Professional Documents
Culture Documents
ECONOMICS
Abstract
Over the past 25 years, higher growth in developing countries has contributed to a dramatic fall in global poverty, although poverty rates in rural
areas remain higher than in urban areas. Unfortunately, projected growth rates have fallen in recent years; this article examines the impact of this
slowdown on the poor, particularly the rural poor. It first uses a global model to assess the impacts of lower productivity on key price and income
variables. It then uses microsimulation models for almost 300,000 households to assess the impacts on their real incomes. Although poverty rates
overall are projected to fall substantially, the poorest countries see the greatest slowdown in poverty reduction, with over 5% of their population
projected to remain below the poverty line. In addition, poverty rates will remain alarmingly high in many countries. Overall, 38 million fewer
people will leave extreme poverty compared to earlier projections. Farm households are at particular risk in middle-income countries, with over
1.5% more of the farming population remaining trapped in poverty than previously estimated. By 2030, average extreme poverty in rural areas is
projected at about 7.5%, rather than 7.1% under the earlier growth projections. Clearly, a strong focus on policies for poverty reduction will be
vital for eliminating poverty by 2030.
JEL classifications: I32, O13, Q12
Keywords: Economic growth; World economy; Rural poverty; Economic slowdown; Poverty reduction
1. Introduction by 2030, the target year for eradicating global poverty under the
United Nations (UN) Sustainable Development Goals (SDGs).
Since the global financial crisis of 2007–2008, the global The approach taken in this study focuses on several of the
economy has suffered a sustained slowdown, with growth lev- key linkages between growth and the real incomes of the poor,
els frequently falling below earlier forecasts. For example, the particularly the rural poor. In particular, it considers the impacts
International Monetary Fund (IMF) World Economic Outlook on the poor of: (1) direct reductions in business income from
(WEO) forecast for global growth in 2017 declined from 4.7% lower productivity growth, (2) changes in wage rates for the
in April 2012 to 3.8% in October 2015, quite close to the end- earnings of unskilled workers, (3) changes in key commodity
2017 estimate of 3.6%. For most countries (although not all), prices for business incomes, and (4) changes in the cost of living.
recent forecasts are substantially lower than they were in more This approach goes beyond the traditional poverty elasticity of
optimistic times, such as 2012. Other international organiza- income growth used in most poverty projections, such as World
tions, such as the Organization for Economic Co-operation and Bank (2015). Adding linkages through changes in commodity
Development (OECD, 2017), have also made sizeable down- prices is likely to be particularly important when considering
ward revisions to their long-run projections, with global growth poverty among farmers and the rural poor.
expected to slow substantially. This article looks at the implica- Using household models for almost 300,000 households
tions that this slowing growth in advanced and emerging market across the developing world, we assess the impacts of changes
economies may have for poverty, and particularly how these less in the growth scenarios on the real incomes of individual house-
optimistic macroeconomic projections may affect rural poverty holds; this includes both changes arising from changes in the
productivity of households’ own activities and those arising
∗ Corresponding author. Tel.: +1 202 862 5628; fax: +1 202 467 4439.
from changes in the prices and wage rates that households face.
E-mail address: w.martin@cgiar.org (W. Martin). Finally, we check whether these income changes result in the
household moving across the poverty line and use the resulting
This is an open access article under the terms of the Creative Commons Attribu-
changes in the poverty headcount as a simple indicator of these
tion License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited. changes’ impact on the poor.
C 2018 The Authors. Agricultural Economics published by Wiley Periodicals, Inc. on
We first develop the growth scenarios to be analyzed. We Therefore, when comparing 2012 projections for the future
then lay out the analytical framework, which uses a global with the 2015-based projections, it is critical to see that both the
computable general equilibrium (CGE) model combined with slope and the level of the GDP trajectories have been affected,
a household modeling approach to properly identify short-term as shown in Fig. 1. For instance, global income growth between
and long-term consequences for poverty. Finally, we describe 2011 and 2015 was expected to be 17.5% in 2012 but turned
the different growth scenarios analyzed and discuss the results. out to be only 13.9%. By 2017, the GDP growth gap reached
5 percentage points (GDP in 2017, as projected in 2012, was
5 percentage points above the 2015 projection for the same
2. Growth scenarios year), while the growth gap (the GDP growth not achieved)
reached 22%. A simple extrapolation shows that global GDP
In this section, we review the different inputs based on the in 2030 will be 15% lower than initially expected. Our objec-
IMF WEO 2012 and 2015 and identify key drivers and nar- tive is to assess the implications of this substantial decline in
ratives for our scenarios. The IMF projections are probably anticipated output for achieving the SDGs.
