ArticlePDF Available

Investigating Growing Inequality in Mozambique

April 2019
South African Journal of Economics 87(4)

License
CC BY-NC-SA 4.0

Authors:

University of Copenhagen

In this paper, we investigate the long‐term trend of consumption inequality in Mozambique. We show that an imbalanced growth path disproportionally benefited the better‐off and caused increasing inequality, especially in more recent years, curbing the necessary reduction in poverty. Using a regression decomposition technique, our results suggest that this trend was strongly associated with the higher attained education of household heads and with the changes in the structure of the economy (with less workers in the public and subsistence sectors). The trend was, however, mitigated by the tendency for the higher level of attained education and the smaller public sector to become associated with less inequality over time. These results point to the importance of accelerating the expansion of education and improving the productivity of the large subsistence sector to lower inequality in line with the sustainable development goals.

Daily real per capita consumption: densities [Colour figure can be viewed at wileyonlinelibrary.com] Note: Daily real per capita consumption values, expressed in constant terms (deflated by the contemporary poverty line). Source: Authors' calculations based on IAF/IOF.

…

Daily real per capita consumption: mean and quantiles [Colour figure can be viewed at wileyonlinelibrary.com] Note: Daily real per capita consumption values normalised by the contemporary poverty line. Source: Authors' calculations based on IAF/IOF.

…

Daily real per capita consumption: (relative) growth incidence curves [Colour figure can be viewed at wileyonlinelibrary.com] Note: Daily real per capita consumption values normalised by the contemporary poverty line. Source: Authors' calculations based on IAF/IOF.

…

Lorenz curves of real per capita consumption [Colour figure can be viewed at wileyonlinelibrary.com] Source: Authors' calculations based on IAF/IOF.

…

Gini index of consumption inequality, alternative consumption estimates [Colour figure can be viewed at wileyonlinelibrary.com] Note: Real consumption is nominal consumption adjusted using intra-survey temporary and spatial price indices. The square root of the household size was used to estimate the corresponding equivalised values. Source: Authors' calculations based on IAF/IOF.

…

Figures - available via license: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International

Content may be subject to copyright.

Available via license: CC BY-NC-SA 4.0

Content may be subject to copyright.

South African Journal of Economics Vol. 0:0 Month 2019

doi: 10.1111/s aj e.12215

INVESTIGATING GROWING INEQUALITY IN

MOZAMBIQUE

CAR LOS GRADÍN*,† A ND FINN TARP†

Abstract

In this paper, we investigate the long-term trend of consumption inequality in Mozambique. We

show that an imbalanced growth path disproportionally benefited the better-off and caused

increasing inequality, especially in more recent years, curbing the necessary reduction in poverty.

Using a regression decomposition technique, our results suggest that this trend was strongly

associated with the higher attained education of household heads and with the changes in the

structure of the economy (with less workers in the public and subsistence sectors). The trend was,

however, mitigated by the tendency for the higher level of attained education and the smaller

public sector to become associated with less inequality over time. These results point to the

importance of accelerating the expansion of education and improving the productivity of the large

subsistence sector to lower inequality in line with the sustainable development goals.

JEL Classification: D63, I24, O55

Keywords: Inequality, Mozambique, decomposition, RIF

1. INTRODUCTION

Mozambique was the poorest country in the world in 1992, when the war that followed

from the early 1980s after independence from Portugal in 1975 came to an end. Per

capita GDP was US$354 (2011 PPP) and poverty was widespread. Economic reforms

were initiated in 1986, and recovery followed with rapid economic growth from the mid-

1990s onwards, reaching a GDP per capita of US$1,128 in 2016. Yet the country still

ranks among the poorest in the world.1

Economic growth brought a substantial reduction in poverty levels. This is so whether

poverty is measured with monetary or non-monetary indicators, as reported, among

others, by the last two national poverty assessments (MPD/DNEAP, 2010; MEF/DEEF,

2016). Poverty, however, continues to be high, in part, because of the persistence of eco-

nomic constraints (i.e. inadequate education, trade and transport systems). They slow

1 GDP data from the World Bank, International Comparison Program database (http://data.

worldbank.org), accessed on 11/08/2017.

* Corresponding author: United Nations University World Institute for Development

Economics Research (UNU-WIDER), Katajanokanlaituri 6 B, FI-00160 Helsinki, Finland.

E-mail: Gradin@wider.unu.edu

† United Nations University World Institute for Development Economics Research

(UNU-WIDER), Helsinki, Finland

This is an open access article under the terms of t he Creative Commons Attribution-NonCommercial-ShareA like License, which

permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and the content

is offered under identical terms.

Economic Society of South Africa.

South African Journal

of Economics

2South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

down further poverty reduction compared with countries, like Vietnam, which have

achieved a more pro-poor growth pattern (Arndt et al., 2012).

The sustainable development goals have made the reduction of poverty in the develop-

ing world a priority, but also the reduction of inequalities to guarantee that no one is left

behind (Goal 10). Sub-Saharan Africa is among the most unequal regions in the world

and a puzzle for traditional development economics models and the popular Kuznets’

inverted-U hypothesis because it is also the least developed. This, however, implies that

there is a risk that inequality increases even more during the initial stages of develop-

ment of the non-subsistence sector in a region with predominantly resource-led growth.

Higher levels of inequality could harm the stability of an extremely fragile region and

undermine the effectiveness of poverty reduction strategies.

The scarce evidence for relative inequality in this region points at no clear pattern in

the last decades. Inequality does not seem to have changed much on average (see review in

Alvaredo and Gasparini, 2015), but is associated with large heterogeneity in levels, trends

and explanatory factors depending on the initial conditions and how inclusive economic

growth was, as recently pointed out in a UNDP report (Odusola et al., 2017). This report

has also summarised the main driving forces of inequality in the region: (i) a highly dual-

istic economy structure, with a large informal or subsistence economy cohabiting with a

small elite working in the formal economy (i.e. public, international and resource sectors);

(ii) the high concentration of land and physical and human capital in certain groups

and regions; and (iii) a limited distributive capacity of the state, leading to the “natural

resource curse,” the urban bias of public policy, and ethnic and gender inequalities.

The last two national poverty assessments in Mozambique documented an increase in

consumption inequality in this country. While Mozambique’s overall initial level of

inequality was high according to world standards, this was not so given the African con-

text, except for the urban areas (Fox et al., 2005). Inequality slightly increased between

the first two post-war households budget surveys (1996/97 and 2002/03) to later remain

barely constant (between 2002/03 and 2008/09).2 However, a much larger increase in

inequality was recently found between the last two surveys (2008/09 and 2014/15).

National poverty assessments emphasised evidence of underreporting in food consump-

tion by the poor that might imply that the actual level of inequality is lower than reported.

But inequality could also be significantly higher if we take account of underreporting

among the relatively better-off (Arndt and Mahrt, 2017) or how the expenditure structure

differs for these households as compared with the poor (Arndt et al., 2015).

This paper contributes to the growing literature on inequality in Mozambique and,

by extension, in sub-Saharan Africa. We analyse the long-term trend in inequality in

Mozambique and characterise its distributional pattern. We also identify some of the

underlying drivers using a regression-based decomposition technique based on the

Recentered Influence Function of the Gini index. In line with Gradín (2016), we first

investigate the role of several household characteristics on inequality in every year. Then,

we construct a counterfactual distribution in which we combine the average characteristics

2 Elbers et al. (2005) showed that in Mozambique, like in Ecuador and Madagascar, most in-

equality in 1996/97 was within the small administrative units. The change in inequality between

1996/97 and 2002/03 has been analysed, for example, by James et al. (2005) and by Fox et al.

(2005), using decompositions by different population subgroups.

3South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

of the initial year while keeping the impact of these on inequality that prevailed in the final

year. Using this counterfactual, we decompose the overall change in inequality over time

into two terms. One is the change in inequality that can be attributed to a change in the

composition of the population by characteristics (characteristics or explained effect). The

other is the change in inequality that can be attributed to the change in the relationship

between these characteristics and inequality (coefficients or unexplained effect). The results

point in the direction that there was a robust increase in inequality, driven by an unbal-

anced growth pattern that has disproportionally benefited the better-off. This growth pat-

tern is characterised by the emergence of a non- subsistence economy in Maputo and other

urban areas, in a resource-based country, with a shrinking public sector, the expansion of

education and the emergence of a small but highly educated elite. The results show that the

increase in inequality can be mostly accounted for by this compositional effect rather than

by a structural change in terms of how household characteristics affect inequality.

In what follows, the next section describes the data and main variables used while

Section 3 discusses the latest trends in inequality. Section 4 introduces the decomposition

methodology while Section 5 presents the empirical results. The last section concludes.

2. DATA AND VARIABLES USED IN THE ANALYSIS

The empirical analysis is based on the four nationally representative households budget

surveys collected by the Instituto Nacional de Estatística (INE) after the end of the post-in-

dependence war: the Inquéritos aos Agregados Familiares (IAF) for 1996/97 and 2002/03,

and the Inquéritos ao Orçamento Familiar (IOF) for 2008/09 and 2014/15.

We use daily real per capita consumption as our main indicator of individual well-

being, although nominal consumption will also be used for robustness analysis.

Consumption is usually preferred to income in inequality analyses in developing coun-

tries, especially in the sub-Saharan region. We use here the same indicator constructed

for the Fourth National Poverty Assessment MEF/DEEF (2016) based on the PLEASe

methodology (see Arndt et al., 2017a for details). In a first normalisation, current real

consumption is obtained by adjusting nominal consumption in each survey to correct for

seasonal and spatial variation in prices using price indices computed separately for 13

geographical regions. We proceed with a second normalisation to produce real consump-

tion in constant terms over time. We divide current real consumption by the contempo-

rary official poverty line (which is also expressed in contemporary real consumption

terms). This deflator allows us to describe the change in household purchasing power of

a (flexible) basket of basic food and non-food goods over time. Note that this second

normalisation does not affect relative inequality measures because the contemporary

poverty line is the same for all individuals within each survey.3

3 The Consumption Price Index in Mozambique is only estimated using prices in a few urban

areas. Poverty lines in contemporary currency (in parentheses 2011 PPP using the World Bank GDP

deflator) are MZM 5,502 (US$1.13) in 1996/97, MZM 8,307 (US$0.95) in 2002/03), MZN 17.93

(US$1.37) in 2008/09 and MZN 29.19 (US$1.76) in 2014/15. The use of a common deflator

(regardless of the level of consumption) is a rather conservative approach. Arndt et al. (2015) have

recently shown that the increase in inequality between 2002/03 and 2008/09 was 0.030 Gini points

higher when using a composite household-specific price index that is sensitive to the differences in

the structure of consumption of the relatively well-off and the poorer households.

4South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

IAF/IOF surveys have been the main source for the analysis of well-being in

Mozambique. They have, however, suffered from a variety of limitations in the collec-

tion of data, shared with other large developing countries and aggravated by the lasting

consequences of the conflict, such as the lack of infrastructure, the presence of land

mines, market fragmentation, flooding in certain areas and the use of non-standard unit

measures. (Arndt et al., 2017b). The surveys are also associated with the well-known

problem of underreporting of food consumption that has been aggravated in the most re-

cent one. Initially confined to urban areas in the South, it is now affecting rural areas too

(MEF/DNEAP, 2016). This underreporting is in part related to infrequent purchases,

especially of rice and corn flour. Like most household surveys, IAF/IOF also suffer from

underestimation of top values due to the underrepresentation and/or underreporting of

consumption by relatively well-off households (Arndt and Mahrt, 2017).

In the first three surveys, we have information about consumption for a total of,

respectively, 8,250, 8,700 and 10,832 households interviewed once over four quarters.

They account for a total of 42,667, 44,083 and 51,177 individuals, respectively. The

design of the most recent survey is different. We have information for around 11,000

households that were interviewed one, two or three times between August 2014 and July

2015 (with a total of 11,505 households interviewed in the first quarter, 10,372 in the

second one and 11,315 in the fourth quarter). In this last case, we use the pool of house-

holds in the analysis (58,342, 50,770 and 55,198 individuals, respectively) to prevent a

seasonal bias (which is not present in the other surveys because the sample of households

was distributed in different quarters). The unit of analysis is always the individual, while

the income sharing unit at which individual well-being is determined is the household.

For that reason, individuals are attributed the per capita consumption and the charac-

teristics of their households. Standard errors are clustered within households and all

estimations used sampling weights.

To explain changes in inequality over time, we consider several characteristics of

households available in these surveys that may have influenced their consumption levels.

