Data Diving on Inflation
So now that we have had some new data releases and I am back from vacation, we get back to what I still consider the meat of this blog, focusing on the data of the Chinese economy. I have developed a general rule of thumb that if you a read about a Chinese statistic in the headline of a news article, it is most likely heavily manipulated. To rephrase this slightly, if it is top line data, it is heavily manipulated. One of the problems for people who have not spent enormous time looking at the data is that, to give credit to Chinese artists, they are very skilled at manipulating the data in a variety of ways that can make it difficult to detect. Many times, you can only detect the manipulation looking at much more granular data, which in all fairness, simply aren’t available to most people. However, in some cases, even just looking at granular data on the NBS website and comparing it to top line data that should be a summation of lower level data, one can see the discrepancies.
One of the biggest ways that Chinese officials manipulate real GDP growth is by manipulating the pricing indexes. There has been recent skepticism about the unchanging nature of the PPI flattening at 5.9% over the past few months given the continuing price declines in a range of products. However, for the moment, I want to focus on the CPI basket weighting.
When building inflation rates, one of the key questions is how to weight each and every product of the basket. Ideally, you have lots of micro-level data that gives you a very clear idea of how to weight individual products and classes of products. For instance, food prices should still comprise a major portion of the CPI and budgetary basket as despite Chinese economic development, it still comprises a significant portion of most people’s spending. The CPI weight should closely resemble people’s spending habits.
There is one other thing that impacts how we want to approach the Chinese CPI basket, there is something called Engel Curve that should provide us a good basis of comparison about how people spend their money. The idea is really pretty simple in that the more money a household or country makes, the less they will spend on food as a percentage of income. It is really quite intuitive. Assume you make $1,000 per year, you will probably spend a pretty high percentage of your income simply keeping yourself nourished. Now assume you make $100,000 a year, you will probably spend a higher amount of money in absolute terms, say $500 increased to $2,000, but you will spend a significantly lower percentage on food in absolute terms, going from say 50% to 2%. As households and countries become wealthier, they will spend less as a percentage of their income on food.
Emi Nakamura and Jon Steinsson of Columbia, currently on leave at MIT, have actually written a paper studying the Engel Curve and Chinese households (PDF). They write about recent Chinese inflation that “official inflation rose in the 2000’s, but our estimates indicate that true inflation was still higher and consumption growth was overstated over this period.” If true inflation was higher this would mean that real GDP growth was in reality overstated. In addition sound empirical and theory based research that inflation is significantly underestimated, we can again look inside the Chinese inflation data calculations.
China began releasing its CPI basket based upon food and non-food items, though in 2011 it released the weightings for the non-food sub items. In 2011, the CPI food weighting was 31.39%. This 2011 weighting had fallen from the 2006 number of 33.60% and was expected to keep falling and relatively rapidly. History, intuition, and wealth of global economic data said that this number should pretty quickly fall beneath 30% of the budget. Right? Wrong.
In 2012 the food weighting in the CPI basket jumped almost to 32.21% and has risen about 0.5% every year since. For 2015, the food weighting of the CPI stands higherthan the 2006 number at 33.61%. This stands is such complete contradiction to everything we know about economics that is belies any credibility.
There are a couple of empirical points to look at here. First, from 2011 to 2014, average urban wages officially grew 34.8%. From 2006 to 2014, urban wages officially grew 170.2%. Oddly though, during this period, the amount of money spent on food, according to official statistics, actually increased as a percentage of income. In reality, this seems distinctly unlikely as it would spell morbid obesity with Chinese characteristics for most of the country. It isn’t rising that fast.
Second, if we actually believe these numbers, from 2011 to 2014, the Chinese consumer will have seen wages go up 34.8% but food spending go up by 44.4%. Put into absolute terms, they would have spent about an additional 2,880 RMB or 5.1% of their income on food than if the CPI weighting had dropped by a similar amount instead of going up. That is not an insignificant amount of income. I know of no other country that would plausibly claim to triple wages and have people spend the same percentage of income on food as before.
Third, I can’t honestly explain why the food weighting has gone up though I have some guesses. Given that food inflation in China runs higher than non-food inflation, which given PPI deflation isn’t a surprise, it could be they are trying to obfuscate realized deflation. Another possibility is that food inflation is actually higher than admitted but by allocating a higher percentage of the CPI weight they can obfuscate the true impact by allocating the same absolute expenditure amount but change the internal percentages. I will be honest in saying, I’m not really sure why.
Fourth, the inflation basket is riddled with other problems but this is one good example. For instance, the residence component is only 17.8% of the CPI basket. Not only is that number incredibly low by really any standard, I think you would be hard pressed to find anyone in China who would say that food has risen as a percentage of their income while housing is only 17.8% of their income. Until 2011, Chinese residents supposedly only spent 13% of their income on housing, an absurdly low number for any country but definitely one in the midst of one of the most rapid price increases in housing in modern history. The point however, is that how a CPI basket is weighted internally can have a major impact on the final number depending on what you are trying to accomplish.
Whenever you read of a headline number in China, do NOT, I repeat do NOT just accept it as fact but try and go beneath and figure out what is going on. There is tremendous and important detail lurking just beneath the surface.