the most comprehensive and carefully scrutinized set of eco- The growth projections were reduced for 134 countries out
nomic projections available. They provide estimates of growth of a total of 189. Fig. 1 displays an overview of the projected
in national GDP, savings rates, and changes in current account 2017 growth rate. Looking at global GDP, the growth rate was
balances (CABs). Within these projections, the main measure revised from 4.66% to 3.80% (–0.86 points or 19%) between
of economic growth is the evolution of GDP at constant prices. 2012 and 2015. For the country sample, the median growth
The current slowdown has led to significant downward revision rate moved from 4% to 3.6% (–0.4 points or 10%). The most
of projections for future growth at national and global levels. recent WEO projections available in November 2017 involve a
Since we consider no change in population dynamics, an in- slight further decline in these levels, but the additional change is
crease of 1 percentage point in real GDP implies an identical small relative to the changes considered in this article; the world
change in GDP per capita. Two factors are important to keep average growth rate projected for 2017 declined from 4.7%
in mind, especially when looking forward and considering the to 3.8% during the period of our analysis, while the October
2030 targets (for the SDGs): 2017 projections involve a further decline of 0.2 percentage
points. Most importantly, the growth projections for developing
r The current projections involve lower growth rates for the countries—both for total growth and for per capita growth—
future (for example, 2016–2017 growth rates). remain substantially higher than for the industrial countries,
r The actual starting point is lower today (that is, the 2015 which has important implications both for poverty reduction
GDP level) than expected a few years ago. and for world food markets (Fukase and Martin, 2017).
Fig. 1. Comparison of 2012 and 2015 country GDP growth projections for 2017. [Color figure can be viewed at wileyonlinelibrary.com]
D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338 327
Design experiments that assess the implicaons of these changes for the world
economy and the prices that affect poor households
Global CGE simulaons: Idenfy implicaons for the poor in different regions of the
different scenarios capturing different levels of price transmission, macroeconomic
adjustment, and global/regional pecuniary externalies
Household modeling: Simulate the implicaons for households of the prices that they
face and of the technological changes from which they benefit
Results: Global, regional, and naonal impacts on real income, poverty, food indicators,
and macro-nutrient consumpon
Source: Authors.
In these projections, not all countries are affected in the same starting from the macroeconomic scenario definitions (based on
way; exporters of raw commodities and crude oil are among the IMF outlook), their translation in the structural CGE model,
the most affected (Persian Gulf countries, as well as Angola, and the poverty and farm analysis.
Equatorial Guinea, and Democratic Republic of Congo), as are
some key emerging economies such as Brazil and China, with 3.1. MIRAGRODEP: a global CGE model
a decline from 8.5% to 6.3%. China plays a very critical role
due to its size, its growth, and its impacts on world markets. This assessment utilizes a recursive dynamic multiregion,
In advanced economies, recovery is slow but still expected for multisector CGE model: the MIRAGRODEP model. This
the most open economies (such as Japan). The U.S. case is model (Laborde et al., 2013) is a CGE model based on
also very relevant; its growth rate, among the highest for de- MIRAGE (Modelling International Relations under Applied
veloped economies, is cut by one-third. Because we are using General Equilibrium; Decreux and Valin, 2007). This model
fully-fledged household models rather than a poverty elastic- allows for a detailed and consistent representation of the eco-
ity approach, we can capture some important indirect linkages nomic and trade relations between countries. International eco-
between economic growth rates and poverty. nomic linkages are captured through international trade in
goods and services, as well as through capital flows. A dynamic,
recursive solution is obtained by solving the model sequentially
3. Methodology and moving the equilibrium from one year to another. In our
study, we assume perfect competition in all sectors, which al-
When evaluating poverty outcomes, it is worthwhile to look at lows us to have a detailed geographic and sector decomposition.