We account for economic opportunities varying by location using information about the

area (rural or urban)4 and province of residence (Niassa, Cabo Delgado, Nampula,

Zambezia, Tete, Manica, Sofala, Inhambane, Gaza, Maputo and Maputo City).

Demographic variables considered include the number of children (aged 14 or less) and

adults in the household, as well as the marital status (single, married, widowed, separated

or divorced), age (in brackets) and sex of the household head.5 We also considered the

education attained by the household head. Several variables accounted for the employ-

ment status of the household head. First, we used the head’s industry, distinguishing

whether the head is (i) employed in the non-subsistence sector, operationalised here as

remunerated work (not being a family helper) outside the primary sector; (ii) working in

4 The definition of urban area in 1996/97 is narrower than in the subsequent surveys. Consistent

definitions of area were used for comparing 1996/97 with 2002/03, and the latter with subse-

quent surveys. Maputo City is entirely urban and Maputo province is mostly urban (70%), while

the majority of the population is rural in the other provinces, with the urban population ranging

between 14% in Tete and 36% in Sofala (2014/15).

5 In the regressions, we will not use a few observations with unknown age and sex of the house-

hold head.

5South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

the subsistence sector; and (iii) not employed.6 We also considered two dummy variables

indicating whether the head is self-employed or works in the public sector, respectively.

Finally, we included the employment rate of household adults (the number of employed

adults in the household divided by the number of adults, taking the value zero in few

cases in which there were no adults at all).

3. TRENDS IN INEQUALITY

The densities of real per capita consumption for the different years are displayed in

Fig. 1. Mean consumption values and several quantiles are reported in Table A1 in the

Appendix. Real per capita consumption increased by two thirds over the entire

1996/1997–2014/2015 period, corresponding to an accumulated average annual growth

rate of 2.9%. Yet this rate was not homogeneous in different subperiods. The average

annual growth rate was 5.2% between 1996/1997 and 2002/2003. Hereafter,

6 We also included a dummy to control for cases in which the household head was employed

while the industry was unknown (less than 1% in any survey). Compared with other options, like

not using the variable or removing the affected observations, this allows us to minimize the loss

of information (preserving all the information that is not missing in those observations).

Figure 1. Daily real per capita consumption: densities [Colour figure can be viewed at

wileyonlinelibrary.com]

Note: Daily real per capita consumption values, expressed in constant terms (deflated by

the contemporary poverty line).

Source: Authors’ calculations based on IAF/IOF.

6South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

consumption remained almost constant until 2008/2009 (0.2%) to increase again at an

annual average rate of 3.3% until 2014/2015.7

Increases in real per capita consumption took place across the entire distribution. The

densities shifted to the right, consistent with the well-established reduction in the inci-

dence and intensity of poverty. The official poverty rate declined from 70% in 1996/1997

to 46% in 2014/2015: a sharp reduction until 2002/2003 (reaching 53%), a more modest

decline (from 52 to 46%) between 2008/2009 and 2014/2015 (see MEF/DNEAP, 2016

for a more detailed analysis of poverty). This was accompanied by an initial decline in

the intensity of poverty too. The median poverty gap, according to our own calculations,

declined from 43 to 35% of the poverty line in the first period, remaining at this level

thereafter.

The increase in real per capita consumption exhibited, however, a clear pattern dis-

proportionally benefitting the relatively well-off as shown in Figs. 2 (selected percentiles

in absolute terms) and 3 (relative growth incidence curves). This is in line with earlier

results (e.g. James et al., 2005 for 1996/1997–2002/2003), but was accentuated during

the last period. Growth in consumption was largest for the highest percentiles in both

absolute and relative terms. While real consumption grew by 73% for the 95th percentile

over the entire period, the median grew by 47%, and the 5th percentile only grew by

33%. There are different patterns across periods, though, with the highest growth for the

7 The closest per capita expenditure aggregate in National Accounts (i.e. households and non-

profit institutions serving households, constant 2010 US$) also shows a large increase over the

entire period, 79%, but with no stagnation in the middle one; the average annual rates were, re-

spectively, 4.9, 2.6 and 2.3% (http://data.worldbank.org, accessed on 06/09/2018).

Figure 2. Daily real per capita consumption: mean and quantiles [Colour figure can be

viewed at wileyonlinelibrary.com]

Note: Daily real per capita consumption values normalised by the contemporary poverty line.

Source: Authors’ calculations based on IAF/IOF.

7South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

well-off occurring during the last subperiod. This inequality increase is also illustrated

by the increase in the p50/p10 and p90/p50 ratios (respectively from 2.2 to 2.4 and from

2.4 to 2.6), with the bulk of the increase in the last period. This unbalanced growth is

compatible with significant improvements at the very bottom in the first and last peri-

ods that contrast with the decline in their consumption during the intermediate period,

characterised by stagnation on average (Fig. 3).

The non-overlapping empirical Lorenz curves shown in Fig. 4, not surprisingly, reveal

an unambiguous increase in inequality in both periods of consumption growth

(1996/1997–2002/2003 and, especially, 2008/2009–14/15). Lorenz dominance was sta-

tistically significant in both periods (Table A9).8 Less obvious is the trend in the inter-

mediate period of stagnation in consumption (2002/2003–08/09). While the 2008/2009

curve crosses the 2002/2003 one from below around the 65th percentile, the cross is not

statistically significant (while the difference between both curves is statistically signifi-

cant only below the 13th percentile). This implies also an increase in inequality in this

period according to the Lorenz criterion. For the entire period, the most recent Lorenz

curve falls entirely below the earliest one, with the differential being statistically signifi-

cant at all percentiles.

The increase in inequality is corroborated using a variety of inequality indices, like

Gini, and the Generalised Entropy and Atkinson families, all of them consistent with

Lorenz dominance.

8 The 2002/03 curve is below the 1996/97 curve everywhere, but only with high statistical sig-

nificance (at 90 or 95%) in the upper tail. The 2014/15 curve falls below that of 2008/09 for all

percentiles, being statistically significant (95%) above the 7th percentile. See Table A9.

Figure 3. Daily real per capita consumption: (relative) growth incidence curves [Colour

figure can be viewed at wileyonlinelibrary.com]

Note: Daily real per capita consumption values normalised by the contemporary poverty line.

Source: Authors’ calculations based on IAF/IOF.

8South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

The Gini index of real per capita consumption increased by 4.6% in the 1996/1997–

2002/2003 period, and by 12.7% in the 2008/2009–14/15 period, remaining constant

in between (−0.1% between 2002/2003 and 2008/2009), as displayed in Fig. 5 and Table

A1. The increase in inequality, as measured by the Gini index, over the entire period was

17. 8 %.9

Figure 5 shows the trend in the Gini index using four alternative well-being indica-

tors for the sake of robustness, with real and nominal consumption (no adjustment for

spatial and temporary variation in prices), and in each case with per capita and per adult

equivalent (using the square root of household members) – see Tables A1–A4 in the

Appendix. The Gini index is substantially higher when consumption is nominal instead

of real. This reflects that in Mozambique, as in other developing countries, geographical

differences in prices are substantial. It is slightly lower, however, when it is equivalised

instead of per capita. The Gini index for Mozambique follows the well-known U-pattern

with the level of economies of scale associated with cohabiting in households, with not

much difference found between the per capita case (no economies of scale) and the most

common equivalent scale found in the literature (of mostly middle- and high-income

countries).

The global trend, however, is similar although with different intensities, indicating

that the increase in inequality is robust to alternative methodological choices. The total

increase in inequality ranges between 16% and 20%, corresponding to 5%–9% during

9 Changes in Gini of real per capita consumption inequality over time are all statistically signif-

icant using bootstrap standard errors (bias-corrected 95%confidence intervals), except between

2002/03 and 2008/09 (see Table A1).

Figure 4. Lorenz curves of real per capita consumption [Colour figure can be viewed at

wileyonlinelibrary.com]

Source: Authors’ calculations based on IAF/IOF.

9South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

the first period, and 11%–17% in the last period. Inequality declines in the intermedi-

ate period between 0% and 5%, depending on the case.10

Arndt and Mahrt (2017) have recently shown that re-scaling consumption by discrep-

ancies with National Accounts would increase the level of inequality as measured by the

Gini index. It would also affect the trend (with a higher increase between 2002/2003 and

2008/2009, and a lower increase in the last period). In the opposite direction, a correc-

tion for underreporting in food consumption in 2014/2015 (based on meals reported by

households that are not reflected in their consumption) would reduce the level of in-

equality in that year. Although one could expect this reduction to be larger than in pre-

vious years, the effect on the trend cannot be computed due to the lack of necessary

information.11 Due to these opposite effects and the impossibility of controlling for all of

them in all surveys, we do not make any adjustment for underreporting, but the results

should be interpreted in the context of these data limitations.

The Generalised Entropy (Fig. 6) family of measures also shows a unanimous increase

in inequality in real per capita consumption during the initial and, especially, final

10 The respective changes in the Gini index for each period are 8.6%, −2.1% and 10.6% for real

equivalised consumption, and 5.1%, −2.3% and 16.7% for nominal per capita consumption.

11 According to our own calculations, the Gini in 2014/15 would be smaller by between 0.004

and 0.020, depending on the assumptions made (for the methodology used in adjusting con-

sumption for under-reporting, see MEF/DEEF, 2016).

Figure 5. Gini index of consumption inequality, alternative consumption estimates [Colour

figure can be viewed at wileyonlinelibrary.com]

Note: Real consumption is nominal consumption adjusted using intra-survey temporary

and spatial price indices. The square root of the household size was used to estimate the

corresponding equivalised values.

Source: Authors’ calculations based on IAF/IOF.

10 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

periods (with similar results for the Atkinson family of indices shown in the Appendix).

These indices also help better characterise the distributional pattern of the inequality

increase. They confirm that these increases in inequality tend to be larger with higher

sensitivity of the indices to top consumption values (higher α). In this line of reasoning,

the increase in inequality using the GE(2) is remarkable. In the last period, there was also

an important increase in inequality with higher sensitivity to the bottom of the distribu-

tion, likely the result of the underreporting of food consumption mentioned above. In

the intermediate 2002/2003–2008/2009 period, on the contrary, there is an increase or

a decline in inequality depending on whether we put more weight on inequality among

the poor or the better-off, respectively (inequality declines for α ≥ 1, increases otherwise),

which does not come as a surprise due to the crossing empirical Lorenz curves.12

4. METHODOLOGY: DECOMPOSING INEQUALITY CHANGES OVER TIME

Let y = (y1,…,yN) indicate a vector of consumption across a population of N individuals.

We will measure inequality using the Gini index, I(y). We assume that the contribution

of the ith individual to the overall inequality I(y) is given by the recentered influence

function of I, estimated for consumption yi, RIF (yi; I) (Firpo et al., 2007, 2009). The RIF

measures the impact on the target statistic of marginally increasing the population with

12 Changes in GE (and Atkinson) indices of real per capita consumption inequality over time

are all statistically significant at 95% using bootstrap standard errors (bias-corrected confidence

intervals), except between 2002/03 and 2008/09, in which case only GE(−2) and GE(−1) (and

A(2)) show statistically significant increases (see Table A1).

Figure 6. General Entropy (GE) indices of real per capita consumption inequality [Colour

figure can be viewed at wileyonlinelibrary.com]

Source: Authors’ calculations based on IAF/IOF.

11South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

consumption yi, with the overall inequality being just the average impact, I(y) = E(RIF

(y; I)).13 The RIF function is a U-shaped transformation of yi, reflecting that a marginal

increase in the population with extremely low and high values would have a dispropor-

tional increase in inequality. A marginal increase in the population around the mean

consumption, however, will tend to have a small or no effect on inequality. The exact

values of the RIF function are index specific, in particular, reflecting the different sensi-

tivities of the indices to the different parts of the distribution. That is, some indices will

increase at a higher rate after a marginal increase in the population at the top or at the

bottom (see the analysis of the RIF of different inequality indices in Gradín, 2016, 2018).

The relationship between these individual contributions to inequality and household

characteristics is given by a N×K matrix X (including a constant) that can be estimated

by OLS14:

We can interpret βk, k ≥ 2, as the expected effect on inequality of a marginal change

in the average value of the kth characteristic,

̄xk

, ceteris paribus while β1 reflects the ex-

pected value of inequality when xk = 0, k ≥ 2. In the case of dummy variables, this means

that βk measures the marginal impact of a slight increase in the proportion of individuals

with xk = 1, and β1 the expected value of inequality for the reference household (defined

by the omitted categories).

