Discussion Paper
No. 2012-62 | December 07, 2012
Rami Ben Haj Kacem
Monetary versus Non-Monetary Pro-Poor Growth: Evidence from Rural Ethiopia between 2004 and 2009

Abstract

The aim of this paper is to contribute to the debate on the pro-poor growth measurement techniques using monetary versus non-monetary indicators. In this context, an alternative method for introducing non-monetary indicators into monetary pro-poor growth analysis is presented. The method is based on the definition of a “Conditional Growth Incidence Curve” for each group of households with a common selected non-monetary characteristic. Additional information provided by the “Conditional Growth Incidence Curve” is useful for a more detailed pro-poor growth analysis. Empirical illustration using data from rural Ethiopia between 2004 and 2009 shows the utility and the limits of each measurement technique.

JEL Classification:

D30, I30, O12

Links

Cite As

[Please cite the corresponding journal article] Rami Ben Haj Kacem (2012). Monetary versus Non-Monetary Pro-Poor Growth: Evidence from Rural Ethiopia between 2004 and 2009. Economics Discussion Papers, No 2012-62, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2012-62


Comments and Questions



Anonymous - Invited Reader Comment
January 07, 2013 - 09:43
Report on "Monetary versus non-monetary pro-poor growth: evidence from Rural Ethiopia between 2004 and 2009" The idea of conditioning the growth-incidence curves on other characteristics, particularly well-being determinants, is straightforward but useful in practice. However I am not sure the author has fully discussed the analytical implication of computing growth-incidence curves of a monetary indicator of well-being conditioned on other indicators. In addition to that, the paper's stated objectives and related discussions are sometimes confusing. I have a few suggestions to offer which I think could improve the quality of the paper: Major suggestions: 1. Introduction: The paper's main contribution is buried in the last paragraph, the one that describes the paper's organization. This is not good practice. There should be an introductory paragraph (at least one) describing what the paper does, what's the contribution, the novelty, etc.This paragraph should appear somewhere before, and be independent of, the last introductory paragraph outlining the rest of the paper. 2. While I understand the author's interest in comparing his method to those of Klasen et al., I don't find the comparison very interesting. Moreover the non-monetary indicators are not even the same in the empirical application, therefore the author cannot say, for instance, that he is exploring what happens when the roles of conditioning and conditioned variable are exchanged. On the other hand, I think the author should compute the unconditional and conditional measures of pro-poor growth discussed in section 2. Now that would be a very interesting comparison. The author did it already for growth-incidence curves. Why not doing the same for all the other measures? Then the paper's emphasis would be in the empirical insights that ensue when conditioning is introduced. In my opinion this focus is more interesting, and relevant, than the comparison against the proposals by Klasen et al. Although the author should still emphasize the difference between his proposal and theirs (but not with an empirical illustration). 3. The author's analysis seems to be anonymous (i.e. he is computing the growth of consumption quantiles as opposed to the growth of household's consumption). However, with a panel dataset like the Ethiopian one, the author could perform both types of analyses and see how the conditioning interacts with anonymity (or lack thereof). The reference to follow and cite for non-anonymity in pro-poor growth is a paper by Michael Grimm which appeared in the Journal of Economic Inequality in 2007 ("Non-anonymous pro-poor growth", or something like that). 4. When discussing his results, the author does not distinguish between absolute and relative pro-poor growth in the context of growth-incidence curves. For instance, growth could be absolutely pro-poor if g(p)>0 for the lowest values of p at least until certain threshold beyond which poverty is deemed to be overcome, whereas growth could be relatively pro-poor ifdg(p)/dp<0 for all p between 0 and 1. The distinction should be made. 5. More importantly the author does not fully explore the implications of his results. What can we say meaningfully about pro-poor growth when other conditioning well-being indicators come to play? For instance, imagine that we use income and we use head's literacy as a dummy conditioning variable.Imagine that the growth rate g(p) is positive and has a negative slope for households with literate heads, whereas it is positive and has a flat slope for households with illiterate heads. But, imagine also that g(p) is higher among illiterate heads for all p! So what should we say? Is growth being pro-poor absolutely? Relatively? This is a major point that the author should address. 6. Another major point related to the previous one (point 5) is that a priori it is not easy to assess the well-being situation of the people who exhibit the poorest income among a group whose conditioning value (for the non-monetary variable) is high. Perhaps these households have a high "absolute" income (e.g. they have enough for their basic needs, etc.) even though they are the poorest within their cohort. So how do we consider them?This point also highlights the importance of distinguishing between relative and absolute poverty. 7. In terms of pro-poor growth and multidimensional poverty, the author may want to check a recent paper by Berenguer and Bresson (2012) that appeared in the Review of Income and Wealth. Minor comments: 1. The author may want to link up his section 2 to the recent work of Deutsch and Silber (2012) on pro-poor growth that appear in Economics E-Journal. 2. Equation 4: A delta is missing in the definition of g in the line below the equation. 3. Page 4: The sentence: "If the curve movesof the income distribution"is confusing. Please check and rewrite if necessary. 4. Equation 6: Shouldn't g be there instead of mu? 5. Page 5: "Decomposition error". What does the author mean by that? Why would the residual term be considered an "error"? The author later refers to it as an interaction term, which is more in tune with the literature. I don't think the interaction term should be considered an "error". 6. Explanation of GIC (e.g. top of page 6, after equation 8): the author should say that if g(p)>0 for all then growth is pro-poor in absolute terms.Then the author could say that if dg(p)/dp<0 for all p then inequality decreases and growth is pro-poor in relative terms. Same issue in equation 9. 7. Page 6: "cannot find the results". What does that mean? Is it maybe "cannot match the results"? 8. Page 6: Where it says "limit" it should read "limitation". 9. Page 8: "alleging poverty". Shouldn't that be "alleviating poverty"? 10. Wherever it says "for introducing" (or something similar) it would be better to say "In order to introduce" (e.g. top of page 9). Similar thing in page 10 with "for ensuring". 11. "method to data reduction". It should read "method for data reduction". 12. Page 10: "determinants of poverty". Which poverty? Monetary poverty? The author should clarify the point. 13. Page 10: "their results are not significant". In what sense, statistically? Or otherwise? The author should clarify what he means. 14. Page 14: "non-monitory". It should read "non-monetary".

