A Quality of Growth Index for Developing Countries: A Proposal

Abstract

This paper proposes a new quality of growth index for developing countries. The index encompasses both the intrinsic nature and social dimensions of growth, and is computed for over 90 countries for the period 1990–2011. The approach is premised on the fact that not all growth is created equal in terms of social outcomes, and that it does matter how one reaches from one level of income to another for various theoretical and empirical reasons. The paper finds that the quality of growth has been improving in the vast majority of developing countries over the past two decades, although the rate of convergence is relatively slow. At the same time, there are considerable cross-country variations across income levels and regions. Finally, empirical investigations point to the fact that main factors of the quality of growth are political stability, public pro-poor spending, macroeconomic stability, financial development, institutional quality and external factors such as FDI.

Appendices

Appendix 1

See Appendix Table 4.

Table 4 Data sources and definition

Appendix 2. Dealing with Missing Observations in the Primary School Completion Rate

Observations for primary school completion rate are missing for some countries. In order to avoid reducing considerably the sample size when using this indicator as a proxy for the social dimension of the quality of growth, we estimate these missing observations according to the following approach:

• 1. First, we compute for each country the primary completion rate (Complest) by dividing the average years of primary schooling (from Barro and Lee 2013) with the average duration of primary school (from WDI, 2010).

• 2. Second, to ensure that the computed completion rates (Complest) are consistent with the actual rates, we proceed as follows:

  • For a given country at a given year, if the actual completion rate (Complact) is available, then we keep this Complact as the completion rate (Compl) to be considered for the calculation of the QGI index.

  • For a given country at a given year, if Complact is missing, then we consider Complest adjusted for the average deviation between Complest and Complact (in the neighborhood of the year for which Complact is missing). The neighborhood considered spans from the 5 years preceding the year for which the observation is missing to the 5 years following the year for which the observation is missing.¹⁹

    $$Compl_{it} = Compl\_est_{it} + \lambda_{it}$$

    where i et t stands for country i and period t, respectively, with $\lambda_{it} = mean(Compl\_act_{ij} - Compl\_est_{it} ),\quad j = \overline{t - 4,\;t + 4}$.

Appendix 3

See Appendix Table 5.

Table 5 Country ranking using geometric mean-based QGI

Appendix 4

See Appendix Table 6.

Table 6 QGI for the full sample over 1990–2011: descriptive statistics

Appendix 5

See Appendix Table 7.

Table 7 Correlation matrix of the alternative QGIs

Appendix 6

See Appendix Table 8.

Table 8 Spearman’s rank order correlation test (Arithmetic mean-based vs. Geometric mean-based QGI)

Appendix 7

See Appendix Table 9.

Table 9 Spearman’s rank order correlation test

Appendix 8

See Appendix Table 10.

Table 10 Correlation matrix between the benchmark QGI and the region-specific QGI

Appendix 9

See Appendix Table 11.

Table 11 Determinants of quality of growth on different subsamples

Appendix 10

See Appendix Table 12.

Table 12 Descriptive statistics for additional robustness checks-related QGI (full sample over 1990–2011)

Appendix 11

See Appendix Table 13.

Table 13 Correlation matrix for additional robustness checks-related QGI (full sample over 1990–2011)