We can thus rewrite I(y) as the sum of the impact of household characteristics on

inequality:

The counterfactual inequality index

(

)

=̄

�

𝛽

indicates the expected value of in-

equality in the final year if the characteristics remained constant over time (superscripts

0 and 1 refer to the initial and final years, respectively). With this counterfactual, we can

decompose the change in inequality over time into two distinct contributions:

The coefficients effect, ̄

X0�

(𝛽

1−

𝛽

)

, is the structural change and it indicates the ex-

pected change in inequality if average characteristics had remained constant over time

13 Let us consider a mixture distribution

y𝜺

that assigns a probability

𝜀

of having a mass

and

1−𝜀

of being the original distribution. The influence function

IF (

yi;I(y)

)

𝜕

𝜕𝜀

I(y𝜺)

𝜀=

the directional derivative of

for this mixture distribution when

𝜀→0

and has zero expectation

(Hampel, 1974). By just adding the value of the target statistic, we obtain the

RIF (

;I(y)

)

(

)

+IF

(

;I(y)

)

, where the expected value is the target statistic.

14 The

for the Gini index was first documented in Monti (1991).

(1)

RIF

(yi;I)=

∑

k=1

̄xk𝛽k+𝜀i

(2)

(y)=

∑

RIF (yi;I)=̄

X�𝛽=

∑

k=1

̄xk𝛽k

(3)

)−

)=̄

�

𝛽1−̄

�

𝛽0=(̄

�

−̄

�

)𝛽1+̄

�

(𝛽1−𝛽0

12 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

and only the coefficients (i.e. their impact on inequality) changed. The characteristics

effect,

(

�

−̄

�)

𝛽

, is the compositional effect and indicates the expected change in

inequality induced by changes in the average characteristics, evaluated with the coeffi-

cients estimated in final year,

𝛽1

Therefore, the evaluation of the individual contribution of each variable

to the

characteristics and coefficients effects can be measured as

WΔX

(

̄x

−̄x

𝛽

and

WΔ𝛽

=̄x0

𝛽1

−𝛽0

, so that the individual effects sum up the corresponding aggregate ef-

fects. Similarly, the sums of the characteristics and the coefficients effects of each charac-

teristic add up to the total contribution of that same characteristic,

WΔX𝛽

=WΔX

Δ𝛽

Note that when the target statistic is the average of

, this procedure would lead to the

one proposed by Blinder (1973) and Oaxaca (1973).

The detailed coefficients effects, however, suffer from a well-known identification

problem (Oaxaca and Ransom, 1999). The contribution of each dummy variable depends

on which is the omitted category, and the contribution of a continuous variable depends

on the chosen scale. Although there are some proposals in the literature suggesting a nor-

malisation of the coefficients of dummies to assure that the detailed effects do not vary

with a change in the omitted category, we do not make any correction because all these

adjustments are ad hoc (Fortin et al., 2011). Thus, the results should be interpreted for

a given specification of the model. Neither the detailed characteristics effects nor the ag-

gregate coefficients and characteristics effects are affected by this identification problem.

Alternatively, a different counterfactual can be derived,

(

)

=̄

1�

𝛽

, changing the

decomposition into characteristics and coefficients effects:

The difference between this decomposition and that in (3) is that now the coefficients

effect is valued with the final average characteristics

, while the characteristics effect

uses the initial coefficients

𝛽0

. Thus, we can interpret the characteristics effect as the

expected change in inequality if only characteristics had changed over time (but not the

coefficients), and the coefficients effect as the change in inequality after changing those

effects in the final year (while keeping the final average characteristics).

5. EXPLAINING THE INEQUALITY TRENDS

We now investigate the trends of inequality in Mozambique in two steps. We first iden-

tify the extent to which household characteristics are associated with inequality. Those

attributes, ceteris paribus, with higher impact on consumption at the extremes of the

distribution, especially at the very top, will tend to be more strongly associated with

higher inequality. Then, we estimate their contribution to the increase in inequality over

time, either through a compositional effect (changes in the proportions) or through a

structural effect (change in their relationship with inequality, i.e. the coefficients).

5.1 Household Characteristics and Inequality

The RIF regressions, reported in Table 1, show that inequality in 2014/15 tends to be

strongly increasing with the proportion of people living in Maputo City, and to a lower

(4)

)−

)=̄

1�

𝛽1−̄

0�

𝛽0=̄

1�

(𝛽1−𝛽0)+(̄

1�

−̄

0�

)𝛽

13South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

extent in other urban areas. This points to the fact that inequality in Mozambique is

closely related with higher inequality in urban areas (higher than in other African coun-

tries, see Fox et al., 2005) and to the large urban-rural gap, even after controlling for

Table 1. RIF regressions, 1996–2014

1996 /97 20 02/03 2002/03 2008/09 2 014/15

Estimate SE Estimate SE Estimate SE Estimate SE Estimate SE

Area

Urban 1996 0.055* 0.022 −0.020 0.032 – – – – – –

Urban – – – – 0.053* 0.026 0.031 0.021 0.053** 0. 017

Province

Niassa −0.048 0.029 0.003 0.028 0.014 0.029 0.168*** 0.049 −0.091* * 0.032

Cabo Delgado −0.014 0.034 0.085 0.063 0.093 0.067 0.055* 0.026 −0.116*** 0.031

Nampula −0.009 0.033 − 0. 011 0.026 −0.012 0.026 0 .118** 0.038 −0.086* 0.034

Zambezia −0.063* 0.027 0.002 0.025 0.017 0.028 0.120*** 0.030 −0.090** 0. 031

Tet e −0 .012 0.027 0.036 0.024 0.047 0.026 0.058* 0.026 −0.129*** 0.031

Manica 0.026 0.037 0. 015 0.027 0.020 0.028 0.050* 0.024 −0.109*** 0.032

Sofala 0.060* 0.027 0.090* 0.037 0.092* 0.038 0.142*** 0.026 −0.034 0.034

Inhambane −0.008 0.027 0.112*** 0.024 0.127*** 0.026 0.111*** 0.025 −0.059 0.032

Gaza −0.026 0.031 0 .015 0.029 0.025 0.030 0.143*** 0.025 −0.032 0.031

Maputo City 0.005 0.042 0.108** 0.038 0.088* 0.037 0.184*** 0.038 0.413*** 0.066

Household size

N. adults 0.002 0.004 − 0.0 01 0.004 − 0.0 01 0.004 −0.0 01 0.006 −0.027*** 0.005

N. children −0.004 0.004 −0.009* 0.004 −0.009* 0.004 −0.009* 0.004 −0.008** 0.003

Age (head)

25–34 −0.029 0.027 0.033 0.032 0.034 0.033 0.046 0.024 −0.02 0.017

35– 44 0.000 0.028 0.04 0.033 0.04 0.033 0.064* 0.025 0.035 0.018

45–54 −0.004 0.028 0.058 0.031 0.057 0.031 0.089*** 0.024 0.065** 0.024

55 or older 0.009 0.027 0.096* * 0.031 0.096** 0.031 0.081*** 0.023 0.071* * 0.024

Sex (head)

Female 0.012 0.025 0 .010 0. 019 0.006 0.019 0.016 0.020 − 0. 015 0.025

Marit al sta tus

(head)

Single 0.082* 0.037 0.14 4* 0.059 0.140* 0.059 0.215 0.118 0.097* 0.044

Divorced 0.002 0.026 0 .031 0.021 0.025 0.021 0.0 03 0.020 −0.021 0.028

Education

(head):

Atta ined

education

Some/lower

primary

0.014 0.016 0.021 0.018 0 .013 0.018 −0.018 0.010 −0.030*** 0.009

Upper primary 0.126*** 0.036 0.139* 0.064 0.12 2* 0.058 0.025 0.024 −0. 001 0.013

Lower secondary 0.454*** 0.100 0.342*** 0.079 0.333*** 0.078 0.168* 0.078 0.114*** 0.025

Upper secondar y 0.478* 0. 241 0.847*** 0.165 0.826*** 0.163 0.456*** 0.083 0.455*** 0.069

Tec h ni c al 0.506*** 0.137 0.641*** 0.148 0.628*** 0.147 0.412*** 0.111 0.558*** 0.095

Higher 2.071*** 0.492 3.061** 0.972 3.063** 0.969 1.738*** 0.325 1.690*** 0.150

Literate −0.013 0.012 0.021 0.012 0.019 0.012 −0.002 0.010 − 0. 019* * 0.007

Employment:

Typ e (hea d)

Public sector −0.0 49 0.033 −0.143* 0.070 − 0.136* 0.069 −0.154*** 0.047 −0.409*** 0.061

Self-employed 0 .017 0.025 0.007 0.043 0.016 0.043 0.005 0.035 0.017 0.014

Sector+ (head)

Subsistence −0.006 0 .031 − 0.057 0.051 −0.058 0.0 51 −0.063 0.053 −0.066* 0.028

Non-subsistence 0.037 0.036 0.047 0.038 0.027 0.038 0.021 0.043 − 0.059* 0.028

Employment

rate

(household)

−0.027 0.028 0.118* * 0.042 0.141** 0.048 0.102 0.053 0.050 0.027

Intercept 0.404*** 0.042 0.230*** 0.052 0.188 ** 0.062 0.186*** 0.056 0.572*** 0.041

N Observations 42 ,143 44,083 44,083 51,175 16 4,359

R25.0 8.0 8.0 10.7 7. 6

F7.25*** 9.08*** 9.26*** 7.49*** 14.01***

Note: p-values: * < 0.05; ** < 0.01; *** < 0.001. +A category was included to indicate that the head’s

industry was missing.

Source: Authors’ calculations based on IAF/IOF.

14 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

other characteristics like education or predominance of the subsistence sector. While

inequality in rural areas is 0.373, it is 0.552 in urban areas, with the highest level of 0.582

in Maputo City, in contrast with the lowest level of 0.312 in rural Tete. There is also a

large gap between the average real consumption of urban and rural areas: 1.3 times the

poverty line in rural areas versus 2.4 in urban areas, with the highest level in Maputo

(4.5) and the lowest in rural Niassa (1.0).

Inequality in 2014/15 tends to decline with the average number of household mem-

bers (especially, adults), and to increase with the proportion of people living in house-

holds in which the head is older or single and, especially, has attained upper secondary or

higher education. The education effect is again the result of higher between-group gaps,

with average real consumption 7.6 times the poverty line when the household head has

higher education, compared with only 1.2 when they have less than primary education.

It is also the result of higher education having more within-group inequality. The Gini

index is 0.566 among those with a head with higher education, and 0.379 when the head

has only achieved lower primary education.

Regarding the labour market status, inequality is reduced with the employment of

the household heads in the subsistence sector and, especially, in the public sector, and

increases with the employment rate of household members. There is a major difference

in the average consumption by sector (2.5 times the poverty line if the head works in the

non-subsistence sector; 1.2 in the subsistence sector), and in within-group inequality as

well (Gini is 0.512 versus 0.361, respectively). These results do however seem to vanish

after controlling for other characteristics like education or area of residence.

Some of the above effects have intensified over time, especially the dis-equalising ef-

fect associated with the proportion of people living in Maputo City, and the equalising

effect of the head working in the public sector. The dis-equalising effect of having some

college education in increasing inequality declined after 2002/03.

At the same time, there was a change in the composition of households by characteris-

tics over time (Table 2). There was a redistribution of population by provinces (with pop-

ulation increases in Tete and Manica, and declines in Sofala, Inhambane, Gaza or Maputo

City). This was accompanied by a decline in the average number of adults, an increase

in the average number of children and in the proportion of people living in households

with middle-aged, divorced or female heads. There was also a general increase in attained

education of household heads, along a reduction of self-employed and workers in the

public sector (from 11 to 6%) or in the subsistence sector (from 70 to 62%), all reflecting

a timid higher development and diversification of the modern sector of the economy after

the end of the war and the economic growth that followed (e.g. Jones and Tarp, 2016),

along the expansion of the educational system (e.g. van der Berg et al., 2017).

