Chinese trend growth: Productivity growth - the Solow perspective
In the first note of this trilogy, we looked at what a declining population means for China’s growth and concluded that the impact was minimal, and outstripped by the effects of urbanisation but even more by the effects of labour productivity growth. Here, we examine productivity growth in more detail.
For the workforce, productivity can be boosted by the simple expedient of adding more capital (while technological progress is also an important driver particularly in more advanced economies).
To give a hugely simplified example, more machines mean workers can produce more goods. This then raises the question of how much capital can be added, and for what return.
Clearly, at some point the addition of more machines in a factory would not raise the output per worker, because each worker would already be fully occupied utilising the existing machinery.
We might then suppose there is some optimum level of capital per worker, beyond which it does not pay to invest in additional units. This is the starting point of the Solow growth model.
In the Solow growth model, economic growth is driven by the accumulation of physical capital until this optimum level of capital per worker, the so-called “steady state”, is reached.
The steady state itself is determined by labour force growth, the savings rate, and the rate of depreciation.
The model predicts more rapid growth when the level of physical capital per capita is low, something often referred to as “catch up” growth.
When the steady state is reached, growth in per capita incomes is determined entirely by technological progress.
All things being equal, therefore, the model predicts that emerging markets should grow faster than developed markets, and this is generally what we see. So what does the model predict for China?
Chart 1 shows the estimated capital stock per worker in China, South Korea and the US.
In the simplest form of the Solow model, we would assume China’s convergence over time with the US, which implies that there is a great deal of investment left to go given a gap of over $90,000 per head.
Yet it is possible that the US is the wrong example to choose for China; the countries have different savings rates in both physical and human capital, so convergence is not assured.
We might instead look at Korea, which bears greater similarities to China in a number of ways, but importantly this includes the behaviour of fertility and saving, rendering it a more likely point for convergence.
It turns out that the scope for catch up is still significant, so again Solow’s model still implies a growth rate above that seen in developed markets for as long as it takes to close the gap.
We would also argue that the disparity in capital stocks demonstrates very clearly that China has not exhausted her investment opportunities, as some claim.
The next question to ask then is how long this catch up process is likely to take.
Given that we would like to use the results of this analysis to provide estimates of trend growth, using growth forecasts to predict the path of investment would be rather circular.
Instead, given that we have identified Korea as a likely candidate for Chinese convergence, let us look at the Korean experience.
Chart 2 shows the accumulation of capital stock in Korea and China, with the timeline beginning when the per capita capital stock stood at approximately $2,500.
The thirty or so years following this point seem to have generated similar experiences in both countries, so convergence looks a plausible theory.
This implies that in around 15 years, China will be close to Korea’s current level of capital per worker.
Hitting the wall
Of course, Korea’s path includes a rather large bump in the form of the Asian financial crisis, and assuming identical timelines would place China just two years from that moment at most.
This is a much cited worry; that no economy has invested at the rate China has without hitting some crisis as a result.
While we would not rule this out, we would note that the Korean experience was in large part due to a high reliance on foreign capital which rendered the economy susceptible to a “sudden stop”.
China, by contrast, has almost entirely self-funded its investment. This is not to say there are no risks, there are many, but there are limits to what history can tell us about the future.
With this caveat in mind, we will look to extrapolate the likely path of Chinese investment growth from the current similarity to Korea’s.
Assuming convergence with Korea by 2030 implies growth in the capital stock of between 6.7% and 7.0% for the next 15 years, declining to the lower bound over time.
At this point, apologetically, we return once more to Solow, but this time to the growth accounting approach he did much to foster.
Total factor productivity
Under this approach, growth can be decomposed into three main elements: growth in the labour force, growth in the capital stock, and growth in what is commonly called total factor productivity (TFP); essentially the efficiency with which we combine labour and capital, and which is boosted by technological progress.
So to implement this approach to estimate future Chinese growth, we will need estimates of each component.
From our analysis above, we have obtained a figure (6.94%) for growth in the capital stock, and the UN produces regular working age population forecasts, which gives a number for labour force growth. So all that remains is to obtain a figure for TFP growth.
This is likely the most contentious part of our analysis given the difficulties in measuring TFP. We draw on data from the Penn World Table in what follows.
The most recent data suggest TFP growth in China was around 3% on average in 2005-10, but this is distorted by the impact of the crisis; removing that effect leaves TFP growth looking more like 2% annually.
For the US, it looks to be around 1%, and in Korea, roughly the same – though the series is volatile.
It seems unlikely that China will be able to maintain higher TFP growth rates indefinitely – much of the larger gains (some years show TFP growth of 6%+) came earlier on, possibly reflecting the massive boost to productivity from adopting radically more advanced technologies as the country industrialised.
As the level of technology converges with that in Korea or the US, TFP growth rates should do the same. Consequently, the contribution of TFP growth to overall GDP growth should diminish over time.
While we have so far framed TFP entirely in terms of technology, other factors can also impact it. Recall that we described TFP as the efficiency with which inputs are combined into outputs.
