3.1.4 Economic growth and convergence
Determinants of long-term GDP per person include labour force and its productivity projections. Labour force utilization depends on factors such as the number of working-age people, the level of structural unemployment and hours worked per worker. Demographic change is still the major determinant of the baseline labour supply (Martins and Nicoletti, 2005). Long-term projections of labour productivity primarily depend on improvements in labour quality (capacity building) and the pace of technical change associated with building up the capital-output ratio and the quality of capital.
The literature examining production functions shows increasing returns because of an expanding stock of human capital and, as a result of specialization and investment in ‘knowledge’ capital (Meier, 2001; Aghion and Howitt, 1998), suggests that economic ‘catch-up’ and convergence strongly depend on the forces of ‘technological congruence’ and ‘social capability’ between the productivity leader and the followers (see the subsequent sub-section on institutional frameworks and Section 3.4 on the role of technological change).
The economic convergence literature (Abramovitz, 1986; Baumol, 1986), using a standard neoclassical economic growth setup following Solow (1956), found evidence of convergence only between the richest countries. Other research efforts documented ‘conditional convergence’ – meaning that countries appeared to reach their own steady states at a fairly uniform rate of 2% per year (Barro, 1991; Mankiw et al., 1992). Jones (1997) found that the future steady-state distribution of per person income will be broadly similar to the 1990 distribution. Important differences would continue to arise among the bottom two-thirds of the income distribution, thus confirming past trends. Total factor productivity (TFP) levels and convergence for the evolution of income distribution are also important. Expected catch-up, and even overtaking per-person incomes, as well as changes in leaders in the world distribution of income, are among some of the findings in this literature. Quah (1993, 1996) found that the world is moving towards a bimodal income distribution. Some recent assessments demonstrate divergence, not convergence (World Bank, 2002; Halloy and Lockwood, 2005; UNSD, 2005).
Convergence is limited for a number of reasons, such as imperfect mobility of factors (notably labour); different endowments (notably human capital); market segmentation (notably services); and limited technology diffusion. Social inertia (as referred to in Chapter 2, see Section 2.2.3) also contributes to delay convergence. Therefore only limited catch-up can be factored in baseline scenarios: while capital quality is likely to push up productivity growth in most countries, especially in those lagging behind, labour quality is likely to drag down productivity growth in a number of countries, unless there are massive investments in education. However, appropriate policies may accelerate the adoption of new technologies and create incentives for human capital formation and thus accelerate convergence (Martins and Nicoletti, 2005). Nelson and Fagerberg, arguing within an evolutionary paradigm, have different perspectives on the convergence issue (Fagerberg, 1995; Fagerberg and Godinho, 2005; UNIDO, 2005). It should be acknowledged that the old theoretical controversy about steady-state economics and limits to growth still continues (Georgescu-Roegen, 1971).
The above discussion provides the economic background for the range of assumptions on the long-term convergence of income between developing and developed countries (measured by GDP per person) found in the scenario literature. The annual rate of income convergence between 11 world regions in the SRES scenarios falls within the range of less than 0.5% in the A2 scenario family to less than 2% in A1 (both in PPP and MER metrics). The highest rate of income convergence in the SRES is similar to the observed convergence, during the period 1950–1990, of 90 regions in Europe (Barro and Sala-i-Martin 1997). However, Grübler et al. (2006) note that extending convergence analysis to national or sub-national level would suggest that income disparities are larger than suggested by simple inter-regional comparisons and that scenarios of (relative) income convergence are highly sensitive to the spatial level of aggregation used in the analysis. An important finding from the sensitivity analysis performed is that less convergence generally yields higher emissions (Grübler et al., 2004). In B2, an income ratio (between 11 world regions, in market exchange rates) of seven corresponds to CO2 emissions of 14.2 GtC in 2100, while shifting this income ratio to 16 would lead to CO2 emissions of 15.5 GtC in 2100. Results pointing in the same direction were also obtained for A2. This can be explained by slower TFP growth, slower capital turnover, and less ‘technological congruence’, leading to slower adoption of low-emission technologies in developing countries. On the other hand, as climate stabilization scenarios require global application of climate policies and convergence in the adoption of low-emission technologies, they are less compatible with low economic convergence scenarios.