5.2 Decomposing Changes in Inequality

In the previous section, we established the trend towards higher inequality that was the

combined effect of changes in the characteristics and the marginal impact they had on

15South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

inequality. Table 3 reports the decomposition of the change in the overall inequality over

time into the characteristics and coefficients effects following the approach previously

described. It turns out that the increase in the Gini index between 1996/1997 and

Table 2. Distribution of characteristics, 1996/97–2014/15. Proportion of the population,

except for household size

Area 199 6/9 7 2002/03 20 08/09 2014/15

Urban (1996−2002) 0.210 0 .198 – –

Urban (2002−2014) –0.321 0.304 0.317

Province

Niassa 0.050 0.051 0.059 0.064

Cabo Delgado 0.077 0.084 0.078 0.0 74

Nampula 0.188 0.18 8 0.192 0.195

Zambezia 0.193 0.19 2 0.190 0.188

Tet e 0.068 0.077 0.090 0.098

Manica 0.057 0.067 0.070 0.075

Sofala 0.101 0.084 0.0 81 0.079

Inhambane 0.073 0. 074 0.061 0.058

Gaza 0.070 0.070 0.063 0.055

Maputo province 0.057 0.056 0.063 0.066

Maputo cit y 0.066 0.057 0.052 0.049

Household size

N. adults 3.150 3.122 2.844 2.994

N. children 3.050 3.094 3.112 3.224

Age (head)

Less than 24 0.068 0.061 0.064 0.060

25−34 0.234 0.255 0.265 0.224

35−44 0.264 0.269 0.264 0.289

45−54 0.216 0.198 0. 212 0. 211

55 or older 0.218 0 .217 0 .196 0.217

Sex (head)

Female 0.174 0.205 0.242 0.242

Marit al sta tus (head)

Married, union (or unknown) 0.850 0.820 0.818 0.808

Single 0.028 0 .015 0.015 0.032

Divorced, separated, widow(er) 0.122 0 .166 0.166 0.160

Education (hea d):

Attai ned education

None/unknown 0.690 0.697 0.255 0. 315

Some/lower primary 0.242 0.17 7 0.552 0.439

Upper primary 0.051 0.069 0.126 0.139

Lower secondary 0.009 0.031 0.031 0.041

Upper secondar y 0.002 0.015 0.016 0.033

Tec h ni c al 0.005 0.008 0.008 0.007

Higher 0. 001 0.003 0.012 0.025

Literate 0.522 0.5 44 0.553 0.568

Employment:

Typ e (hea d)

Public sector 0.112 0.077 0.063 0.058

Self-employed 0 .719 0.797 0.784 0.685

Sector (head)

Non-employed 0.073 0.043 0.038 0.010

Subsistence 0.700 0.688 0.714 0 .617

Non−subsistence 0. 219 0.268 0.248 0.283

Missing sector 0.008 0.001 0.001 0.000

Employment rate (hou sehold) 0.805 0.834 0. 881 0.782

Source: Authors’ calculations based on IAF/IOF.

16 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table 3. Decomposition of the increase in Gini inequality, 1996/97–2014/15

Final Gini 19 96 /9 7−2 014/ 15 2002/03−20014/15 199 6 /9 7−2 002 /0 3 2002/03−08/09 200 8/0 9−14 /15

0.468*** (0.006) 0.468*** (0.006) 0.415*** (0.008) 0.415*** (0.007) 0.468*** (0.006)

Initial Gini 0.397*** (0.006) 0.415*** (0.008) 0.397*** (0.006) 0.415*** (0.008) 0.415*** (0.007)

Change in Gini 0.071*** (0.008) 0.052*** (0.010) 0.018 (0.010) 0.000 (0.010) 0.053*** (0.009)

Cha r. E Coef. E C har. E Coef. E Char. E Coef. E Cha r. E Coef. E Char. E Coef. E

Tot al Ef fect 0.070*** (0.010) 0.0 01 (0.010) 0.045*** (0.008) 0.007 (0.010) 0.037*** (0.008) −0.019* (0.008) 0.015* (0.007) −0.015 (0.011) 0.031*** (0.005) 0.021** (0.008)

Area 0.006** (0.002) −0.001 (0.006) 0.000 (0.000) 0.000 (0.010) 0.000 (0.000) −0.016 (0.008) −0.001 (0.000) −0.007 (0.011) 0.001 (0.000) 0.006 (0.008)

Province −0.012*** (0.002) −0.032 (0.038) −0.006** (0.002) −0.089* (0.037) −0.001 (0.001) 0.047 (0.032) −0.002 (0.001) 0.068* (0.032) −0.002 (0.001) −0.160*** (0.036)

Household size

N adults 0.004** (0.001) −0.092*** (0.020) 0.003** (0.001) −0.081*** (0.020) 0.000 (0.000) −0.008 (0.018) 0.000 (0.002) −0.001 (0.023) −0.004*** (0.001) −0.075** (0.022)

N children −0.001* (0.001) −0.010 (0. 014) −0.001 (0.001) 0.003 (0.016) 0.000 (0.001) −0.015 (0.017) 0.000 (0.001) −0.002 (0.019) −0.001 (0.000) 0.005 (0.016)

Age (head) 0.0 01 (0.001) 0.040 (0.029) 0.002* (0.001) −0.019 (0.032) 0.000 (0.001) 0.057 (0.036) 0.000 (0.001) 0.013 (0.034) 0.003** (0.001) −0.032 (0.026)

Sex (head) −0.001 (0.002) −0.005 (0.006) − 0.001 (0 .001) −0.004 (0.006) 0.000 (0.001) 0.000 (0.005) 0.001 (0.001) 0.002 (0.006) 0.000 (0.000) −0.008 (0.008)

Marital status (head)

Single 0.000 (0.000) 0.000 (0.002) 0.002* (0.001) −0.001 (0.001) −0.002* (0.001) 0.002 (0.002) 0.000 (0.000) 0.001 (0.002) 0.002* (0.001) −0.002 (0.002)

Divorced −0.0 01 (0.001) −0.003 (0.005) 0.000 (0.000) −0.008 (0.006) 0.0 01 (0.001) 0.004 (0.004) 0.000 (0.000) −0.004 (0.005) 0.000 (0.000) −0.004 (0.006)

Education+ (head) 0.052*** (0.007) −0.024** (0.007) 0.038*** (0.006) −0.054*** (0.011) 0.027*** (0.005) 0.022* (0.011) 0.011 (0.006) −0.040** (0.012) 0.034*** (0.004) −0.021 (0.011)

Employment:

Type (head)

Public s. (head) 0.022*** (0.004) −0.040*** (0.008) 0.008** (0.002) −0.021** (0.007) 0.005 (0.003) −0.010 (0.009) 0.002 (0.001) −0.001 (0.006) 0.002 (0.002) −0.016** (0.005)

Self-employed (head) −0.001 (0.000) −0.001 (0.021) −0.002 (0.002) 0.000 (0.036) 0.001 (0.003) −0.007 (0.036) 0.000 (0.000) −0.009 (0.044) −0.002 (0.001) 0.009 (0.029)

Sector (head)

Subsistence S. (head) 0.005* (0.002) −0.042 (0.029) 0.005* (0.002) −0.005 (0.040) 0.001 (0.001) −0.036 (0.042) −0.002 (0.001) 0.003 (0.051) 0.006* (0.003) −0.002 (0.042)

Non-subs. (head) −0.004 (0.002) −0 .021* (0.010) −0.001 (0.0 01) −0.023 (0.013) 0.002 (0.002) 0.002 (0.011) 0.000 (0.001) −0.001 (0.015) −0.002 (0.001) −0.020 (0.013)

Employment rate −0.001 (0.001) 0.062* (0.032) −0.003 (0.001) −0.075 (0.046) 0.003* (0.001) 0.117** (0.041) 0.005 (0.003) −0.032 (0.059) −0.005 (0.003) −0.046 (0.052)

Intercept 0.168** (0.059) 0.385*** (0.075) −0.174* * (0.0 67) −0.002 (0.084) 0.387*** (0.070)

Notes: p-values: * <0.05; ** <0.01; *** <0.001. Robust standard errors in parentheses below. + Education and literacy.

Source: Authors’ calculations based on IAF/IOF.

17South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

2014/2015 can be mostly explained by a compositional effect evaluated with the latest

survey’s estimated coefficients.15 Even if the initial high level of inequality in an under-

developed country like Mozambique does not fit very well in the Kuznets hypothesis, the

increase in inequality does, however, because it went along the emergence of a more di-

versified and skilled modern sector of the economy.

Indeed, the entire increase in inequality can be explained by the higher education of

household heads (0.053 or 75% of the total change) and the declining public sector

(0.022 or 31%).16 In the same line of reasoning, other factors that significantly contrib-

uted to increasing inequality were the declining employment in the subsistence sector

(7%) and the smaller average number of adults per household (6%) or the larger share of

single heads in the population (2%).17 These inequality-enhancing changes in character-

istics were compensated by other changes that helped mitigate the increase. They include

changes in the distribution of the population by province (with a contribution of −17%),

or the increase in the average number of children per household (−2%).

Apart from these compositional changes, there were also structural changes in the

relationship between these characteristics and inequality. The net coefficients or unex-

plained effect is negligible and statistically insignificant. However, this is the result of

some negative and significant effects compensated by a larger intercept. More specif-

ically, we find significant and substantial negative coefficients effects associated with

heads’ education, heads working in the public and non-subsistence sectors, and the num-

ber of adults. These effects are the result of these characteristics being associated with less

inequality in 2014/2015 compared with 1996/1997. This is consistent with the regres-

sions of (log-)consumption on the same set of characteristics showing returns to attained

education of the head in 2014/2015 lower than in 1996/1997 (but slightly higher than

in 2008/2009; see Table A8). That is, the impact on inequality of higher education or a

smaller public sector was mitigated by these facts becoming less dis-equalising over time.

The only positive and significant effect is associated with the household employment

rate, a dis-equalising factor in 2014/2015 that had an equalising effect in 1996/1997.

A closer look at the decomposition in Table 3 brings out the fact that education played

a fundamental role in explaining increasing inequality in all periods, reflecting a con-

solidated long-term trend. The effect of education on inequality was larger than the

15 The characteristics effect explains 99% of the increase in inequality, although 8 percentage

points are due to the change in the proportion of urban population, driven by a change in the

definition of the variable. The other 91 percentage points are reliable because the impact of the

change in the definition of area on the other coefficients is small in 2002/03 (Table 1).

Furthermore, the decomposition is consistent with the findings from comparing every pair of

consecutive surveys or the changes between 2002/03 and 2014/15, using comparable definitions

of area of residence (Table 3). In the latter case, the compositional effect accounts for 87% of the

increase in inequality.

16 Odusola et al. (2017) included the public sector among the skilled labour sectors that tend to

raise inequality in sub-Saharan countries (along mining, finance, insurance and the real estate).

In the Mozambican case, it seems that it was the reduction in this sector that contributed to in-

crease inequality, instead.

17 While the difference in the proportion of urban population contributes to 8%, we know that

the variables are not entirely comparable in both years, and the changes between comparable

years point to this factor not being especially relevant to explain the increasing inequality.

18 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

increase in inequality observed in 1996/1997–2002/2003, and it explained 64% of the

rise between 2008/2009 and 2014/2015. Nevertheless, the absolute increase in the last

period (0.034) was larger than the contribution during the first period (0.027). In the last

period, other factors significantly contributed to increasing inequality, like the decline in

the subsistence sector (12%) and some demographic changes (higher proportion of single

heads, 3%, or age structure, 6%). The reduction in the number of adults contributed to

reducing inequality (by 8%). Regarding the coefficients effect, we see that the effect of

the household employment rate occurred in the first period. The negative coefficients

effect of education took place in the intermediate period, when education became less

associated with inequality (as opposed to the positive effect during the first period, and

the negative but statistically not significant effect in the last one). The coefficients ef-

fect associated with the average number of adults in the household and with the share

with heads employed in the public sector occurred in the last period, however, due to

an important increase in the equalising effects associated with both characteristics (that

followed the reduction in their average values).

The results are robust to some changes in the approach. Table A6 in the Appendix

shows that the results for the 1996/97–2014/15 period (using the RIF regressions re-

ported in Table A5) are very similar had we used nominal instead of real consumption.

The increase in inequality (0.089) can be explained by the combination of higher edu-

cation (0.061) and the decline in public (0.026) and subsistence sectors (0.05). However,

in this case, the total explained component is around 82% of the total change due to a

larger unexplained effect (captured by the intercept).

Finally, with the alternative counterfactual (Table A7) in which we evaluate the change

in characteristics using the initial values of the coefficients, the explained effect for the

increase in real consumption between 1996/97 and 2014/15 is even larger (higher than

100%18), with a much stronger contribution from education (due to the more dis-

equalising effect of this characteristic in 1996/97) and a weaker contribution from the

sectoral composition (because of its smaller equalising effect in the initial year). The co-

efficients effects valued using the final characteristics are very similar to the previous case,

when the initial ones were used (the educational coefficients effect is stronger, though).

6. CONCLUDING REMARKS

In this paper, we have investigated the trend in consumption inequality in Mozambique

since the end of the post-independence violent conflict. We have shown that the growth

pattern that led to a reduction in poverty over time went along with a substantial increase

in inequality, especially in most recent years. This was due to consumption dispropor-

tionally increasing among the better-off.