rami bhk - Reply
January 10, 2013 - 21:44
Thank you very much. I will take into consideration your comments and suggestions. I would like to reply to some points:2. - The comparison is based on the limits of the Klassen and all’s method in general terms and not specific to the empirical application given in their paper. Indeed, the Klassen and all’s method is limited to non monetary characteristics having a significant variability over time. For that, for example, they use the average years of schooling and restrict the sample to adult household members aged between 20 and 30 to capture more dynamics of changes in the educational system. This paper presents an alternative method which allows using all kind of non monetary indicators without any restriction and for insuring the generality aspect we can use other variables than those used by Klassen and all, and thus taking also into consideration the prolems related to the divergence of the information provided by households’ surveys. - The paper presents a direct comparison between Conditional and unconditional measure of pro-poor growth discussed in section(2) using the Composite Welfare Index (CWI) in page(9) and figure(2).- Given that the klassen and all’s method can only be applied using the growth Incidence Curve (GIC), the general comparison is essentially based on the use of this measurement technique. However, as the alternative method presented in this paper is supposed to allow introducing non-monetary indicators into pro-poor growth analysis by applying all measurement techniques presented in the literature, in addition to GIC, we use another measurement method which is the “Datt-Ravallion Decomposition” as an example to confirm its applicability without any restriction.5. Generally, if the growth rate g(p) is positive for all p, all the curve will be above the x-axes and indicates that growth is pro-poor absolutely. If in addition g(p) has a negative slope, the curve will be decreasing and indicates that the growth rate of the poorest is higher than the growth rate the richest ( i.e. inequality between households has fallen) . The Growth is thus Pro-poor relatively. Thus, the response to your example is : - If the growth rate g(p) is positive and has a negative slope for households with literate heads: growth for households with literate heads is pro-poor in the absolute and the relative terms. - If the growth rate g(p) is positive and has a flat slope for households with illiterate heads : growth is pro-poor for households with illiterate heads absolutely and not relatively.- If g(p) is higher among illiterate heads for all p compared to g(p) of literate heads: growth is globally pro-poor in absolute sense and relatively stronger for households with illiterate heads than the others. Thus, poverty rate has fallen more in the group of households with illiterate heads than the other group. -6. Yes, by comparing the slope and the position of the curve in different percentiles we can give response to this kind of questions and as indicated in the paper, specific information is useful for a more detailed analysis of pro-poor growth. It can be used for a better identification of any economic policy impact on poverty for each group of households.

