The Barclays Capital note we discussed earlier this week, postulating that China had already begun to see a critical increase in its household consumption-to-GDP ratio, has generated a few questions over email.
So, we’re going to go into it in a little more detail. But stay tuned — this is interesting. Really.
In short, China’s official data are problematic. The official household survey on consumption has problems of sample and disclosure; and this forms part of the basis of the official consumption/GDP ratio figures. Meanwhile retail sales data — which do not differentiate between private and public consumption, and exclude services — appear to account for a rising share of GDP.
As BarCap analysts Yiping Huang, Jian Chang and Lingxiu Yang write:
“The proportion of consumption-related retail sales to GDP actually picked up forcefully, from 26.7% in 2008 to 32% in 2010 (Figure 14). This probably means that consumption has actually strengthened during the past couple of years, both in absolute size and relative GDP. This is consistent with our general observation. But the official data point to a continuous decline in the consumption share of GDP. If the retail sales figures are reliable, the official data could be correct only if non-retail consumption expenditure collapsed in 2009-2010. But this is highly implausible.”
We won’t dwell on that IF the retail sales figures are reliable. The authors go on to point out, quite reasonably, that services consumption is very likely rising in absolute and relative terms, as more Chinese become middle class or wealthy — because that’s what people tend to do as they earn more. The BarCap strategists also use the non-official household survey data contained in a report by Xiaolu Wang and Wing Thye Woo as a basis for a new estimate of GDP.
Anyway. This is what they came up with (our emphasis):
So far, we can draw two qualitative conclusions from our analysis of Chinese consumption data. One, the official consumption data are likely underreported. And, two, growth rates of Chinese consumption are probably also underestimated. In order to gauge the true picture of consumption in China, we try to calculate four sets of consumption shares of GDP for the period 1997-2010 by adopting different assumptions under two scenarios and two different assumptions about GDP data and consumption growth (Figure 16)
We adopt two proxies for consumption growth. The first is the growth rate of consumptionrelated retail sales (ie, retail sales excluding producer goods such as petroleum, chemical, etc). The second is a weighted average growth rate of consumption-related retail sales and growth rate of service sales in China. Between the two, we place a greater emphasis on the results derived from the second proxy growth rate.
When applying Wang’s data, we first extrapolate the GDP underestimation by an additional percentage point in every three years, based on findings of underestimation by 9% in 2005 and 10% in 2008. Wang suggested that household consumption was underestimated by 20% in 2008. In that year, the official statistics reported total consumption share of GDP at 48%, of which household consumption was 35% of GDP and government consumption was 13% of GDP. These imply that the adjusted household consumption was 38.2% of GDP and total consumption was 50% of GDP. We take this as the starting point and then derive consumption data for the other years by applying various growth rates. In order to check for robustness, we also include consumption underestimation by 10% and 30%
Here are the results. Bear in mind that BarCap themselves say the scenarios using retail sales growth as a proxy are less useful, so concentrate on the right-hand charts which use a weighted average of retail sales and services growth:
You get the idea. They all point to an upturn in household consumption’s share of GDP from about 2008, in contrast to the official data which have shown the ratio declining.
None of this sounds especially implausible, nor does the BarCap conclusion that the official household consumption/GDP ratio of 33.8 per cent in 2010 may be understated by 3 percentage points.
Michael Pettis has a bigger problem with these sorts of exercises. He writes to us:
The current account surplus is simply the excess of savings over investment. Countries with extraordinarily high investment rates should, therefore, have high current account deficits. China has the highest investment rate perhaps ever, and yet it has a huge current account surplus. That means by definition that its savings rate must be truly extraordinary, and of course savings is simply GDP minus consumption, so consumption must be extraordinarily low.
In Pettis’ view, it comes back to the household income. If households are effectively subsidising investment via artificially low interest rates, the central government’s renewed enthusiasm for measures to stimulate consumer purchasing won’t, in aggregate, change the household consumption share of GDP.
Sure, consumption of automobiles and white goods surged back then, just as you would have expected given the subsidies, and they will surge again no doubt, but since those subsidies were ultimately paid for by the household sector, the policies did not translate into an overall surge in consumption because there was no net increase in household wealth. In fact during both of those years consumption continued to decline sharply as a share of GDP.
That’s the case against.
But what if BarCap really is onto something and consumption share has improved a few points since 2008? As this chart kindly provided by Mark Williams of Capital Economics shows, a three percentage point increase in household consumption would still leave China’s household consumption-to-GDP ratio at a rate lower than any country has seen in modern times.
The chart plots both the official and the BarCap numbers for China, against the lowest household consumption/GDP ratio recorded by the World Bank (in most cases this goes back to 1961, Williams says):