This increase in inequality, in line with the classical predictions for initial stages in the

development of dualistic economies, can be explained as the result of the emergence of

18 The difference with the previous counterfactual is that in this case the change in characteris-

tics is evaluated using the initial coefficients (generally associated with higher inequality for the

same characteristics). This implies that the aggregate coefficients effect (evaluated with the final

characteristics) is negative, indicating that if characteristics in 1996/97 were the same as in

2014/15, inequality would have been reduced over time.

19South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

an increasingly skilled population working in the small but expanding non-subsistence

private sector of the economy. The impact of the enhancement of this class was curbed

by its weaker association with inequality in consumption, mitigating the final increase

in inequality. However, initial inequality levels are too high to fully fit in the Kuznets’s

story, and the growth path predominant in the sub-Saharan region is quite different

from the one followed by other countries, due to its much weaker manufacturing indus-

try, its strong dependence on natural resources and its emerging service sector character-

ised by low productivity (e.g. Addison et al., 2017). In this context, rising inequality is the

most likely result in economies with growth taking place in sectors characterised by high

asset concentration, high capital absorption and skilled labour intensity, such as mining,

finance, insurance and real estate (Odusola et al., 2017). The opposite would be expected

if growth were based in labour-intensive sectors such as manufacturing, construction and

agriculture.

Growing inequality occurs in an already unequal country with a large divide between

urban and rural areas and among regions, with a limited redistributive capacity of the

state. This raises legitimate concerns about the implications of the current distributional

growth pattern if it accentuates the duality of the economy, especially when the country

is still facing major challenges in improving the living conditions for most of its popula-

tion, nearly half of which remains below the poverty line. For fulfilling the “no one left

behind” target encouraged by the sustainable development goals, Mozambique needs to

accelerate the expansion of education. This expansion should reach, especially, the least

developed rural communities to improve their productivity and living conditions above

the average, helping to narrow the large urban-rural gap.

REFERENCES

ADDISON, T., PIKKAR AINEN, V., RÖNKKÖ, R. and TAR P, F. (2017 ). Development and Poverty in Sub-Saharan

Africa. Working Paper 2017/169. Helsinki: UNU-WIDER.

ALVAREDO, F. and GASPARINI, L. C. (2015). Recent trends in inequality and poverty in developing countries.

In A. B. Atkinson and F. Bourguignon (eds), Handbook of Income Distribution, Vol. 2, Chapter 9. Amsterdam, The

Netherlands: Elsevier, 697-805.

ARNDT, C., BECK, U., HUSSAIN, M. A., SIMLER, K., TARP, F. and MAHRT, K. (2017a). User guide to Poverty

Line Estimation Analytical Software (PLEASe). In C. Arndt and F. Tarp (eds), Measuring Poverty and Wellbeing in

Developing Countries. Oxford, UK: Oxford University Press, 305-324.

, GARCI A, A., TARP, F. and THURLOW, J. (2012). Poverty reduction and economic structure: comparative

path analysis for Mozambique and Vietnam. Review of Income and Wealth, 58(4): 742-763.

, JONES, S., M AHRT, K., SALVUCCI, V. and TARP, F. (2017b). A review of consumption povert y estimation

for Mozambique. In C. Arndt and F. Tarp (eds), Measuring Poverty and Wellbeing in Developing Countries. Oxford, UK:

Oxford University Press, 108-120.

, JONES, S. and SALVUCCI, V. (2015). When do relative prices matter for measuring income inequality? The

case of food prices in Mozambique. Journal of Economic Inequality, 13(3): 449-464.

and MAHRT, K. (2017). Is Inequality Underestimated in Mozambique? Accounting for Underreported

Consumption. UNU-WIDER Working Paper 2017/153. Helsinki: UNU-WIDER.

BLINDER, A. S. (1973). Wage discrimination: Reduced form and structural estimates. Journal of Human Resources,

8(4): 436- 455.

JONES, S. and TARP, F. (2016). Mozambique: Growth experience through an employment lens. In H. Bhorat and

F. Tarp (eds), Africa’s Lions: Growth Traps and Opportunities for Six African Economies. Washington DC: Brookings

Institution Press, 145-80.

ELBERS, C., LA NJOUW, P., MISTIAEN, J., ÖZLER, B. and SIMLER, K. (2005). Are neighbou rs equal? Estimating

local i nequality in t hree developing c ountries. In R . Kanbur and A . J. Venables (ed s), Spatial Inequality and Development.

New York: Oxford University Press, 37-76.

FIRPO, S., FORTIN, N. M. and LEMIEUX, T. (2007). Decomposing Wage Distributions using Recentered Influence

Function regressions. Unpublished Manuscript, University of British Columbia.

, FORTIN, N. M. and LEMIEUX, T. (2009). Unconditional quantile regressions. Econometrica, 77: 953-973.

20 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

FORTIN, N. M., LEMIEU X, T. and FIRPO, S. (2011). Decomposition methods in economics. In O. Ashenfelter and

D. Card (eds), Handbook of Labor Economics, Vol. 4 Part A. Amsterdam: Elsevier, 1-102.

FOX, L., BARDASSI, E. and VAN DEN BROECK, K. (2005). Poverty in Mozambique: Unraveling Changes and

Determinants. Africa Region Working Paper Series 87, Washington, DC: World Bank.

GRADÍN, C. (2016). Why is income inequality so high in Spain? In L. Cappellari, S. Polachek, and K. Tatsiramos

(eds), Inequality Around the World, Research on Labor Economics, Vol. 44. Bingley, West Yorkshire: Emerald Group

Publishing Limited, 109- 177.

(2018). Explaining Cross-state Earnings Inequality Differentials in India: An RIF Decomposition Approach.

WIDER Working Paper 2018/24. Helsinki: UNU-WIDER.

HAMPEL, F. R. (1974). The influence curve and its role in robust estimation. Journal of the American Statistical

Association, 60: 383-393.

JAMES, R. C., ARNDT, C. and SIMLER, K. R. (2005). Has Economic Growth in Mozambique Been Pro-poor?

Washington: International Food Policy Research Institute.

MEF/DEEF. (2016). Pobreza e bem-estar em Moçambique: Quarta avaliação nacional Inquérito ao Orçamento Familiar-

lOF 2014/15. Maputo: Ministerio de Economia e Finanças, Direcção de Estudos Económicos e Financeiros.

MONTI, A. C. (1991). The study of the Gini concentration ratio by means of the influence function. Statistica, 51(4):

561-577.

MPD/DNE AP. (2010). Pobreza e bem-estar em Moçambique: Terceira avaliação nacional. Maputo: Ministério da

Planificação e Desenvolvimento, Direcção Nacional de Estudos e Análise de Políticas.

OAX ACA, R. L. (1973). Male-female wage differentials in urban labor markets. International Economic Review, 14(3):

693-709.

and RANSOM, M. R. (1999). Identification in detailed wage decompositions. Review of Economics and

Statistics, 81: 15 4-157.

ODUSOLA, A., CORNIA, G. A., BHORAT, H. and CONCEIÇÃO, P. (EDS) (2017). Income Inequality Trends in

Sub-Saharan Africa: Divergence, Determinants and Consequences. New York: United Nations Development Programme

(UNDP).

VAN DER BERG, S., DA MAIA, C. and BURGER, C. (2017). Educational Inequality in Mozambique. WIDER

Working Paper 2017/212. Helsinki: UNU-WIDER.

21South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

APPENDIX

Table A1. Inequality indices and quantiles (real per capita consumption, divided by the poverty line)

Level Change (%)

1996 /97 20 02/03 20 08/09 2014/15 19 96/ 200 2 2002/2008 2008/2014 1996/ 2014

Mean 0.972 1. 315 1.329 1.613 35. 2 1.1 21.4 65.9

Quantiles

0.254 0.333 0.316 0.337 31.2 −5.0 6.6 32.8

p10

0.329 0.427 0 .417 0.445 29. 8 −2.2 6.7 35.5

p25

0.485 0.630 0.637 0.683 29.9 1.1 7. 2 40.8

p50

0.726 0.958 0.977 1.0 69 31.9 2.0 9.5 4 7.4

p75

1.114 1.471 1.518 1.699 32.0 3.2 12.0 52 .5

p90

1.721 2.320 2.318 2.781 34.8 −0.1 20.0 61.6

p95

2.356 3.170 3.138 4.086 34.6 −1.0 30.2 73.4

Inequality

p50∕p10

2.2 2.2 2.3 2.4 1.7 4.2 2.7 8.8

p90∕p50

2.4 2.4 2 .4 2.6 2.2 −2.0 9.6 9.7

Gini

0.397 (0.002) 0.415 (0.004) 0.415 (0.003) 0.468 (0.002) 4.6 −0.1#12.7 17. 8

GE(−2)

0.565 (0.012) 0.679 (0.025) 1.127 (0.06 4) 15.06 8 (4.775) 20.2 65.9 1 2 37. 9 2568.5

GE(−1)

0.322 (0.004) 0.357 (0.007) 0.409 (0.008) 0.532(0.016) 10.8 14 .4 30.2 65. 2

GE(0)

0.268 (0.003) 0.297 (0.007) 0.303 (0.005) 0.381 (0.004) 10.7 2.2#25.7 42.3

GE(1)

0.313 (0.006) 0.380 (0.016) 0.367 (0.010) 0.520(0.013) 21.1 −3.4#41.8 66.0

GE(2)

0.601 (0.031) 1.132 (0.162) 0.887 (0.061) 2.242 (0.322) 88.5 −21.6 #152 .8 273.4

A(.1)

0.030 (0.001) 0.036 (0.001) 0.035 (0.001) 0.049 (0.001) 19.0 −2 .7#38.7 60.7

A(.25)

0.072 (0.001) 0.084 (0.003) 0.082 (0.002) 0.111 ( 0.00 2) 16.4 −1.7#34.7 54.0

A(.5)

0.133 (0.002) 0.151 (0.004) 0.150 (0.003) 0.194 (0.003) 13.1 −0.4#29.2 45.5

A(.75)

0.187 (0.002) 0.207 (0.005) 0.209 (0.003) 0.261 (0.003) 10.8 0.8#24.8 39.3

A(.9)

0.216 (0.002) 0.238 (0.005) 0.241 (0.004) 0.295 (0.003) 9.8 1.4 #22.5 36.5

A(1)

0.235 (0.003) 0.257 (0.005) 0.262 (0.004) 0.317 (0.003) 9.2 1.9#21.1 34.8

A(2)

0.392 (0.003) 0.417 (0.005) 0.450 (0.005) 0.516 (0.008) 6.3 7. 9 14.6 31.6

Note:

𝜀

)

= Atkinson family;

GE (

𝛼

)

= Generalised Entropy family. Bootstrap standard error in parentheses (500 replications). (#) Statistically not

significant at 95%; the rest of the changes in Gini, Atkinson and GE indices are statistically significant (bias-corrected confidence intervals).

Source: Authors’ calculations based on IAF/IOF.

22 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A2. Inequality indices (real equivalised consumption using square root of the

household size)

Level Change (%)

1996 /97 20 02/03 2008/09 2014 /15 1996/2 002 2002/2008 2008/2014 199 6/2014

p50∕p10

2.1 2.1 2.3 2.3 2.2 8.7 0 pt 14.0

p90∕p50

2.1 2.4 2.3 2.5 12.1 −6.2 9.1 14. 8

Gini

0.373 0.405 0.397 0.439 8.6 −2.1 10.6 17.6

GE(−2)

0.4 31 0.579 0.820 10. 221 34.5 41.6 1146 2274

GE(−1)

0.271 0.327 0.357 0.481 20.7 9.3 34.7 7 7. 6

GE(0)

0.237 0.279 0.277 0.336 17. 6 −0.9 21.6 41.7

GE(1)

0.285 0.346 0.332 0.431 21.1 −3.8 29.7 51.1

GE(2)

0.529 0.761 0. 751 1.419 43.9 −1.4 88.9 168.0

A(.1)

0.028 0.033 0.032 0.041 20.2 −3.8 28.1 48.2

A(.25)

0.065 0.078 0.075 0.095 19.1 −3.5 25.9 44.7

A(.5)

0.121 0.142 0.138 0.169 17.5 −2.9 2 2.9 40.3

A(.75)

0.169 0.196 0.192 0.232 16.3 −1.9 20.3 3 7.3

A(.9)

0.195 0.225 0.222 0. 265 15.6 −1. 2 19.0 35.9

A(1)

0.211 0.244 0.242 0.286 15.3 −0.8 18 .2 35. 2

A(2)

0.351 0.396 0. 417 0.491 12.5 5.4 17.7 39.6

Note:

𝜀

)

= Atkinson family;

GE (

𝛼

)

= Generalised Entropy family.