While technology undoubtedly plays a large role in this, government regulation also has an effect particularly give its impact on the efficiency of resource allocation.
In this simple model, the resources in question are labour and capital. On labour, restrictions on labour mobility (in the form of the hukou registration system) will reduce efficiency and hinder TFP.
On capital, a system under which credit is directed by a series of quotas and mandated lending is also likely to lead to inefficient allocation, as is evidenced by the excess capacity in China’s heavy industry and property sectors.
Reforms on both fronts (as are planned) can therefore boost TFP; the success or failure of market friendly reforms could have significant growth consequences.
In Chart 3, we combine our estimates for labour and capital stock growth with a number of TFP scenarios.
In our best case, China manages to successfully implement market-based reforms, such that although TFP drops from the current 2% level in 2020, it continues to see growth in line with America’s TFP of around 1% per annum.
We assume that China continues to benefit from innovation both domestically and globally, with reforms enabling the full integration of technology on a par with the rate of adoption elsewhere.
Before readers scoff at this assumption, we would refer to China’s experience in e-commerce and consumer telecommunications, where in many respects they have equalled or even surpassed Western counterparts.
Alongside this, in the best case scenario, is the assumption that China manages to boost labour force participation, particularly among the older segment (50-64) of its workforce.
We discussed this in more detail in the first part of this series, but essentially participation amongst this segment is far below that in the West or Korea.
A gradual (but not complete) catch-up on this front helps cancel out the effects of a declining workforce.
The other scenarios are versions of our best case, but dropping one or more assumptions – no labour force participation increase, no reform progress, or both.
The no reform assumption reduces TFP growth to 1% in the first five-year window, and 0% thereafter, with significant GDP growth consequences. This gives us a range of potential trend growth numbers.
At one end of the scale, this provides our best case numbers of 5.7% trend growth for the next five years which then drops to 4.4% by 2030. In the worst case, trend growth today is only 4.4%, and falls to 3.1% by the final period.
The chief driver of the differences lies in the TFP assumptions we have made, so radical reforms, technological breakthroughs, or political crisis all present obvious risks to the forecast.
Overall though, it is encouraging to see a similar range of predictions as generated by our previous work looking at the demographic implications for trend growth.
Once again, we have a trend growth number some way below the official GDP number. So given that GDP tends to revert to trend, this presumably implies a significant correction, perhaps even a hard landing to come.
But as we have said, the official numbers are somewhat in doubt. With apologies to those who read our last note, we reproduce the argument below.
Depending on who you talk to, some would have Chinese growth as low as 2% (and possibly even lower).
But in that case, the correction has already happened, and we should expect an upturn in Chinese growth to return to a trend growth number of over 5%. Our view would be that Chinese growth is somewhat overstated, though chiefly through statistical errors and perhaps a tendency to round up, instead of down.
Capital Economics, for example, estimate that incorrect calculation of the GDP deflator might be overstating real GDP growth by 1-2% in the first half of 2015 - which brings the true growth number much closer to our estimate of trend growth.
Aside from this, there are few of the usual signs of an economy growing above trend; inflation pressures are extremely muted, and the Producer Price Index (PPI) is in fact in deflationary territory.
Analysis from the International Monetary Fund (IMF) of the Chinese output gap, estimates a negative gap under the usual methodology (i.e. the economy is below trend), but a positive gap when using a “credit augmented” measure1, and this is where we see the risk.
So, by traditional metrics, China is below trend, in that spare capacity is rampant. But this spare capacity has been achieved through above-trend credit and investment growth which has to now unwind.
We include a “China hard landing” risk scenario in our economic forecasts, but it arises from financial instabilities rather than an overheating economy.
The high and growing level of indebtedness and the opacity of balance sheets create concerns over lurking asset quality issues beyond those officially stated by banks.
A financial crisis would have the potential to take growth significantly below trend, and could, as in the West, see a prolonged period of below-trend growth as the rubble is cleared. But again, this is not a result of the usual overheating typical of the end of a business cycle
Still, it would be extremely unusual for the economy to undergo such an adjustment without some cyclical volatility, and a recession at some point seems a certainty.
Further, investors will need to get used to a world in which China in all likelihood posts a lower growth number every year for at least a decade, depending on the speed of adjustment.
Rather than taking comfort from government stimulus, we would become concerned if policymakers became modern day King Canutes, trying to halt, in their case, a receding tide by unleashing stimulus after stimulus.
The more adjustment is deferred, the more painful it will ultimately prove, and there is no policy that can successfully prevent this kind of structural slowdown.
To our minds, every attempt to hold the growth rate above trend increases the risk of crisis, notwithstanding the measurement errors we mentioned above.
One other implication of our work here is that China still has significant scope for capital investment.
Yet media reports of Chinese spare capacity are almost as abundant as the country’s steel production, so how can the two be compatible – is this just extreme naiveté on our part?
Our third and final instalment will examine where we think this capital could go, and the potential for consumption growth and service sector development in the new China.
1. Maliszewski, W., Longmei, Z. “China’s Growth: Can Goldilocks Outgrow Bears?” IMF Working Paper 15/113↩