both short- and long-run effects. Key short-run impacts will be In each country, a representative consumer maximizes a CES-
on business (especially farming) incomes and commodity prices LES (Constant Elasticity of Substitution – Linear Expenditure
(directly affecting both households’ incomes and their costs of System)1 utility function under a budget constraint to allocate
living). In the longer run, changes in wage rates for unskilled expenditures across goods. The LES system allows for different
workers will have major impacts. To disentangle the short- income elasticities of demand, with those for food typically
and long-term impacts of these changes, the MIRAGRODEP lower than those for manufactured goods and services. Once
global economy-wide model (Laborde et al., 2013) is coupled total consumption of each good has been determined in the top
with country-specific household models, following Ivanic and level, the origin of the goods consumed is determined by another
Martin (2018). This top-down approach, starting from broad CES nested structure, following the Armington assumption.
macroeconomic changes and fine-tuning the analysis through
the identification of key impacts, is essential to be able to com- 1 The CES-LES is a variant of a CES function where minimal consumptions
bine the ambition of this project with the very tight constraints are introduced. It is equivalent to replacing the Cobb-Douglas structure of the
on time and resources. Fig. 2 summarizes our analytical steps, Stone-Geary function (that is, LES) by a CES structure.
328 D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338
On the production side, value added and intermediate goods projections will affect demand and supply on world markets in
are complements under the Leontief hypothesis that interme- heterogeneous ways.
diate inputs are used in fixed proportion to output. Total value We consider three scenarios between 2011 and 2030:
added is represented as a CES function of unskilled labor and a r Scenario 0 (S0), the baseline, is based on the old, “opti-
composite of skilled labor and capital; this allows us to specify
mistic” projections. We consider the growth trajectory from
a lower degree of substitutability between the last two produc-
the WEO 2012. Starting from our base year (data set 2011),
tion factors. In agriculture and mining, production also depends
we implement the growth trajectory from WEO 2012 for
on land and natural resources. New capital is perfectly mobile
2012–2017. Demographic projections are taken from the UN
across sectors, while installed capital is immobile. Skilled la-
(2015), and the change in the population in its terms of its
bor is perfectly mobile across sectors, while unskilled labor is
prime working age population is used to define the growth of
imperfectly mobile between the agriculture sector and nona-
the labor force. Total factor productivity (TFP) is computed
griculture sectors. Investment is savings-driven, and the real
at the country/regional level to match the GDP trajectories.
exchange rate adjusts endogenously such that the current ac-
To build the growth trajectory to 2030, we assume that the
count is constant in terms of world GDP. In some scenarios, we
average annual TFP growth rate projected for the 2015–2017
exogenously decrease (increase) the current account of some
period is maintained between 2017 and 2030.
countries and restore global balance by proportional increases r Scenario 1 (S1) is our “first” change scenario looking at the
(decreases) to the CABs of other countries with initial im-
reduced growth projections. Starting from the base year, we
balances. As is common in modeling long-term productivity
implement the actual growth rate from 2011 to 2014 for all
growth or trade reform, the supply of labor is treated as exoge-
countries. The WEO 2015 projections are used from 2015
nous. While the supply of labor clearly adjusts to changes in
to 2020 for our group of “leading” economies, defined as
wages in complex ways (Blundell and Macurdy, 1998; Keane
high-income countries + Russia + Brazil + China. For this
and Rogerson, 2012), any resulting welfare gains are second-
group, we assume that the average annual TFP growth rate
order once the transitional dynamics have died down, coming
projected between 2020 and 2030 is the average over the
at the cost of reductions in nonmarket activities such as leisure.
2017–2020 period. For all other middle- and low-income
By contrast with MIRAGE and most global general equilib-
countries, we maintain the TFP growth rates computed in S0
rium models, the government is explicitly modeled as distinct
between 2014 and 2030. In addition, we implement structural
from the private agent in MIRAGRODEP. The income of the
reforms in China, leading to a “rebalancing” of its economy
government consists of taxes collected on production, on factors
through an increase in domestic consumption and a decrease
of production, on exports, on imports, on consumption, and on
in domestic savings, leading to a reduction in domestic in-
households’ income. The government is assumed to maximize
vestment but also a reduction in the current account surplus.
a Cobb-Douglas utility function so that spending on each com-
Other changes from S0 include reductions in energy prices,
modity remains a fixed value share of total public expenditure
savings rates in oil-exporting countries, CABs, and natural
on goods and services. The model structure ensures that both
resource endowments for energy products.