Source: Authors’ calculations based on IAF/IOF.

Table A3. Inequality indices (nominal per capita consumption)

Level Change (%)

1996 /97 20 02/03 2008/09 2014 /15 19 96/200 2 2002/2008 2008/2014 1996/ 2014

p50∕p10

2.3 2.3 2 .5 2.6 −0.9 7.9 3.5 10.6

p90∕p50

2.7 2.7 2.6 3.2 −0.5 −1.9 20.9 18.0

Gini

0.448 0.471 0.460 0.537 5.1 −2.3 16 .7 19.8

GE(−2)

0.781 0.921 1.601 21.23 17.9 73.9 1226 2618

GE(−1)

0.419 0.460 0.513 0.734 9.7 11.5 43.2 75.1

GE(0)

0.341 0.380 0.371 0.505 11.4 −2 .4 35.9 47.8

GE(1)

0.416 0.520 0.465 0.701 25.2 −10.7 50.8 68.5

GE(2)

0.966 2.008 1. 315 3.166 10 7. 8 −34.5 141 228

A(.1)

0.040 0.049 0.044 0.0 65 22.5 −9.6 47.6 63.5

A(.25)

0.094 0.111 0.102 0.147 19.2 −8 .1 43.3 56.9

A(.5)

0.170 0 .195 0.184 0. 252 14.8 −5.9 37.2 48.3

A(.75)

0.234 0.262 0.252 0.332 11.7 −3.8 32 .1 41.8

A(.9)

0.268 0.296 0.288 0.372 10.2 −2.7 29.4 38.8

A(1)

0.289 0.316 0.310 0.396 9.4 −2.0 27. 7 37.0

A(2)

0.456 0.479 0.50 6 0.595 5.1 5.7 1 7.5 30.4

Note:

𝜀

)

= Atkinson family;

GE (

𝛼

)

= Generalised Entropy family.

Source: Authors’ calculations based on IAF/IOF.

23South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A5. RIF Regressions, 1996/97–2014/15 (nominal per capita consumption)

1996 /97 2 014/15

Area

Urban 1996 0.081*** –

Urban –0. 015

Province

Niassa −0.100 * −0.277***

Cabo Delgado −0.095* −0.296***

Nampula − 0. 051 −0.202***

Zambezia −0.130*** −0.205***

Tet e −0.07 −0.284***

Manica −0.044 −0.260***

Sofala 0.007 −0.172***

Inhambane −0.100* −0.232***

Gaza −0.090* −0.206***

Maputo City 0.132* 0.548***

Household size

N. adults −0.002 −0.030***

N. children −0.005 −0.006*

Age (head)

25−34 0.022 − 0.011

35−44 0.061** 0.062***

45−54 0.065** 0.090***

55 or older 0.067** 0.098***

Sex (head)

Female 0.031 −0.022

Marit al sta tus (head)

Single 0.087* 0.08

Divorced −0.023 −0.033

Education (hea d):

Attai ned education

Some/lower primary 0.022 −0.030***

Table A4. Inequality indices (nominal equivalised consumption using the square root of the

household size)

Level Change (%)

1996 /97 20 02/03 2008/09 2014 /15 1996/2 002 2002/2008 2008/2014 199 6/2014

p50∕p10

2.2 2.2 2.4 2 .5 0.1 11. 0 2.8 14.1

p90∕p50

2.7 2.8 2.6 3.1 2 .1 −8.2 21.5 13.8

Gini

0.442 0.468 0.445 0.511 5.8 −5.0 14 .9 15. 6

GE(−2)

0.662 0.815 1.141 14.386 23.1 40.1 1161 2074

GE(−1)

0.388 0.440 0.456 0.671 13.4 3.6 47. 3 73.1

GE(0)

0.330 0.371 0.345 0.456 12.6 −7.2 32.2 38.2

GE(1)

0.406 0.484 0.421 0.596 19. 2 −13.0 41.4 46.6

GE(2)

0.859 1.267 1.013 2.013 47. 5 −20.1 98.8 134

A(.1)

0.039 0.046 0.040 0.056 17.9 −12.4 39.4 43.9

A(.25)

0.091 0.106 0.094 0.128 16. 2 −11.5 36.5 40.4

A(.5)

0.166 0 .189 0.171 0.226 13.8 −9.8 32.3 35.9

A(.75)

0.228 0.255 0.235 0.303 11.9 −7.9 28.7 32.6

A(.9)

0.261 0.289 0.270 0.342 11. 0 −6.8 26.7 31.1

A(1)

0.281 0.310 0.292 0.366 10.4 −6.0 25.6 30.3

A(2)

0.437 0.468 0.477 0.573 7.1 1.9 20.2 31. 2

Note:

𝜀

)

= Atkinson family;

GE (

𝛼

)

= Generalised Entropy family.

Source: Authors’ calculations based on IAF/IOF.

(Continued)

24 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A6. Decomposition of the increase in Gini inequality, 1996/97–2014/15 (nominal

per capita consumption)

1996 /97 2 014/15

Change in Gini 0.089*** (0.009)

Cha r. E Coef. E

Total Effect 0.073*** (0.011) 0.016 (0.011)

Area 0.002 (0.002) −0.014* (0.006)

Province −0.016*** (0.003) −0.102* (0.051)

Household size

N adults 0.005** (0.001) −0.087*** (0.021)

N children −0.001* (0.001) −0.003 (0.015)

Age (head) 0.001 (0.001) 0.005 (0.023)

Sex (head) −0.002 (0.002) −0.009 (0.007)

Marit al sta tus (head)

Single 0.000 (0.000) 0.000 (0.002)

Divorced −0.0 01 (0.001) −0.001 (0.005)

Education+ (head) 0.061*** (0.008) −0.044*** (0.007)

Employment:

Typ e (hea d)

Public s. (head) 0.026*** (0.004) −0.045*** (0.009)

Self-employed (head) −0.001 (0.001) −0.013 (0.022)

Sector (head)

Subsistence S. (head) 0.005* (0.002) − 0.053 (0.030)

Non-subs. (head) −0.006** (0.002) −0.030** (0.010)

Employment rate −0.002* (0.001) 0.096** (0.034)

Intercept 0.318*** (0.069)

Notes: p-values: * <0.05; ** <0.01; *** <0.001. Robust standard errors in parentheses below.

Source: Authors’ calculations based on IAF/IOF.

1996 /97 2 014/15

Upper primary 0.178*** 0.012

Lower secondary 0.648*** 0.152***

Upper secondar y 0.941* 0.502***

Tec h ni c al 0.914*** 0.647***

Higher 2.902*** 1.920***

Literate 0.000 −0.028***

Employment:

Typ e (hea d) −0.079* −0.485***

Public sector

Self-employed 0.044 0.026

Sector+ (head)

Subsistence 0.01− −0.066*

Non-subsistence 0.051 −0.087**

Employment rate (hou sehold) −0.0 51 0.068*

Intercept 0.428*** 0.746***

N Observations 42,143 164 ,359

R28.7 10.1

F9.65*** 13.08***

Note: p-values: * <0.05; ** <0.01; *** <0.001. +A category was included to indicate that the head’s

industry was missing.

Source: Authors’ calculations based on IAF/IOF.

Table A5. (Continued)

25South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A7. Decomposition of the increase in Gini inequality, 1996/97–2014/15 (alternative

counterfactual)

1996 /97 2 014/15

Change in Gini 0.071*** (0.008)

Cha r. E Coef. E

Total Effect 0.103*** (0.016) −0.033 (0.019)

Area 0.006* (0.002) −0.001 (0.009)

Province − 0.001 (0 .001) − 0.043 (0.037)

Household size

N adults 0.000 (0.001) −0.087*** (0.019)

N children −0.001 (0.0 01) −0.011 (0.015)

Age (head) 0.000 (0.000) 0.040 (0.029)

Sex (head) 0.001 (0.002) −0.007 (0.008)

Marit al sta tus (head)

Single 0.000 (0.000) 0.000 (0.002)

Divorced 0.000 (0.001) −0.004 (0.006)

Education+ (head) 0.093*** (0.016) −0.064** (0.020)

Employment

Typ e (hea d)

Public s. (head) 0.003 (0.002) −0.021*** (0.004)

Self−employed (head) −0.001 (0.001) −0.001 (0.020)

Sector (head)

Subsistence S. (head) 0.000 (0.003) −0.037 (0.026)

Non−subs. (head) 0.002 (0.002) −0.027* (0.013)

Employment rate 0.0 01 (0.001) 0.060* (0.031)

Intercept 0.168** (0.059)

Notes: p-values: * <0.05; ** <0.01; *** <0.001. Robust standard errors in parentheses below.

Source: Authors’ calculations based on IAF/IOF.

Table A8. (Log-)Real per capita consumption regressions, 1996/97–2014/15

1996 /97 20 02/03 2002/03 2008/09 2014 /15

Area

Urban 1996 −0.055*** −0.088*** – – –

Urban – – −0.107*** −0.061*** −0.119***

Province

Niassa −0.125*** 0.346*** 0.356*** 0.392*** −0.557***

Cabo Delgado 0.118*** 0.125*** 0.138*** 0.317*** −0.296***

Nampula −0.091*** 0.216*** 0.240*** 0.135*** −0.519***

Zambezia −0.018 0.308*** 0.315*** 0.003 −0.536***

Tet e −0.261*** −0.006 0.000 0.231*** −0.217***

Manica 0 .174*** 0.289*** 0.300*** 0.055* * −0.285***

Sofala −0.537*** 0.512*** 0.520*** 0.004 −0.242***

Inhambane −0.225*** −0.301*** −0.288*** 0.202*** −0.359***

Gaza 0.167*** 0.294*** 0.305*** 0.039* −0.372***

Maputo City 0.129*** 0.316*** 0.311*** 0.349*** 0.227***

Household size

N. adults −0.032*** 0.016*** 0.016*** −0.014*** − 0.0 01

N. children −0.118*** −0.108*** −0.108*** −0.108*** −0.125***

Age (head)

25−34 −0.116*** −0.060** −0.058* −0.006 −0.002

35−44 −0.065*** −0.093*** −0.093*** − 0.014 0.062***

45−54 −0.003 −0.032 −0.03 0.026 0.068***

55 or older −0.067*** −0.048* −0.044 0.054** 0.076***

Sex (head)

Female 0.026 0.094*** 0.095*** 0.074*** 0.013*

Marit al sta tus (head)

Single 0.038 −0.03 − 0.0 21 0.101* * −0.008

Divorced −0.007 −0.117*** −0.110*** −0.128*** −0.040***

(Continued)

26 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

1996 /97 20 02/03 2002/03 2008/09 2014 /15

Education (hea d):

Attai ned education

Some/lower primary 0.131*** 0.043*** 0.052*** 0.005 0.067***

Upper primary 0.347*** 0.310*** 0.322*** 0.120*** 0.169***

Lower secondary 0.542*** 0.635*** 0.650*** 0.344*** 0.385***

Upper secondar y 0.848*** 0.994*** 1.008*** 0.637*** 0.678***

Tec h ni c al 0.755*** 0.873*** 0.884*** 0.720*** 0.777***

Higher 1.390*** 1.324*** 1.327*** 1.106*** 1.229***

Literate 0.127*** 0.097*** 0.096*** 0.168*** 0.114***

Employment:

Typ e (hea d)

Public sector 0.014 −0.032 −0.042* 0.079*** −0.039***

Self-employed 0.084*** 0.059*** 0.051*** 0.178*** 0.039***

Sector+ (head)

Subsistence −0.116*** −0.179*** −0.178*** −0.123*** −0.085***

Non-subsistence 0.139*** 0.054* 0.067** 0.120*** 0.168***

Employment rate

(household)

0.02 0.180*** 0.162*** −0.048* 0.237***

Intercept 0.151*** −0.082* −0.063 0.042 0.451***

N Observations 42 ,143 44,083 44,083 51,175 164 ,359

R230.0 27. 4 27. 6 24.0 34.8

F394*** 269*** 268*** 281*** 2,026***

Note: p-values: * <0.05; ** <0.01; *** <0.001. + A category was included to indicate that the head’s

industry was missing.

Source: Authors’ calculations based on IAF/IOF.

