tells us that the price of breads has doubled. If the inflation compilers had applied the theory of product improvement (disimprovement in this case) faithfully, the price of a loaf of the standard quality bread will be adjusted back up to $4 and thus matches the common sense. However, there are suspicions that inflation compilers do not adjust the price of bread up in this example, but diligently and aggressively adjust the price of entry-level PC down to artificially show a subdued inflation picture.

The discrepancies and shortcomings of inflation indicators discussed in the previous two paragraphs are systematic in nature. That means that the distortion in the indicator will show up month after month in the same direction. However, this kind of distortions of systematic in nature does not affect the analysis of this comment. Suppose we use PCE price indicators in our discussions and graphs instead of CPI indicators. The inflation trend lines A, B and C will still exist, showing an accelerating inflation trend for the overall price index; only the growth rates represented by lines A, B and C will be lowered by certain amounts. Of course, when inflation indicator is changed, it will have significant effect in the statistics like real GDP, real personal income and consumption, and so on. If CPI is adjusted up by 1% a year in average as demanded by skeptics, and if this new CPI is used to compute the growth rate of real GDP instead of currently used underestimated GDP deflator, the stellar performance of 3.5 to 4.0% growth rate per year of real GDP in the past few decades will be reduced to about 2.0% a year; a picture of a robust US economy under the aura of globalization will be turned into a picture of stagnation and frustration.