domestic and external constraints are respected in each country. r Scenario 2 (S2) includes all elements of S1 but also con-
As in Lakatos et al. (2015), we begin our global macroeco-
siders additional drivers for low- and middle-income coun-
nomic analysis with an assessment of the key macroeconomic
tries other than Russia, Brazil, and China. Particularly, the
adjustments under review. This builds on the analyses of the
TFP growth rates between 2014 and 2030 are adjusted to
macroeconomic situation by the IMF, comparing projections
our new scenario using the 2017–2020 projections for these
made in 2012 with the latest 2015 release. New trends are
countries, as well as for the “driver” economies identified in
identified at the global, regional, and national level. A very im-
Scenario 1.
portant task is to identify in these new scenarios the drivers at
play in the OECD economies, in the BRICs (Brazil, Russia, In- These scenarios attempt to capture, in a very simple way,
dia, and China), and in other low- and middle-income countries. the decline in optimism about the outlook for many countries
The case of China is particularly relevant for poor countries in between 2012 and 2015. Scenario 1 captures the impacts of
Africa, Asia, and Latin America. Our analysis considers the slowing growth in the large “locomotive” economies on poverty
direct impacts of the economic slowdown in China, as well as in the developing would. Scenario 2 broadens the coverage to
the rebalancing of its economy between consumption, invest- include the impacts of slowing growth in the rest of the devel-
ment, and domestic and foreign demand. An important element oping world. Since all the scenarios are based on projections
of this scenario is changes in the opportunities for other coun- beyond the sample period of the WEO projections, none of
tries to take up production of labor-intensive goods that are them should be interpreted as a forecast for the period to 2030.
currently produced in China but that are expected to migrate However, they do allow us to build on the detailed and carefully
to other developing countries. The turning point in 2015 from scrutinized country-level projections prepared for the WEO and
growth to decline in China’s labor force reduces China’s over- to perform a sensitivity analysis on the impacts of a decline in
all growth rate, as well as the country’s ability to continue to growth on achievement of SDG 1, the elimination of global
expand exports of labor-intensive goods. These elements of the poverty by 2030.
D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338 329
Population and active population growth rates for each region expenditures on agricultural products (particularly food), as
follow the UN projections (central scenario). Depending upon well as on household income received from labor markets. We
the change in the population and its age structure, this implies do not form “representative” households using these data, which
increases or decreases in the labor forces of particular countries. necessarily aggregates away a great deal of valuable informa-
The case of China is particularly interesting, with the workforce tion. Rather, we retain and use directly the information available
declining by 4% over the period to 2030. The labor force evolves from household surveys on the income sources and expenditure
over time but is constant across scenarios. patterns in each of our more than 285,000 sample households.
Overall, the main exogenous driver of differences between The variables at the household level evolve with their match-
the scenarios is TFP. It captures changes in both per-unit pro- ing concept in the CGE. For example:
ductivity at a constant utilization rate and in the utilization rate
of productive factors (that is, for the same population, a re-
r Prices for goods and services consumed by the household
duced number of effective working hours or underutilization of evolve as in the CGE.
the physical capital stock). The latter interpretation is particu-
r Wages for all households evolve as in the CGE.
larly relevant when addressing downturns in a macroeconomic
r Nonlabor factor endowments (capital, land, natural re-
cycle; both have the same effects on per capita income out- sources) and prices evolve as in the CGE, on a per capita
comes when considering homogenous changes in utilization basis.
rate across factors and agents.
r Input/output ratios for household business activities are kept
constant.
r Labor productivity for each activity/crop rises with produc-
3.1.1. Impacts at the household level
tivity in the CGE.