Table A8. (Continued)

27South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A9. Differentials in Lorenz curves between years

Percentile 1996/ 200 2 2002/08 2008/14 199 6/2 014 percentile 199 6/20 02 2002/08 20 08 /14 199 6/2014

10.000 0.000* 0.000 0.000*** 51 0.009 0.002 0.025*** 0.036***

20.000 0.001** 0.000 0.001*** 52 0.009 0.001 0.026*** 0.036***

30.000 0.001** 0.000 0.001*** 53 0.009 0.001 0.027*** 0.037***

40.000 0.001** 0.000 0.002*** 54 0.009 0.001 0.028*** 0.038***

50.000 0.002* 0.000 0.002*** 55 0 .010 0.001 0.028*** 0.039***

60.0 01 0.002* 0.001 0.003*** 56 0.010 0.0 01 0.029*** 0.040***

70.0 01 0.002* 0.001 0.003*** 57 0.010 0.0 01 0.030*** 0.041***

80.0 01 0.002* 0.001* 0.004*** 58 0.010 0.001 0.031*** 0.042***

90.0 01 0.002* 0.002* 0.005*** 59 0.010 0.001 0.032*** 0.043***

10 0.0 01 0.002* 0.002** 0.005*** 60 0.011 0.000 0.033*** 0.044***

11 0.0 01 0.002* 0.002** 0.006*** 61 0.011 0.000 0.033*** 0.045***

12 0.0 01 0.002* 0.003** 0.006*** 62 0.011 0.000 0.034*** 0.045***

13 0.002 0.002* 0.003** 0.007*** 63 0.011 0.000 0.035*** 0.046***

14 0.002 0.002 0.004** 0.008*** 64 0.011 0.000 0.036*** 0.047***

15 0.002 0.002 0.004** 0.009*** 65 0.012 −0.001 0.037*** 0.048***

16 0.002 0.002 0.005*** 0.009*** 66 0.012 −0 .001 0.038*** 0.049***

17 0.002 0.002 0.005*** 0.010*** 67 0.013 −0.0 01 0.039*** 0.050***

18 0.002 0.003 0.006*** 0.011*** 68 0.013 −0.001 0.039*** 0.051***

19 0.002 0.003 0.006*** 0.011*** 69 0.013 −0.002 0.040*** 0.052***

20 0.003 0.003 0.006*** 0.012*** 70 0.013 −0.002 0.041*** 0.053***

21 0.003 0.003 0.007*** 0.013*** 71 0.014 −0.002 0.042*** 0.053***

22 0.003 0.003 0.007*** 0.013*** 72 0.014 −0.002 0.043*** 0.054***

23 0.003 0.003 0.008*** 0.014*** 73 0.014 −0.003 0.044*** 0.055***

24 0.004 0.003 0.009*** 0.015*** 74 0 .015 −0.003 0.044*** 0.056***

25 0.004 0.003 0.009*** 0.016*** 75 0 .015 −0.003 0.045*** 0.057***

26 0.004 0.003 0.010*** 0.016*** 76 0 .015 −0.003 0.046*** 0.058***

27 0.004 0.003 0.010*** 0.017*** 77 0 .015 −0.003 0.047*** 0.059***

28 0.004 0.003 0.011*** 0.018*** 78 0.016 −0.004 0.048*** 0.060***

29 0.005 0.003 0.011*** 0.019*** 79 0. 017* −0.004 0.049*** 0.061***

30 0.005 0.003 0.012*** 0.019*** 80 0 .017 −0.005 0.050*** 0.061***

31 0.005 0.003 0.012*** 0.020*** 81 0 .017 −0.005 0.051*** 0.062***

32 0.005 0.003 0.013*** 0.021*** 82 0.017 −0.005 0.051*** 0.063***

33 0.005 0.003 0.013*** 0.021*** 83 0.017 −0.005 0.052*** 0.064***

34 0.005 0.003 0.014*** 0.022*** 84 0. 017 −0.005 0.053*** 0.065***

35 0.005 0.003 0.015*** 0.023*** 85 0.017 −0.005 0.053*** 0.065***

36 0.006 0.003 0.015*** 0.023*** 86 0. 017 −0.006 0.054*** 0.066***

37 0.006 0.003 0.016*** 0.024*** 87 0.017 −0.006 0.055*** 0.066***

38 0.006 0.003 0.016*** 0.025*** 88 0. 018 −0.006 0.055*** 0.067***

39 0.006 0.002 0.017*** 0.026*** 89 0.018 −0.006 0.056*** 0.068***

40 0.006 0.002 0.018*** 0.027*** 90 0.017 −0.005 0.056*** 0.068***

41 0.007 0.002 0.018*** 0.027*** 91 0.018 −0.006 0.056*** 0.069***

42 0.007 0.002 0.019*** 0.028*** 92 0.018 −0.006 0.056*** 0.069***

43 0.007 0.002 0.020*** 0.029*** 93 0.018 −0.006 0.056*** 0.068***

44 0.007 0.002 0.020*** 0.030*** 94 0.019 −0.006 0.055*** 0.068***

45 0.008 0.002 0.021*** 0.031*** 95 0 .019 −0.006 0.054*** 0.067***

46 0.008 0.002 0.022*** 0.031*** 96 0.019 −0.005 0.052*** 0.066***

47 0.008 0.002 0.022*** 0.032*** 97 0.0 21* −0.005 0.049*** 0.065***

48 0.008 0.002 0.023*** 0.033*** 98 0.021* −0.003 0.044*** 0.063***

49 0.008 0.002 0.024*** 0.034*** 99 0.022** −0.0 01 0.035*** 0.055***

50 0.009 0.002 0.025*** 0.035***

Note: p-values: * <0.05; ** <0.01; *** <0.001 (bootstrap standard errors).

Source: Authors’ calculations based on IAF/IOF.

28 South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Table A10. Daily consumption shares (deciles, quintiles and top 5%) with standard errors

Per ca pita Equivalised (square root of household size)

1996 /97 20 02/03 2008/09 2014 /15 19 96/97 2002/03 2008/09 2014/15

Real* Sha re St.E. Share St.E. Share St.E. Share St.E. Share St.E. Share St.E . Share St.E. Share St.E.

D1 2.49 0.02 2.40 0.03 2.18 0.02 1.97 0 .01 2.80 0.02 2.57 0.03 2.31 0.02 2.13 0.01

D2 3.95 0.03 3.77 0.04 3.72 0.03 3.28 0.02 4.32 0.03 3.98 0.03 3.92 0.03 3.54 0.02

D3 4.97 0.03 4.77 0.04 4.75 0.03 4.23 0.02 5.31 0.03 4.92 0.04 5.01 0.03 4.54 0.02

D4 5.87 0.03 5.70 0.05 5.75 0.04 5 .15 0.02 6.26 0.03 5. 81 0.04 6.02 0.04 5.50 0.02

D5 6.92 0.04 6.71 0.06 6.78 0.04 6.10 0.03 7.22 0.04 6 .74 0.04 7. 02 0.04 6.50 0.02

D6 8.08 0.04 7.8 8 0.06 8.00 0.05 7.19 0.03 8.28 0.04 7.8 5 0.05 8.20 0.05 7.5 9 0.03

D7 9.48 0.05 9.19 0.07 9.43 0.05 8.59 0.04 9.58 0.04 9.19 0.06 9. 62 0.05 8.95 0.03

D8 11.50 0.05 11.19 0.08 11. 46 0.07 10.60 0.05 11.27 0.05 11.12 0.06 11.52 0.06 10.89 0.04

D9 14.82 0.07 14.68 0.11 14.82 0.08 14.15 0.06 14.15 0.06 14.43 0.08 14.63 0.07 14.32 0.05

D10 31.91 0.27 33.71 0.46 33.10 0.35 38.73 0.26 30.82 0.26 33.39 0.34 31.75 0.33 36.03 0.20

Q1 6.45 0.05 6 .17 0.06 5.90 0.05 5.25 0.03 7.1 2 0.05 6.55 0.05 6.23 0.05 5.66 0.03

Q2 10.84 0.06 10.47 0.09 10.51 0.07 9.38 0.04 11.57 0.06 10.73 0.07 11.03 0.07 10.05 0.04

Q3 15.00 0.08 14.58 0.12 14.7 8 0.09 13.30 0.06 15 .4 9 0.07 14.59 0.09 15. 22 0.08 14 .09 0.05

Q4 20.98 0.10 20.39 0.15 20.89 0.12 19.19 0.08 20.85 0.09 20.31 0.12 21.14 0 .11 19.85 0.07

Q5 46.73 0.24 48.39 0.38 47.92 0.29 52.88 0.21 44.97 0.24 4 7.82 0.30 46.38 0.28 50.35 0.17

Bottom

0.98 0. 01 0.94 0.01 0.78 0.01 0.75 0.01 1.11 0.01 1.01 0.01 0.85 0.01 0.80 0.01

Top 5% 21.76 0.27 23.63 0.50 23.04 0.37 28.46 0.29 21.28 0.26 23.07 0.36 21.96 0.35 25.83 0.22

Val ue St.E. Valu e St.E. Va lu e St.E. Va lue St.E. Value St.E . Val ue St.E. Va lu e St.E. Va lu e St.E.

Mean 5,350 29 10,924 78 23.829 0.140 47.0 88 0.246 12,165 61 25, 211 148 53.617 0.290 106.180 0.4 41

Median 3,993 7,9 55 17.5 09 31.214 9,4 26 18,409 40.413 74 .491

Pover ty

line

5,502 8,307 17.90 0 29.200

Constant

Mean

0.972 1 .315 1.329 1.613

Nominal Share St.E. Share St.E. Share St.E. Share St.E. Share St.E. Share St.E . Share St.E. Sha re St.E.

D1 2.12 0.02 2.09 0.03 1.91 0.02 1.58 0.01 2.29 0.02 2.23 0.03 2 .01 0.02 1.71 0.01

D2 3.42 0.03 3.33 0.03 3.23 0.03 2.63 0.02 3.59 0.03 3.40 0.03 3.40 0.03 2.84 0 .01

D3 4.39 0.03 4. 21 0.04 4.19 0.03 3.46 0.02 4.54 0.03 4.18 0.03 4.39 0.03 3.70 0.02

D4 5.25 0.03 5.07 0.05 5.16 0.04 4.25 0.02 5.37 0.03 5.03 0.04 5.33 0.03 4.53 0.02

D5 6.19 0.04 5.99 0.05 6.20 0.04 5.14 0.03 6.24 0.04 5.93 0.04 6.37 0.04 5.46 0.02

D6 7.38 0.04 7.03 0.06 7.4 2 0.05 6. 21 0.03 7.29 0.04 6 .97 0.05 7.5 7 0.04 6.57 0.03

D7 8.97 0.05 8.35 0.07 8.99 0.05 7.6 6 0.04 8.68 0.05 8.27 0.05 9.11 0.05 7.9 8 0.03

D8 11.0 2 0.05 10.37 0.09 11.14 0.07 9.8 8 0.05 10.71 0.06 10.31 0.06 11.24 0.06 10.21 0.04

D9 14.86 0.08 14.17 0.11 14.73 0.08 14.16 0.07 14.73 0.07 14.48 0.09 14 .88 0.07 14.59 0.06

D10 36.38 0.28 39.38 0.46 37.0 4 0.34 45. 03 0.27 36 .56 0.27 39.19 0.34 35.70 0.31 42 .41 0.21

Q1 5.55 0.04 5.42 0.06 5 .14 0.04 4.22 0.03 5.88 0.04 5.63 0.05 5.41 0.04 4.55 0.02

29South African Journal of Economics Vol. 0:0 Month 2019

Economic Society of South Africa.

Q2 9.65 0.06 9.27 0.08 9.34 0.07 7.71 0.04 9.91 0.06 9.21 0.07 9.72 0.06 8.23 0.04

Q3 13.57 0.08 13.02 0.11 13.62 0.09 11.3 4 0.06 13.54 0.08 12.90 0.09 13.94 0.08 12.04 0.05

Q4 19.99 0.10 18.72 0.15 2 0.13 0.12 17.5 4 0.09 19.38 0.09 18.58 0.11 20.35 0.11 18.18 0.07

Q5 51.2 4 0.25 53.55 0.37 51.7 7 0.28 59.19 0 .21 51. 29 0.24 53.68 0.28 50.58 0.26 57.0 0 0.17

Bottom

0.85 0.01 0.83 0.01 0.69 0.01 0.61 0.01 0.93 0.01 0.88 0.01 0.75 0 .01 0.65 0.01

Top 5% 25.55 0.28 28.88 0.51 26.52 0.37 33.98 0.31 25.69 0.26 28.41 0.35 25.27 0.33 31.0 9 0.24

Val ue St.E. Valu e St.E. Va lu e St.E. Va lue St.E. Value St.E . Val ue St.E. Va lu e St.E. Va lu e St.E.