Anonymous - Referee Report
January 14, 2013 - 10:25
The paper’s main contribution is to disaggregate (monetary) growth incidence curve analysis by various socio-economic categories according to household size, household head’s education, and household head’s gender. Counter to the author’s contention – and the title of the paper – this sheds no light on non-monetary pro-poor growth (i.e. changes in non-monetary dimensions of well-being). This type of disaggregation has probably already been conducted elsewhere, although I am not aware of any specific example. This analysis is a form of within-category poverty and inequality analysis. The contribution is thus very limited.A second major problem with the paper is the illustration. The choice of a period of negative economic growth (i.e. a period of contraction) to illustrate an analysis of pro-poor growth confuses the entire illustration, rather than elucidating the conceptual approach proposed. Many misinterpretations results from this in the empirical analysis.The composite welfare index raises a number of issues:• How were the component variables selected?• Were the same weights used in both years or did these vary?• These weights should be presented, as they determine the relative importance of each of the component variables in the index and help to understand its evolution.• In the same vein, some information should be provided on how the individual component variables evolved between 2004 and 2009.The paper requires a comprehensive linguistic revision.

Rami Ben Haj Kacem - Reply
January 23, 2013 - 18:23
Thank you for your comments.In the literature, as discussed in the paper, there have been several studies and attempts to measure and analyze pro-poor growth. However, despite the importance of the multidimensional aspect of poverty, little attention has been devoted to introducing non monetary indicators into pro-poor growth measurement. In this context, the contribution of this paper is to present an alternative method based on the fact that instead of ranking household by income and then constructing the GIC using a non monetary indicator (Klasen et al. (2008)), we classify the households by non monetary indicators and then we construct the GIC using income, which leads to different results and it is more faithful to the fundamental principle of pro-poor growth measurement.The choice of the Ethiopian case between 2004 and 2009 for the empirical illustration is first explained by the fact that Ethiopia is one of the poorest countries, so it is logically more interesting and useful to studying poverty in such case than in a reach country. Second, all the methods presented in the literature are supposed to be used wherever the case is .i.e. pro-poor or anti-poor growth situation. Third, the available data used in this paper are from the Ethiopia Rural Household Survey (ERHS) which consists in 8 waves (1989, 1994, late 1994, 1995, 1997, 1999, 2004 and 2009). However, as indicated in the paper in page (8), the consumption behavior of rural Ethiopian households varies considerably between seasons. For that, only the tow waves of 2004 and 2009 are used for the empirical illustration as they were constructed during the same period of the year. Concerning the composite welfare index (CWI), it is a classical method following the spirit of the Human Development Index to estimate household’s welfare when we do not dispose of monetary information (consumption or expenditure) in the survey, like in Demographic and health Surveys (DHS). The most indicators used to build the composite welfare index (CWI) in general are used in this paper (of course there is a little difference between one study to another depending on the availability of information and the agregation method). See for example: Harttgen . K, Klasen .S and Misselhorn. M (2010) `Pro-Poor Progress in Education in Developing Countries?`* Review of Economics and Institutions Vol. 1 – No. 1.Klasen el .al (2004) `Operationalizing Pro-Poor Growth Country Case Study: Bolivia` AFD, BMZ, DFID, and the World Bank Report.The CWI are calculated in the two years separately, so the weights are different. For having the same weight we should stacking the variables data of the tow dates and then calculating the CWI. This point is discussable.

Grigorios Zarotiadis - Invited Reader Comment
January 16, 2013 - 09:57
See attached file
[ Download attached file ]

Anonymous - Invited Reader Comment
January 22, 2013 - 09:50
See attached file
[ Download attached file ]

Rami Ben Haj Kacem - Reply
January 23, 2013 - 18:57
Thank you for your comments All definitions are correct and all used variables are difined in the paper. The diffrence between absolute and relative pro-poor growth are defined clearely in the paper. For more details, please see the reply to the first Invited Reader Comment.

rami bhk - Revised version
May 16, 2013 - 10:09
Please find enclosed the revised version of the paper.
[ Download attached file ]