If there is one area where even the most battle hardened China analyst is left with nothing it is labor market data. Even the most Beijing loving, data believing kool-aid drinkers refuse to defend the accuracy of Chinese labor market data. Even if you believe the long term average is accurate, which in all fairness, I believe there is a reasonable probability that the long run average is somewhat accurate, it is beyond ludicrous to believe the complete lack of fluctuation.
The problem however, is that there is virtually nothing else that people can use for data to look at employment. Even in the data I’ve been able to find, about three to four years ago, industries reported aggregate numbers of people employed and that was stopped. There simply is no good data or even solid proxies for employment. Whatever you think of the Li Keqiang Index and electricity production as a good proxy of GDP, there was very little that could even come close to providing useful insight on Chinese labor markets.
I would like to propose a proxy for measuring Chinese employment. Before I tell you what it is, let me emphasize that that there a numerous shortcomings that I can think of and probably some that I have not, so please do not take this as a perfect measure that is authoritative. However, I do believe it can provide us some useful information and insight that should track some dynamics of the Chinese labor market.
For the industrial sector, there is a dataset that covers nearly 400,000 firms that will achieve a projected 11-13 trillion RMB in sales revenue for 2015 and manages almost 20 trillion in assets. This is a very comprehensive and large dataset of firms and key financial metrics. One of the variables is expenditures on administrative costs. I would propose that changes in administrative costs will be able to provide us some insight into changes in Chinese labor market dynamics.
Let me explain some of the positives and negatives of this measure. First, administrative costs will be strongly linked to labor inputs. While there are hard administrative costs like information technology, travel, or office space, labor costs dominate administrative costs and are the most variable when compared to the other primary input office space. Second, the internal components of administrate costs will be highly correlated with each other. For instance, if you hire more office workers, you will on average need more office space, computers, and travel budget. This is important, because we are less likely to see negative correlation between the internal components of administrative costs that will result in a funny top line number. By that I mean, if labor costs rise by 10% and the other costs fall by 10%, resulting in no change. That is less of a risk in this specific scenario.
Third, the administrative cost measure does not tell us the initial employment level. Consequently, while we can impute changes in labor market using some base, say 100 at whatever initial time period I choose, it will not tell us the number of people that have been employed or unemployed. Fourth, we can however impute the level of employment, again with some base say 100, based upon estimated wage changes. To take a simple example if, we estimate that wages have gone up by 10% and administrative costs have also gone up by 10%, we can essentially impute there has been no change in the volume of labor inputs. I have never seen academic or popular research on administrative costs, labor, and the other cost components so for now we have to use intuition.
Fifth, administrative costs provides insight but into a narrow slice of the industrial sector. I say this because at this point, I am loathe to project this onto the entire Chinese labor market. I am not saying it isn’t related to other sectors but until I have done more research, I want to recognize it for what it is: administrative costs in the industrial sector. In fact, I am not yet even entirely sure how this relates to workers directly related to product output. For instance, if a coal mine is facing a decline, it is possible administrative personnel will be laid off sooner than miners. At this point, I am simply reserving judgment until more work has been done. It is very unlikely however that the two are negatively correlated so there is most likely a positive correlation.
Sixth, I fully recognize that this is somewhat of a blunt measure. By that I mean, it would be best if we knew how many people were employed and how much they made and didn’t have other cost items in our measure. Even if all other cost measures are perfectly correlated with labor, their inclusion will blunt our understanding and require additional calculations to impute details we are really interested in. However, just as we shouldn’t project the importance of this measure too much, we need to understand that similar to electricity production, this is a blunt proxy.
In the past year, administrative costs in the industrial sector have increased by 5.2%. Now if we take the accepted wisdom that wage are still going up by 7-10% and use the midpoint of 8.5% this would imply declines in overall labor inputs. Even in a best case scenario this would appear to indicate wage growth with no growth in labor input implying some degree of increasing unemployment.
If we venture out a little into applying this the larger industrial sector, we can propose some tentative analysis. First, it would seem likely that there is growing pressure on industrial employment levels but not major decline in levels. Output in most industrial sectors is flat or seeing small declines. The prices of inputs are declining and especially in capital intensive industries where labor will play a smaller role. Given declines in capital and product input will offset some to a significant degree of wage increases, though there is clearly not only significant pressure but some declines in industrial employment. Consequently, it is probable the bigger impact on the labor market is overall declining/flat industrial output and falling prices. This combination is placing a lot more pressure on wages and employment levels, but so far unlikely to cause widespread unemployment.
Second, it is very unlikely that formal service sector labor growth is offsetting changes in industrial employment levels. Financial services, transportation, hotel, retail sales, and real estate (not real estate construction) are not experiencing hiring booms by any means. In fact most have flat or low growth revenue and output. Financial services the only service sector with strong revenue and output growth for numerous reasons, is not witnessing a hiring boom and definitely not enough to offset declines elsewhere.
I plan to flesh out some of these issues but for the moment, just wanted to put out some of these ideas. I think this does provide some insight into the Chinese labor market, definitely far more than anything else out there but please note the caveats.