What Drives the Linkage?

At least three general factors are widely believed to explain differences in the poverty reduction response to economic growth across countries, namely:

1. Sectoral effects (composition of output and of output growth; sectoral growth rates)
2. Public investments (e.g., in health and education, housing, agriculture)
3. Quality of governance

In the discussions below, the effects of these factors are first examined individually, using graphical (scatterplot) and pairwise simple regression analysis. A multiple regression analysis is further undertaken to examine possible interactions among the explanatory variables in determining the magnitudes of the PEG.

4.1 Sectoral Composition of GDP and GDP Growth

Will an agriculture-dominated economy tend to have more inclusive growth than one dominated by manufacturing, or by services? Will agriculture-driven growth lead to more inclusive growth?

It is commonly believed that an agriculture-led economy and agriculture-led growth promotes faster rural poverty reduction, and because rural poverty tends to dominate overall poverty in most countries, overall poverty is expected to fall faster with an agriculture-driven economic growth. It has been noted, however, that this does not necessarily apply for all countries. Indeed, as mentioned earlier, Hasan and Quibria (2004) cautioned that the experience has varied across regions, and while agriculture had been an important driver of employment creation and poverty reduction in East/Southeast Asia, it was manufacturing that appeared more instrumental in South Asia for generating much employment and driving down poverty.

Table 7 [ PDF 46.2KB | 1 page ] shows that most Asian economies have in fact been dominated by the services sector in the past decade. The PRC is a notable exception, where manufacturing has been the largest sector in the economy. Based on sectoral contribution to economic growth, PRC, Thailand, and to some extent Viet Nam, have also been exceptions in the region, with manufacturing being a prominent driver of economic growth. Agriculture had been a relatively minor contributor, except in the case of Myanmar, where it dominates, as well as Mongolia and Nepal, where the manufacturing sector is relatively miniscule.

An examination of the relationship between the poverty elasticity of growth and sectoral shares in GDP based on scatterplots and simple regressions yields no clear systematic relationship in both periods (see Figure 2a-2b [ PDF 46.2KB | 1 page ], Figure 3a-3b [ PDF 46.3KB | 1 page ], to Figure 4a-4b [ PDF 45.7KB | 1 page ]). Simple regressions yield insignificant coefficients and very little explanatory power (see Table 8 [ PDF 47.2KB | 1 page ]).

Similarly, there is weak evidence of any systematic relationship between sectoral contributions to GDP growth and the PEG in both periods, especially for agriculture and services (see Figure 5a [ PDF 45.7KB | 1 page ], Figure 5b [ PDF 45.7KB | 1 page ], Figure 5c [ PDF 45.6KB | 1 page ] to Figure 6a [ PDF 45.6KB | 1 page ] - Figure 6b [ PDF 45.7KB | 1 page ]). For agriculture's growth contribution, isolating the Southeast Asian countries (Figure 5c [ PDF 45.6KB | 1 page ]) does not yield the expected relationship based on the earlier observations by Hasan and Quibria (2004) (see further below). Indeed, Figure 5c [ PDF 45.6KB | 1 page ] and Table 9 [ PDF 52.5KB | 1 page ] even suggest a puzzling perverse effect whereby a stronger agriculture growth contribution is associated with less inclusive growth (i.e., a positive coefficient). These results run counter to the widely held belief that agriculture-driven growth is crucial to poverty reduction.

A similar result is obtained when services sector contribution to growth is plotted against the PEG (Figure 6a [ PDF 45.6KB | 1 page ] - Figure 6b [ PDF 45.7KB | 1 page ]); no systematic relationship is readily apparent. On the other hand, some semblance of an influence may be seen in the case of manufacturing's contribution to growth, particularly in 2000–2006 (Figure 7b [ PDF 45.6KB | 1 page ] - Figure 7c [ PDF 45.6KB | 1 page ]), but not in 1990–1996 (Figure 7a [ PDF 45.7KB | 1 page ]). The relationship appears to be stronger and regression coefficients become significant (Table 9 [ PDF 52.5KB | 1 page ])¹¹ when the analysis is confined to Southeast Asian countries (Figure 7c [ PDF 45.6KB | 1 page ]).

The above finding provides some indication that the manufacturing sector may have taken a more important role as driver of employment and poverty reduction especially in Southeast Asia in recent years. This is a departure from the earlier experience observed by Hasan and Quibria (2004) for the 1990s, when agriculture was seen to have been more instrumental to inclusive growth in Southeast Asia, while light manufacturing played the same role in the case of South Asia.

Finally, using sectoral growth rates directly as explanatory variables fails to yield any significant relationship with the PEG, as indicated by dispersed scatterplots and insignificant regression estimates (Annex Figure 1a-c [ PDF 47.3KB | 2 page ]; Annex Table 2 [ PDF 47.7KB | 1 page ]).

4.2 Public Expenditures

Data on public expenditures derived from the annual ADB Key Indicators of Developing Asian and Pacific Countries are summarized in Table 10 [ PDF 52.4KB | 1 page ]. For our purposes, we examine public expenditures on health, education, housing, and the agricultural sector as candidate variables that would influence the size of the PEG.

The following observations emerge from the table:

• Malaysia and Mongolia have made the largest public investments in education, exceeding 6% of GDP, while on the other extreme is Indonesia, with less than 1% of GDP being spent by government on education.
• Mongolia has made the largest public investments in health, while Indonesia and the Philippines have made the smallest, at less than half a percent of GDP.
• Malaysia has made the largest public investments in education and health combined, at over 7% of GDP, while Indonesia emerges as having made the smallest.
• Singapore has made the largest public investments in housing (at more than 2% of GDP), while the Philippines emerges as having made the smallest, with a miniscule 0.09% of GDP.
• Nepal and Thailand have made the largest public investments in agriculture, while Singapore (not surprisingly) has made a relatively miniscule public investment in the sector.