The impacts of macroeconomic changes on the incomes of r Input and output prices for household businesses follow the
the poor are perhaps simplest when a decline in productivity
domestic CGE price trajectory.
lowers the efficiency of production within an unincorporated
firm (such as a family farm) owned by a household. Another Our sample of countries is geographically widespread and
relatively simple case arises from a shock that results in a size- includes 76% of the world’s poor people (Ivanic and Martin,
able change in a commodity price for which a poor household is 2018), so it seems reasonable to assume that it can generate
a net seller or net buyer. In the short run, the effect of this shock realistic estimates of the global poverty impacts of changes
on real income can be identified with knowledge of the house- such as the macroeconomic shocks under consideration.
hold’s net seller or net buyer status, although recent evidence To obtain estimates of poverty changes at the global level, and
from household surveys suggests that our knowledge of these for meaningful groupings of countries, we clustered countries
net trade shares may be less reliable than we formerly thought based on four variables: the poverty headcount rate at $1.90 a
(Headey and Martin, 2016). The longer-term impact incorpo- day, the rural population share, adjusted net national income per
rates impacts of productivity changes on the real wages obtained capita (constant 2005 USD), and cereal yields (kg per hectare).
by household members from their sales of labor (and other fac- If the poverty headcount at $1.90 per day was not available,
tors) sold outside any family-owned firm. Major changes in we used the previous $1.25 at 2005 PPP per day measure or,
income in different countries may also be expected to result in if this was not available, the poverty rate of 2–3 countries with
changes in remittances from family members working abroad,2 a similar level of GDP per capita. The needed measures were
which are projected to change in line with changes in unskilled obtained from the World Bank’s PovcalNet database and World
labor earnings in the source countries (Laborde and Martin, Development Indicators. With these measures, we used a par-
2016). Thus, macroeconomic shocks can be expected to affect titioning method, assigning each country to the group with the
the real income of households through different channels with nearest mean values while maximizing differences between the
macro- and microeconomic implications. country clusters.3
Assessing impacts at the household level for both rural and Using this method, five country clusters were created, making
urban households is much more complicated than assessing sure that each group contained at least one of the countries in
aggregate impacts since household impacts depend not only the Ivanic and Martin (2018) household data set. We then used
on the characteristics of the household as a producer but also the change in the poverty rate for the 31 countries and applied it
on the cost of living and on household members’ participa- to the remaining countries in their respective groups, allowing
tion in labor markets. Fortunately, a vast amount of work has us to generate estimates of changes in worldwide poverty rates.
been done by Ivanic and Martin (2014, 2018) and by Bouet
et al. (2011) to collect data on income from agriculture and on
3.2. Data sets
2 Information on remittance income by household is obtained from house-
The main source of data for the economy-wide projections
hold survey data and is assumed to be sourced from origin countries in is the GTAP 9 database. For the household simulations, we
line with the bilateral remittance flow data provided by the World Bank
http://www.worldbank.org/en/topic/migrationremittancesdiasporaissues/brief/
migration-remittances-data. 3 Specifically, we used the k means option in STATA.
330 D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338
rely on the data set developed by Ivanic and Martin (2018), to 5.0% for developing countries as a whole, from 8.1% to 6%
which includes 31 countries and represents about 80% of the in China, from 4.4% to 2% in Brazil, and from 4.1% to 1.4%
world’s poor. This data set covers large emerging economies, in Russia. Growth remains slow in many advanced economies
including exporters and importers of food products (for ex- (European Free Trade Association (EFTA) is about 1.7%) and
ample, Brazil, China, India, Indonesia, Nigeria, Pakistan, and is reduced by around one-third in some of the more dynamic
Cote d’Ivoire), and smaller or more vulnerable economies (for OECD economies (for example, the United States, Canada,
example, Tanzania, Uganda, Rwanda, Zambia, Malawi, Niger, Japan, and Korea). Middle-income countries are less impacted
Yemen, Albania, Moldova, Georgia, Bangladesh, Nepal, Mon- (for example, a reduction by one-tenth in the growth rate for
golia, Vietnam, Cambodia, Peru, Ecuador, Guatemala, Belize, Morocco). For some West and East African countries, the more
Nicaragua, and Costa Rica). The consolidated data set in- recent projections are more optimistic than those of three years
cludes more than 285,000 sample households and provides a ago and more than compensate (S2) for the global slowdown
basis for assessing the global impacts of price and productiv- (S1). When we consider the implications of the slowdown in
ity changes on poverty. Details of the regional and commod- the leading economies (comparing S1 to S0), we find that the
ity aggregates used in the study are presented in Appendix negative impacts are more pronounced in West Africa (where
Tables A1 and A2. the annual growth rate declines from 7.1% to 5.7%) and South
Asia than in East Africa and India, clearly showing different
3.3. Baseline projections based on 2012 estimates trade and investment linkages and regional dynamics.
Another important proxy for the impact of these alternative
The simplest indicator of our different growth trajectories is growth paths on the poor is the evolution of the real wage for
the real GDP growth rate between 2011 and 2030, measured as unskilled labor, a major source of income for poor households.
