# 3.2 Evidence of convergence

Evidence from the APEC region is somewhat consistent with the convergence hypothesis that under certain conditions low-income economies will tend to grow faster than high-income economies. However, evidence is mixed and some economies are converging while others are not.

If incomes of APEC economies were converging we would expect to see a negative correlation between initial income levels and the subsequent growth rate of incomes, since low-income economies would be growing faster than high-income economies. Figure 3 plots the real average per capita income level of APEC economies in 1989 against the real average annual growth rate of per capita income for the period 1989 to 2005. The figure shows some evidence of convergence in the region. However, progress is patchy and there is a cluster of economies with low initial incomes that are not “catching up”.

- Figure 3: Convergence of APEC economies’ incomes, 1989-2005
- Source: United Nations Statistical Database.
- Note: Data for Russia for period 1990-2005 due to data gaps. Excludes Chinese Taipei due to data gaps. Key included in Appendix 2.

Figure 4 plots the real average per capita income levels of APEC economies relative to the United States in 1989 and 2005. If an economy was “catching up” to average income levels in the United States over the period then they would appear to the right of the 45º line, and if they were “falling behind” they would appear to the left of the 45º line. Again, while there is some evidence of convergence in the region, there are a number of economies whose per capita income levels have fallen behind relative to the United States over the period 1989 to 2005.

- Figure 4: Relative income levels of APEC economies, 1989 and 2005
- Source: United Nations Statistical Database.
- Note: Excludes Chinese Taipei and Russia due to data gaps. Key included in Appendix 2.

Of the higher-income economies, the relative incomes of New Zealand, Canada and Australia have stayed broadly the same over the period. Brunei Darussalam and Japan have performed relatively poorly. Brunei Darussalam’s growth performance has been poor since the early 1980s and Japan’s economic performance has slowed substantially since the early 1990s. Average per capita incomes in The Republic of Korea; Hong Kong, China; and Singapore have moved ahead relative to the United States.

Of the lower-income economies, per capita incomes in China, Chile, Indonesia, Malaysia, Thailand and Viet Nam moved ahead relative to the United States. However there are some low-income economies that are making little progress or are continuing to fall behind, such as Papua New Guinea, Peru and the Philippines.

It is evident from Figure 4 that economies starting from a low base of average income per capita, will, even with very high growth rates, take a very long time to converge toward income levels of developed economies. For example China’s very high annual average growth rate of 8.7% from 1989 to 2005 has moved their relative income from 1.5% to 4.2% of that of the United States over the period.

It is possible to empirically test whether convergence is occurring in the APEC region. We can do this by estimating the following cross-economy regression model, derived from the neoclassical growth model:

(1) _{ }

where *t* is the time period (annual), *y*(0) is the initial level of per capita income and *y*(*t*) is the level of per capita income in the final period. This model has been used by various researchers (for example, Baumol, 1986).[8] If per capita income convergence (often described as “catch-up”), was occurring, then we would expect to see a negative relationship between initial per capita income levels and subsequent growth rates (ie, we would expect the *b* coefficient in regression model (1) to be negative).

The results of regression (1) are summarised in Table 1. Looking at the 16 years prior to the formation of APEC (1972-1988), there is no evidence of per capita income convergence in APEC or the rest of the world. However, in the period since APEC was formed in 1989, there is evidence of convergence in the APEC region. The estimates suggest that the annual average rate of convergence in the APEC region over the period 1989-2005 is 1.47%. This means that the estimated “half-life” (the time taken to close half of the gap between income levels of high- and low-income economies) for APEC is 47 years. These results are based on real per capita incomes expressed in $US.

If per capita GDP data are measured in purchasing power parity (PPP) terms (to account for differences in the relative prices of goods and services in different economies) then the estimated speed of convergence is slower. For the period 1989-2005, the annual average rate of convergence in the APEC region (for the 18 APEC economies for which data are available) is 0.52%. This means that the estimated half-life is 134 years.[9]

Estimated absolute β-convergence | ||||
---|---|---|---|---|

1972-1988 | 1989-2005 | |||

APEC Economies* | Rest of the World** | APEC Economies* | Rest of the World** | |

GDP (b) | -0.0044(0.1542) | 0.0003(0.8488) | -0.0131(0.0021) | 0.0023(0.1988) |

Annual average rate of convergence | 0.456% | -0.030% | 1.470% | -0.235% |

Half-life | 158 years | 47 years | ||

R^{2} |
0.1157 | 0.0003 | 0.4363 | 0.0104 |

Note: Probability values are in parentheses.

* APEC member economies excluding Russia and Chinese Taipei due to data gaps.

** Economies from the rest of the world (excluding APEC economies) where data are available for the period 1972-2005 from the United Nations Statistical Database.

Engelbrecht and Kelsen (1999) also empirically test whether there is evidence of convergence amongst APEC economies during from 1965 to 1990. They find that APEC economies were indeed converging towards one another over the sample period. This result holds for both “conditional β-convergence” (ie, convergence conditional on institutional and structural characteristics) and “unconditional or absolute β-convergence”.[10] This suggests that APEC constitutes a “convergence club” similar to the OECD and the European Union. This result is of particular note, given the diversity of APEC economies.

Our regression results suggest that in general convergence has continued. However, Figures 3 and 4 indicate that the process is characterised by some low-income economies achieving very high growth, some merely matching the growth of higher-income economies, and some economies falling behind.

Another way to think about income convergence is in terms of how the dispersion of per capita income levels changes over time. This is commonly referred to as σ-convergence, and is calculated as the standard deviation of the log of per capita income. s-convergence is exhibited among a group of economies when the dispersion of per capita income decreases over time, ie, where:

σ(t) < σ(0). (2)

Figure 5 graphs the dispersion of per capita income levels (σ) for the period 1972-2005. The graph shows that the dispersion of per capita income is falling in APEC, while it is increasing in the rest of the world, ie, there is evidence of σ-convergence in APEC, but not in the rest of the world.

- Figure 5: Dispersion of per capita income levels
- Source: United Nations Statistical Database.
- Note: Excludes Chinese Taipei and Russia due to data gaps. Includes 161 economies from the rest of the world (excluding APEC), where data are available for the period 1972-2005 from the United Nations Statistical Database.

#### Notesocc

- [8]“Absolute b-convergence” is the notion that low-income economies will tend to grow faster than high income economies, independent of any other characteristics of the economy. It is also possible to estimate “conditional b-convergence”: the notion that low-income economies will tend to grow faster than high income economies, if other factors are held unchanged (such as initial levels of human capital). We have estimated absolute β-convergence in this paper.
- [9]The GDP per capita purchasing power parity (PPP) data comes from the World Development Indicators Database. The sample of 18 economies excludes Russia, Brunei Darussalam and Chinese Taipei due to data gaps.
- [10]Engelbrecht and Kelsen (1999) obtain estimates of the rate of absolute β-convergence that range around our estimate (from -0.0079 to -0.0178) depending on the method of estimation that they use. Their results are likely to be different from ours due to the different time periods examined, as well as the different estimation methods used. We use Ordinary Least Squares (OLS) estimation.