Scatterplots and simple regression results on pairwise relationships between these categories of expenditures and PEG are presented in Figure 8 [ PDF 49.2KB | 1 page ], Figure 9 [ PDF 46.4KB | 1 page ] to Figure 10 [ PDF 43.7KB | 1 page ] and Table 11 [ PDF 49.2KB | 1 page ].

The data point to a close correlation between public expenditures on health and education and the poverty elasticity of growth (see Figure 8 [ PDF 49.2KB | 1 page ] and Table 11 [ PDF 49.2KB | 1 page ]). With R² of 0.61 (solid line in Figure 8 [ PDF 49.2KB | 1 page ]), the variation in education and health expenditures expressed as a percentage of GDP would appear to account for close to two thirds of the variation in the PEG. This correlation is even stronger if one eliminates outlier Sri Lanka from the analysis; in this case, the R² rises to 0.81 (dotted line in Figure 8 [ PDF 49.2KB | 1 page ]). These results support the findings of Anand and Ravaillon (1993); Bidani and Ravaillon (1997); Self and Grabowski (2003); and Baldacci, Guin-Siu, and De Mello (2003), among others, which are all contrary to the observations made by Carrin and Politi (1996), Filmer and Pritchett (1999), Landau (1986), and Al- Samarrai (2002), who found weak correlation between public spending and social development outcomes. It must be noted that the present analysis differs from the others cited in that it examines the influence of public expenditures not directly on social outcomes, but on the responsiveness of social outcomes to economic growth.

Similarly, public expenditures on agriculture do not appear to have any discernible systematic relationship with the value of the PEG (see Figure 10 [ PDF 43.7KB | 1 page ]), yielding insignificant coefficient estimates and a very low R² of 0.01. This observation need not be surprising, though, as agriculture expenditures can take a wide variety of forms, and the nature and quality of such expenditures differs widely across countries and across time, thereby negating the appearance of any systematic relationships that aggregate figures alone could reveal. One cannot therefore conclude readily from this that public investments in agriculture are not warranted and must assume lower priority. That is, the above result may simply be a reflection of the wide scope for variation in the nature and quality of expenditures undertaken by governments for agriculture, including likely differences in attribution of various types of expenditures to the sector.

4.3 Quality of Governance

The best available measure of quality of governance is the Kaufmann and Kraay (2008) series on World Governance Indicators now published annually by the World Bank, and based on a compilation of results of available regular perception surveys. Inasmuch as the earliest year for which the governance index has been estimated is 1996, the analysis could not be done for the 1990-1996 interval, hence is only done for the 2000–2008 period. For this purpose, the index reported for 2005 was used for the analysis, which was taken to adequately reflect the general state of governance during the time interval analyzed. Table 12 [ PDF 45.4KB | 1 page ] gives the data used for the analysis, while Annex Tables 3a to 3f [ PDF 63.9KB | 6 page ] give the detailed governance index data for 1996 to 2007 for the Asian countries studied.

To check for systematic relationships between the PEG and governance indicators, the PEG is plotted and regressed against the six component indicators of quality of governance as defined in Kaufmann and Kraay (2008) (see Box [ PDF 48KB | 1 page ]).

Figures 11 to 17 [ PDF 57.8KB | 4 page ] show the respective scatterplots, while Table 13 [ PDF 47.3KB | 1 page ] summarizes the regression results. The specific governance indicators that emerge as having significant bearing on the poverty elasticity of growth are political stability/control of violence, government effectiveness, and rule of law, although their explanatory power only ranges from 16–26% of the variation in PEG. Overall quality of governance, i.e., the average of the six indicators, also has a significant relationship with the PEG. Among the indicators, political stability/absence of violence has the strongest explanatory power, with 26% of the variation in PEG explained by the model (i.e., R2 of 0.26).

4.4 Multiple Regression Analysis

Multiple regression equations were estimated to consider the joint effects of sectoral contributions to growth, public expenditures, and quality of governance. Table 14 [ PDF 48.1KB | 1 page ] gives the best regression results obtained from different combinations of the three variables. Best results were obtained with contribution of agriculture to GDP growth, overall (average) governance index, and public expenditures in education and health as explanatory variables, i.e.,

PEG = F(AgrCont, Gov, EH)

where AgrCont is the contribution of agriculture to GDP growth, Gov is the overall (average) governance index, and EH is public expenditures in education and health. With an adjusted R² of 0.80, the joint effect of agriculture-driven growth, good governance, and social expenditures by the government appear to well explain the variation in PEG across Asian countries. Contrary to the puzzling results obtained under pairwise correlation analysis, agriculture's role this time emerges as a significant determinant of the poverty elasticity of growth, in the expected direction. However, its impact on the PEG is still considerably weaker than those of governance and public expenditures on education and heath, with governance having the strongest effect.

These results affirm the importance of sectoral contributions to growth (particularly that of agriculture), public expenditures in education and health, and quality of governance in determining the rate of poverty reduction that accompanies economic growth. That is, inclusive growth in Asia has been enhanced when agriculture has a greater contribution to overall economic growth, when there is better quality of governance, and when more public investments are made in education and health, and housing.

Download this Paper [ PDF 324KB| 59 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.