Mean 5,355 36 10,903 104 23.882 0 .171 46.618 0.289 12,406 79 25,376 19 2 53.7 31 0.338 104.674 0.518

Median 3,594 7,09 0 16.181 26.267 8,323 16, 334 3 7.07 8 62.498

Note: Mean and median in MZM (1996/97 and 2002/03) and MZN (2008/09 and 2014/15). *Real values are obtained normalising nominal con-

sumption to account for variation in prices over the months of the survey and over 13 geographical areas. Def lated values are obtained by dividing

the current real consumption by the corresponding poverty line.

Source: Authors’ calculations based on IAF/IOF.

Stochastic dominance testing of poverty in Mozambique

Article

Full-text available

Jul 2024

National assessments produced regularly from 1996/97 to 2019/20 provide standard estimates of the evolution of poverty in Mozambique. Stylised facts point to significant reduction in monetary and multidimensional poverty until 2014/15, even though characterised by persistent urban–rural and regional inequalities in several dimensions, followed by worsening deprivation during 2014/15–2019/20. In this paper, we rely on first-order dominance (FOD) methodology to check the robustness of standard poverty estimates while providing deeper insights into what happened in this poverty-stricken country. Our main findings include that the capital city, Maputo, and surrounding areas progressively came to dominate all other provinces in terms of well-being, and the provincial rankings from spatial FOD analysis are very stable over time. Rural areas, mainly those in the northern and central regions, are, with a few exceptions, always at the bottom of the ranking. Moreover, our FOD temporal analysis reveals that 2019/20 dominates, with a high degree of confidence, the results from earlier years, pointing to positive development over time, but with one important exception. The comparison between 2019/20 and 2014/15 is indeterminate, indicating that no significant advancement took place during this period. What stands out is the compelling need for concerted policy action to address the strong and persistent spatial gaps in welfare and poverty, including the capacity to respond effectively to external and internal shocks.

Social inequality in Mozambique

Preprint

Full-text available

Jun 2024

This study presents the profile of social inequality in Mozambique from 1997 to 2017. Despite progress in reducing poverty within the government's efforts to achieve Sustainable Development Goals, the country still faces persistent challenges in reducing inequality and policy effectiveness, particularly in employment and access to basic services such as education, electricity, water, and sanitation. The study indicates that inequality in access to basic services tended to decrease until 2017. Gender inequality shows a trend of reduction, but households headed by women represent the most dis-advantaged group with lower access to basic services. At the spatial level, rural areas, especially districts located far from the provincial capital cities, exhibit lower access to basic services.

Assessing the Impact of Covid-19 in Mozambique in 2020

Article

Full-text available

Dec 2023

Taking advantage of the 2019/2020 Mozambican household budget survey, in the field both before and during the first phases of the Covid-19 pandemic, we assess the impact of Covid-19 on welfare in 2020, aiming to disentangle this impact from the effect of other shocks. Comparing a number of welfare metrics, and applying propensity score matching and inverse probability weighted regression adjustment approaches, we find that consumption levels are significantly lower and poverty rates substantially higher during the first phases of Covid-19 than in the pre-Covid-19 period. Moreover, the impact was greater in urban areas and accordingly in the more urbanised southern region. Non-food expenditures suffered relatively more than food expenditures, likely a coping strategy, while the impact on consumption levels was greater for people working in the secondary and tertiary sectors than for workers in the primary sector, mainly agriculture. Stunting among under-5 children also suffered. Only a limited number of countries have actual, collected in-person, survey data that span across the initial phases of the Covid-19 pandemic. Thus, the present analysis adds value to our understanding of the welfare consequences of Covid-19 in a low-income context, where automatic social safety nets were not in place during the early phases of the pandemic. More specifically, it helps in assessing the results of previous welfare impact simulations, compared to real data. Even though our main findings are broadly in line with existing estimates based on simulations or phone surveys, important differences between the predictions and the actual results emerge. We conclude that it is critically important for Mozambique and its development partners to develop stronger and more targeted policies and tools to respond to temporary shocks.

On Income Inequality in Urban Areas in China During the Period 2002–2013: Comparing the Case of Urban Locals With That of Rural Migrants

Chapter

Jan 2023

To explore income inequality in urban China, this paper investigates disparities between- and within-urban locals and rural migrants from 2002 to 2013, using three waves of the China Household Income Project (CHIP) data. While the existing literature concentrates on the wage disparity between these two groups, our results show that the Gini among the migrants increased by 17.86% between 2007 and 2013 and that among the locals increased by 15.54% from 2002 to 2007. The urban–migrant average income gap decreased during the whole period mainly due to higher growth in migrants’ average income. Estimates based on Mincerian earnings functions for both groups reveal the significant role of the education, occupation and type of contract in determining the within-group inequality. In addition, using a recentred influence function (RIF), we observe that short-term and other types of contracts, duration of the job, in-system ownership, marriage and skill have inequality-enhancing effects for migrants. The variation of skills has a larger impact on the income disparity among migrants than on that among urban locals. The RIF-based Blinder–Oaxaca decomposition of the mean difference of incomes shows that labour market discrimination between the two groups is not significant; however, both pure explained and unexplained differences are significant when applying the RIF decomposition to the variance of the logarithms of incomes. While the type of contract significantly reduces the pure explained difference between migrants and urban locals, occupation has a positive impact on this difference between these two groups. The heterogenity analysis shows that the factors influencing incomes in these two groups are different. We recommend labour market intervention to reduce unreasonable occupational and sectoral disparities, especially in the net inflow provinces, to mitigate urban inequality in China effectively.

Overcoming poverty traps in Mozambique: Quantifying inequalities among economic, social and environmental capitals

Article

Nov 2022
J CLEAN PROD

Assessing sustainable development strategies and alternatives depends on placing human well-being and the environment at the core of policy efforts and developing accounting methods to determine whether countries are progressing towards increasingly equal well-being. In this work, emergy accounting was used within the five-sector sustainability model (5SenSu) for measuring resource use inequality. As a diagnostic proposal, the 5SEnSU model based on the theory of ecologically unequal exchanges and emergy synthesis were used to assess Mozambique's poverty traps. Inequalities were measured considering the flows of goods, energy, and money between the social, economic, and environmental sectors as representative of the national economy. The 5SEnSU model showed that the environment directly supports a large part of the Mozambican population and that the country has growth potential according to the natural environment over 38 sej/sej, which must be used to ensure sustainable and equitable growth. Mozambique also has the potential (23:1 in emergy units) to maintain carbon neutrality and preserve forests and soil resources while undergoing economic development and growth. Mozambique has carbon credit and a resource surplus but needs to develop mechanisms to improve economic sectors through cleaner and well-being-centered development policies. High inequality between the social sectors and the economic system exposed a high dependency on local environmental resources characterizing a condition of vulnerability. The reduction in the number of people living below the poverty line and the effective implementation of the sustainable development goals may be achieved by balancing the capacity of the economic system with the country's environmental/social capacitiy.

The evolution of inequality in Mozambique 1996/97–2019/20*

Article

Sep 2024

The impact of energy prices on inflation and economic growth in Mozambique: A wavelet approach and OLS estimator

Article

Mar 2024

Gilmar Fernando Dias da Conceição

The paper is using the Continuous Wavelet Transform (CWT) and the OLS estimator to examine the relationship between energy prices (coal and gas), inflation and economic growth in Mozambique. Coal and gas are the primary commodity and source of revenue for the country. The research uses quarterly data from 2001 to 2022. Among other results, we found evidence of high coherence between macroeconomic variables (inflation and real GDP growth rate) and energy prices (thermal coal and natural gas prices) only for the frequency intervals between 1.5 and 4 years (short term). The results hold even after controlling for the real GDP growth and exchange rate; we concluded that only in the short term (1.5–4 years), there is high coherence between energy prices and inflation in the time interval between 2004 and 2012. This finding indicates no relation between energy prices, inflation and economic growth in Mozambique. Furthermore, the OLS estimates substantiate the wavelet result, showing no impact. In this context, we can conclude that there is probably a ‘Dutch disease’ or ‘resource curse’ that negates the positive effects of exploiting these resources on short‐ and long‐term economic growth.

DESENVOLVIMENTO SUSTENTÁVEL E DESIGUALDADES SOCIAIS EM MOÇAMBIQUE: PERSPETIVAS ÉTICAS E POLÍTICAS

Chapter

Full-text available

Jan 2023

Edgar Manuel Cambaza

Esta obra constituiu-se a partir de um processo colaborativo entre professores, estu-dantes e pesquisadores que se destacaram e qualificaram as discussões neste espaço formativo. Resulta, também, de movimentos interinstitucionais e de ações de incenti-vo à pesquisa que congregam pesquisadores das mais diversas áreas do conhecimento e de diferentes Instituições de Educação Superior públicas e privadas de abrangência nacional e internacional. Tem como objetivo integrar ações interinstitucionais nacio-nais e internacionais com redes de pesquisa que tenham a finalidade de fomentar a formação continuada dos profissionais da educação, por meio da produção e socializa-ção de conhecimentos das diversas áreas do Saberes. Agradecemos aos autores pelo empenho, disponibilidade e dedicação para o desen-volvimento e conclusão dessa obra. Esperamos também que esta obra sirva de instru-mento didático-pedagógico para estudantes, professores dos diversos níveis de ensino em seus trabalhos e demais interessados pela temática.

Agriculture Production and Transport Connectivity: Evidence from Mozambique

Article

Oct 2022

Atsushi Iimi

Despite relative richness of the existing literature, it remains a challenge to consistently estimate the impacts of transport connectivity on agricultural production. The paper constructs pseudo-panel data with transport infrastructure defined at high resolution in two periods of time and examines spatially heterogeneous impacts of improved transport connectivity. The paper takes advantage of the unique circumstances in Mozambique where the Government invested intensively in road infrastructure during a relatively short period of time in the 2010s. Combining the highly disaggregated location-specific fixed effects with the instrumental variable method, the paper controls for the endogeneity to show that the improved road connectivity increased agricultural production significantly. Rural connectivity and domestic market accessibility are found to be of particular importance, but substantial heterogeneity exists across regions. The northern provinces, where transport connectivity is limited, have the potential for agricultural growth, exhibiting increasing returns to scale.

The Politics and Political Economy of Violent Conflicts in Post-war Mozambique

Chapter

Mar 2022

Manuel Francisco Sambo

The outbreak of the insurgency in Mozambique’s Cabo Delgado province in 2017 and RENAMO’s return to violence in 2013 reveal that nearly three decades after the end of the civil war and the implementation of post-war peacebuilding often regarded as a success story, new and old violent conflicts in Mozambique persist. This chapter examines the political and political economy dynamics that fuel these violent conflicts in post-war Mozambique, considering the increasingly globalized world. The chapter argues that, in Mozambique, the expectation that democratization would result in inclusive politics and a free-market economy would produce national prosperity, which together would sustain a lasting peace, did not materialize. Instead, democratization backslid producing and reproducing forms of political exclusion while the economic model implemented did not address or even exacerbated socio-economic problems such as poverty and inequalities. Moreover, political exclusion, poverty, and inequalities are, ultimately, a breeding ground for the occurrence and recurrence of violent conflicts in the country.KeywordsMozambiquePolitical exclusionPovertyInequalitiesViolent conflicts

User Guide to Poverty Line Estimation Analytical Software—PLEASe

Chapter

Full-text available

Dec 2016

Detailed analyses of poverty and wellbeing in developing countries, based on household surveys, have been ongoing for more than three decades. The large majority of developing countries now regularly conduct a variety of household surveys, and their information base with respect to poverty and wellbeing has improved dramatically. Nevertheless, appropriate measurement of poverty remains complex and controversial. This is particularly true in developing countries where (i) the stakes with respect to poverty reduction are high; (ii) the determinants of living standards are often volatile; and (iii) related information bases, while much improved, are often characterized by significant non-sample error. It also remains, to a surprisingly high degree, an activity undertaken by technical assistance personnel and consultants based in developed countries. This book seeks to enhance the transparency, replicability, and comparability of existing practice. It also aims to significantly lower the barriers to entry to the conduct of rigorous poverty measurement and increase the participation of analysts from developing countries in their own poverty assessments. The book focuses on two domains: the measurement of absolute consumption poverty and a first-order dominance approach to multidimensional welfare analysis. In each domain, it provides a series of computer codes designed to facilitate analysis by allowing the analyst to start from a flexible and known base. The volume covers the theoretical grounding for the code streams provided, a chapter on ‘estimation in practice’, a series of eleven case studies where the code streams are operationalized, a synthesis, an extension to inequality, and a look forward.

A Review of Consumption Poverty Estimation for Mozambique

Chapter

Full-text available