I don’t normally write two blog posts in one day but given the CPI data released today, I wanted to detail in brief why these numbers are so absurd because we can clearly tell these are absurd numbers with no basis in reality. When I first saw the data on CPI increase to 2.3% I was stunned and when I saw that food was up 7.3% more than twice the change of any month in the past year my mouth hit the floor. I never rule anything out in China but given the magnitude of the change only a few days after Beijing released a 3% inflation target, I thought it would be wise to just double check some numbers.
Before I explain why, let me give you a little background on China and CPI. Numerous studies have found large problems with Chinese price data. Emi Nakamura and Jon Steinsson (PDF) of Columbia found that official price data was much too smooth when compared to how household surveys indicated that households were spending money. From my own work (PDF), the housing component of Chinese CPI shows that urban housing prices only increased 6% (let me emphasize that it total not annually and not asset price but rather how much more housing costs to obtain) from 2000-2011 and even if we added on 2012-2014 would only slightly raise that number. There are many many problems with Chinese inflation and CPI data.
These problems have not gone away. As Nakamura and Steinsson point out, as a country grows richer, they should spend less on food as a percentage of their income. In simple terms, if you make $1,000 in a year, you might spend 40% of your income on food. However, if you make $100,000 you would probably only spend say 5%. This is important because since 2011, China has raised the weight of food in the CPI basket every year from 31.4% to 33.6%. You would be hard pressed to find another country in the world that experiences rapid growth and every year spends a higher percentage of its income on food.
All this is good background but also ties in directly to why we can tell today’s CPI data is nothing short of a fabrication. With food comprising 33.6% of the CPI basket and food CPI running at 7.3%, this means that food inflation was responsible for 2.45% of the 2.3% CPI rate. In other words, if you strip out food, inflation was actually -0.15%.
However, that’s not the worst of it. It gets sooooo much better. If you look beneath the hood of the CPI basket you will see that actually most food product prices were actually quite small. Grains were up 0.6%, oils and fats 0.7%, aquatic products 3.5%, milk 0.1%, and beef 0.3%. Mutton and eggs were down 7.8% and 3.6% respectively.
So what was up enough to account for the food CPI number? Fresh vegetables and pork, registering increases of 30.6% and 25.4% respectively, were the only products with increases above the average. At this point, it would be tempting to say that over the Chinese New Year holiday people celebrated with friends and bought the best food so makes sense.
There are only two problems with this theory. First, do you want to tell me no other food prices went up? The only food people ate more to justify such a shift in prices was pork and fresh vegetables? Skeptical that they would be such extreme outliers but let us set that aside for a moment. Second, the related data on pork and vegetables tell a radically different story.
Let stop a second and talk about ways to verify Chinese data. A key tenant of mine is to look for numbers that should come close to the number in question. This does not mean we are looking for a perfect match but numbers that are I the ball park of each other. Let me give you a simple example. If electricity production declines 10% but industrial consumption goes up 10%, you have to ask what is going on. If electricity production goes up 1% and consumption goes up 2.4%, that’s good enough for me and move on. For almost any data point, there are many related data points that we can match up that will give us an idea what up and downstream things are doing.
What is interesting is that in reality, fresh vegetable and pork prices appeared to have gone up by roughly the amount indicated. For instance, the Wholesale Price Index of Qianhai Agricultural Products of Vegetables from February 2015 to February 2016 was up 34%, only slightly more than the official CPI amount for fresh vegetables. There are a wide variety of prices to choose from on that cover pork and pig prices from conception through to table and they are also roughly in line the 25% increase.
So what’s causing the discrepancy? Bad math. If we take the official CPI weights for each category and plug them into the price increase of each product for the food category (remember food was responsible for more than 100% of the official CPI or 2.4% of the total 2.3% CPI) equals only 1.41%. (By this I mean for instance, meat was up 15% and comprises 4.33% of the total CPI basket and repeat for other food categories.) What makes this even worse is that if we remove pork and fresh vegetables, food prices for which we have official basket weights and official inflation numbers shrank slightly at 0.06%. Given that the CPI basket covers meat, grains, produce, dairy, flavoring and others and the food that we do not have official weights for was flat like the non-pork and fresh vegetable categories, it is very difficult to see where this supposed food inflation came from. However, as the CPI contribution of pork and fresh vegetables only amount to 1.4%, this means we still need to find an additional 1% from the food category as the implication is, it does not come from the non-food basket.
What is so amazing about this is that inflation statistics in China do not add. China says non-food inflation in February was 1% and food inflation was 7.3%. Given a non-food weight of 66.39% and a food weight of 33.61%, this would sum to inflation of 3.12% or more than 0.8% more than the official rate. In other words, Beijing tops in Communist efficiency has already met the 3% inflation target it set just last Saturday and they exceeded their target. The only way you can have food inflation of 7.3% with a 34% food weight is to have deflation throughout the rest of the CPI.
However, even then, given the weights of the CPI basket and the price increases in those specific products, the numbers do not add to what Bejing says. If we take Beijing numbers and assume that non-food need to deflate slightly and pork and fresh vegetables were the only food items that really increased in price, we still can only arrive at inflation around 1.4%.
There simply is no way for the numbers to reconcile. Let me emphasize in closing that everything here is official data and statistics. The only math used was simple math done in an Excel spreadsheet. All you have to do is try and do something as simple as add the numbers up to see that they do not add.
Update: I have been told by Alex Frangos of the Wall Street Journal Hong Kong that the NBS reduced the CPI weight of food by 3.4% by my calculation to 30.2% of the CPI basket. I have not seen that and given that this would lower the combination of non-food and food inflation rates to 2.9%. Given that I was working with 2016 component weights for products like meat or pork, this does not alter the other parts of the analysis.