nominal GDP deflated by the price of absorption, as shown in Results displayed in Fig. 4 show the increase between 2011
Fig. 3. For the OECD countries and Brazil, China, and Russia, and 2030. In our initial scenario (S0), the projections are quite
the changes in productivity growth rates under S1 and S2 are the optimistic, ranging from +57% in East Africa to +349% in
same by design. For other developing countries, the S1 results China. The most rapid growth was expected to take place in
reflect changes in real income due to changes in their terms Asia (South and East), Brazil, and West Africa, with a doubling
of trade as larger economies slow down, while the S2 results or more in the real incomes of unskilled workers. Less opti-
add the effects of changes in their own productivity growth. mistic scenarios display highly heterogeneous outcomes, with
The differences between scenarios S0 and S2 illustrate the very countries strongly impacted but still having impressive perfor-
substantial slowdown faced by emerging economies: from 6.5% mance (for example, China), weakly affected but still seeing
9.0
Average annual growth rate, %
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0.0
MENA
EFTA
Central Africa
HI Asia
CIS
Russia
West Africa
Brazil
Nigeria
East Africa
South Africa
Oceania
South Asia
India
China
Morocco
Mexico
World
USA-Canada
rest of LAC
rest of SACU
Ghana
MI East Asia
Central America
Gulf Countries
S0 S1 S2
.
Source: MIRAGRODEP model projections.
Fig. 3. Average annual growth rates for real GDP, 2011–2030. [Color figure can be viewed at wileyonlinelibrary.com]
D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338 331
350
300
250
150
100
50
Oceania
Mexico
Brazil
EFTA
Russia
MENA
rest of LAC
Nigeria
West Africa
East Africa
Central Africa
South Asia
South Africa
India
Central America
CIS
rest of SACU
Morocco
HI Asia
China
World
Ghana
MI East Asia
USA-Canada
Gulf Countries
Fig. 4. Real unskilled wage rates, % change from 2011 to 2030. [Color figure can be viewed at wileyonlinelibrary.com]
When agricultural productivity rises in line with productivity we will see, these price effects can be offset by the consequences
in other sectors, the increase in national income results in a of declines in productivity for the incomes of the poor.
smaller-than-proportional increase in demand for agricultural
products simply because of Engel’s Law. With equal produc- 4.2. Impacts on incomes
tivity growth across sectors, higher income results in a decline
in the relative price of agricultural products. Thus, a slowdown The key variables presented in Table 2 show a wide range
in the rate of growth results in a modest increase in agricultural of responses to the slowdown. In the countries most seriously
prices relative to the S0 baseline, the effects being stronger affected by the slowdown, real incomes decline considerably
for staple foods (which have low income elasticities relative to relative to the outcome under the more optimistic, earlier sce-
those of high-value products such as meat and dairy products). nario represented by S0. In Brazil, for instance, the decline of
Based on the results of Ivanic and Martin (2014) and Jacoby 2.4 percentage points in the growth rate turns into a shortfall
(2016), the increase in agricultural prices is likely to have a of 35% relative to the income projected under S0. Under S1,
long-run positive impact on real wages and a favorable effect output declines sharply in all of the leading economies such as
on poverty reduction. Brazil, China, Russia, the United States, and Canada. The im-
The prices of manufactured goods rise slightly relative to pacts of this scenario on other economies tend to be small and to
the numeraire. The slowdown in growth reduces the demand depend upon whether those countries are suppliers of minerals
for these goods more than for agricultural products, tending or agriculture; suppliers of minerals, such as Morocco, tend to
to push their prices up more than for agricultural products; lose, while suppliers of agricultural products tend to gain.
however, the decline in the prices of minerals and fuels helps Regions depending on global demand and strongly oriented
to hold the price of manufactured goods down. The average to OECD markets are particularly negatively affected by these
price of services rises under the lower productivity scenario for pecuniary externalities. These include West Africa WAEMU
a similar reason. (West African Economic and Monetary Union) countries
Therefore, we see that the global commodity price effects of (–18% of real income compared to the baseline in 2030), Ghana
the economic slowdown captured in S1 and S2 have contrasting (–8%), South Asia (–18%, except India, which is very resilient
outcomes for the poor. Agricultural prices are more resilient to to the external shock), and Central America (–9%). It is im-
the crisis and will generate higher incomes for farmers, but portant to underline that these countries are also affected by a
they also mean higher costs for food consumers. However, as reduction in availability of global savings to fund their growth.
