RDP 2012-04: Chinese Urban Residential Construction to 2040 5. Sensitivity Analysis

This section evaluates the sensitivity of the baseline projection for urban residential construction to different assumptions. We evaluate these assumptions in isolation, but it is certainly possible that they could be correlated. For example, a more rapid increase in the urbanisation rate than projected may be accompanied by higher than projected GDP growth. The implications of these different assumptions for steel use are presented in Appendix B.

5.1 Urbanisation Assumptions

While our urbanisation projections are close to the current UN projections, they differ substantially to the projections the UN offered in 2009 (Figure 12). This difference primarily reflects data updates over the past two years, which have shown an unexpectedly strong increase in the urbanisation rate. Under the UN methodology this recent rapid growth has a persistent effect, so that the gap between the two projections reaches 7½ percentage points by 2025. There is some chance that the recent data are anomalous, and that the UN's 2009 extended projections better capture the prospects for urbanisation. Therefore, we consider an alternative where the urbanisation rate approaches the UN's 2009 numbers by 2040.^[21] Under this alternative path, floor space construction does not exhibit any appreciable growth from here on in and it begins to fall in 2014 (Figure 13). The downward trajectory arises because the growth of the urban population slows rapidly.

5.2 Floor Space per Capita Assumptions

The estimates of the relationship between floor space per capita and income are imprecise. We consider a higher and lower alternative. On the upside we impose a coefficient of 15 on the log of GDP per capita in its long-run relationship with floor space per capita, as opposed to 10 in the baseline. This is at the lower end of the coefficients found from IEA data, but at the upper end of those found from the Taiwanese data (Table 2). We use a coefficient of 7.5 as a downside alternative, which is closer to the relationship found in China after 2003.

The resulting paths tell quite different stories (Figure 14). The upside projection sees the growth of the past decade taper off quite slowly, with construction peaking in 2023, 30 per cent above the recent levels. Alternatively, if floor space is less responsive to income, construction remains roughly flat for the next five years before declining.

5.3 GDP Growth Assumptions

Given uncertainty about China's trend rate of growth, we consider two possibilities: a high-growth and a low-growth alternative.

• First we assume that the Chinese GDP per capita grows at 8 per cent per annum for the next 20 years, rather than around 6 per cent in the baseline. Lin (2011) suggests that China has the potential to do so considering China's stage of development relative to that of the United States, and the growth performance of Japan, Taiwan and Korea when they were at a similar stage of development. In this case we project that China's growth rate will follow the same deceleration after 2032 as the growth rate from World Bank and Development Research Center (World Bank and Development Research Center of the State Council, the People's Republic of China) (2012). Accordingly, China's PPP-adjusted GDP per capita reaches almost US$80,000 by 2040 (at 2005 prices), compared to around US$40,000 in the baseline projection.
• In the low-growth alternative we assume that GDP per capita growth is 2.7 percentage points lower than the baseline growth rate. This difference is equivalent to one standard deviation of the distribution of growth rates experienced in China between 2000 and 2009.^[22] In this case, China's PPP-adjusted GDP per capita reaches US$20,000 by 2040 (at 2005 prices).

Under the high-growth assumption, floor space construction peaks in 2023, 25 per cent above recent levels, and remains above current levels for all but the last couple of years of the projection period (Figure 15). Under the low-growth assumption, construction is effectively at its peak now and will be almost 40 per cent below the baseline by the end of the projection period.

Footnotes

To do this, we assume that the rapid growth over the past 2 years only influences the projection for the next 10 years rather than the next 25 as specified in footnote 4, i.e. W_t declines by 0.1 each year rather than by 0.04. [21]

This standard deviation is calculated from growth rates from the Penn World Tables. [22]