There is plenty of disappointment floating around concerning the big Chinese finance policy meeting which ended on Saturday. The general criticism is the lack of a signal of any concrete moves – but particularly, on interest rates for depositors.
Or as SocGen’s Wei Yao wrote:
Slightly surprising, interest rate liberalisation was nowhere to be seen in the plan, and there was hardly any trace of deposit insurance system except for a few words on enacting more financial regulation to protect individual investors.
Artificially low interest rates are one of the one of the policies supporting China’s relatively low household consumption; by being lower than the real inflation rate, it effectively punishes households for saving. It also pushes household savers to find alternative investments, much of which ends up, directly or indirectly, in property development, which is a whole other kettle of fish.
To recap: Chinese household consumption as a proportion of GDP is low — not unusual for a nation undergoing rapid development, but China’s rate of about 34 per cent of GDP is stonkingly low and apparently without historical precedent. The flipside is stonkingly high investment to GDP ratio – as this graph from Nomura shows:
Michael Pettis is one of those who argues this investment/consumption ratio is unsustainable, and the longer it is delayed, the messier the rebalancing is likely to be. Although it’s been CCP policy to rebalance the economy since 2006, official data has failed to show anything of the sort if happening — quite the reverse, in fact.
However, analysts at Barclays Capital say that things might not be so bad. They say those Chinese statistics might be just as unreliable as all the other Chinese statistics:
We try to piece together the true picture of Chinese consumption by examining statistical data:
First, according to one major study (Wang and Woo 2011)3, China’s household income was underreported by 66% in 2008, which could be translated into underestimation of GDP by 10%. The underreported income was concentrated mainly in high-income households.
Second, the same study suggested that household consumption was probably underreported by 20% in the same year. These meant a much higher household saving ratio but a marginally lower national saving ratio. Therefore, according to the study, both total consumer spending and consumption share of GDP were underestimated.
Third, the widening gap between consumption growth and retail sales growth likely implies that the growth rate of consumption has also been underestimated, at least in recent years. We use a weighted average of retail sales growth and service sales growth to proxy consumption growth.
Application of the above consumption growth suggests that: 1) the consumption share of GDP declined during much of the past decade; but that 2) the consumption share actually picked up forcefully after 2008.
First, they look at the gap between growth rates of retail sales, and of consumption as reported in the official GDP data:
Until 2000, retail sales grew slower than consumption expenditure. But after 2000, this relationship reversed (Figure 9). During 2001-2010, retail sales grew faster than consumption expenditure by an average of 4.1pp. The gap was 7.6pp in 2008 and 2009 and narrowed somewhat to 5.9pp in 2010.
And, they say, the differences are bigger than the difference in definition would suggest:
Differences between the two variables may not be surprising given their different definitions. But the two have significant overlaps. About 80% of the retail sales value is for consumption and accounts for 60% of total consumption (in GDP data). The remaining 40% of consumption consists mainly of service products (plus self-supplied farm products). Using the above shares, we derive an average growth rate of 1.4% for service consumption. In some years, such as in 2008 and 2009, the growth rate was significantly negative. Such results contradict common sense.
Then, expenditure on housing. This is really a question about definitions, but they make a case of sorts:
However, if the houses purchased are for ‘own use’, then they are not entirely for investment and the rental equivalent cost should be treated as consumption expenditure. In a typical market economy such as the US, spending on rental equivalent accounts for about 30% of total consumption expenditure. The problem is that China does not have a welldeveloped secondary housing market. In particular, the home ownership ratio (inclusive of self-built and state-allocated apartments) ratio is close to 90%, so the household survey data show very small spending on actual housing rental. Household survey data confirm actual spending on housing-related items at around 2.5% of disposable income. In the national accounts data, housing-related expenses are about 18% of total consumer spending. This is certainly more realistic than the survey data but, we think, probably still underestimates actual expenditure
The authors were also quite taken with surveys by Xiaolu Wang at the National Research Institute in 2005 and 2008:
By selecting a set of more than 1,500 households from the NBS sample, Wang went out to conduct his own surveys. In order to avoid the problem of some households’ unwillingness to report full income, Wang’s surveyors started by asking about detailed expenditure items. He then used Engel’s law, the well established correlation between household income and food expenditure, to estimate households’ actual income levels (Figure 11).
By doing this, he reached at least two very important conclusions. One, if Wang’s results are to be trusted, then China’s household income was significantly underestimated in official statistics, by 77.7% in 2005 and 90.4% in 2008. And, two, the underreporting was disproportionately concentrated in the high-income groups. In fact, the top 10% group’s income was underreported by 218.7% and accounted for 62.5% of the total income underestimation.
Wang’s findings, and a subsequent report co-authored with Wing Thye Woo have big implications, write BarCap: that income distribution is much more unequal than official statistics suggest, and that the Chinese economy itself is much bigger than thought.
BarCap then compile four scenarios of consumption’s proportionate growth for the period 1997 – 2010, using two measures of GDP (official data, and growth extrapolated from Wang’s surveys) and two measures of consumption growth (one using retail sales growth as a proxy, another using a weighted average of retail and service sales growth).
In all four scenarios, they say, consumption appeared to be falling until 2008, when it began to rise as a proportion of GDP. They conclude that investment was overreported and is more like 42 per cent in 2010, rather than the 48.6 per cent officially reported. And:
Our analyses also found underreporting of consumption. Total consumption expenditure was, on average, underreported by 5.7ppt, while the consumption share of GDP was underestimated by 3ppt. The extent of underestimation rose significantly over 2009-10. We imagine that most of the unreported consumption expenditure occurred at the high end of consumer market, in line with Xiaolu Wang’s finding that hidden household income was concentrated in high-income household groups.
Related links:
Not liking those odds of a Chinese hard landing – FT Alphaville
If (when) China slows down - FT Alphaville
Podcast with Michael Pettis - FT Alphaville