Table 2
Key macroeconomic results, % change from baseline
Country/Region Real income Agriculture VA Unskilled wages Rural unskilled wages
S1 S2 S1 S2 S1 S2 S1 S2
Under S2, real incomes fall relative to the baseline scenario in ing the opportunity for poverty reduction in these areas. Other
a wider range of countries where growth prospects have been poor countries are not as strongly impacted, and some even
marked down, such as Ghana (–23%), Nigeria (–26%), South experience higher real wages (for example, Southeast Asia and
Africa (–21%), and South America. Economies such as those in the Southern African Customs Union/SACU countries).
East Africa, where prospects have been upgraded, experience an In rural areas, the same pattern appears but with generally
increase in real income, while India sees essentially no change better outcomes: losses relative to the baseline are smaller
in income growth rates. For most of the regions already nega- (–7.4% in West Africa, –13% in South Asia other than India,
tively affected, the outcome is worse (except for West Africa, and –3.2% in Central America) or substantive gains emerge
which benefits from updated productivity gains), especially for (+7% in SACU, +2.5% in Ghana, +5.4% in South Amer-
South American countries (up to –38% of projected real in- ica, +2.4% in MENA, and +10% in Central Africa). When
come) and the MENA (Middle East and North Africa) region the updated productivity projections are applied to all coun-
(–18.5%). tries, these positive effects tend to vanish (except in WAEMU
Under S1, real agricultural value added falls relative to countries, South Asia, and East Africa, where productivity is
the baseline in countries directly affected by the shock (lead- updated upward). Real rural unskilled wages are cut by more
ing economies), despite the rise in real agricultural prices. In than 10% in many middle-income economies/regions such as
countries not directly affected by the productivity shock, how- Ghana, Nigeria, South Africa, and MENA and by up to 30% in
ever, agricultural prices and incomes rise modestly. For these South America.
economies, the situation depends on their exact specialization. Remittances from people working in other countries are an
If they mainly sell products to advanced economies and are not important source of income for many poor households in devel-
in competition with products originating in these countries (for oping countries and tend to respond heavily to changes in wage
example, cocoa), they are net losers, with shrinking markets income in the originating countries. As a consequence, inbound
and stable or declining prices. On the other side, if they pro- remittances to Mexico decline relative to the baseline projec-
duce similar products to the rich countries, they benefit from tion by almost 18% under S1, even though real domestic in-
the deceleration of productivity in developed economies that come rises by 3.1% under this scenario. For smaller economies
pushes up prices for their exports. Gains in agricultural value more dependent on this income source in Central America, real
added relative to the baseline are substantial in middle-income inbound remittances fall by 16% in S1, affecting families de-
countries such as Ghana (+4.4%), South Africa (+6.8%), and pendent on them and contributing to the average fall in income
Mexico (+4.8%), as well as in Southeast Asian (+4.8%) and nationwide (–8.8%).
South American (+5.6%) countries. A few countries, such as For other regions, the fall in remittances becomes very signif-
those in West Africa and South Asia (except India), have net icant (between 10% and 20%) when the productivity reduction
losses in real agricultural value added, but these results are affects developing countries as well (S2). The crisis in South
mainly driven by the overall macroeconomic slowdown. In Africa, Nigeria, and Brazil leads to reduction of 18% in remit-
all cases, agriculture remains more resilient to the crisis, and tances for other SACU countries, 15% for Central Africa, and
its value added is much less affected than national income. 23% for the rest of South America.
For instance, in Central America, real value added in agri-
culture remains stable (–0.2%), while national income falls 4.3. Impacts at the household level
by 8.9%.
Once the (generally) adverse shocks are extended to a broader Fig. 5 shows the estimated global poverty rate in 2030 under
set of countries in S2, real agricultural value added also falls four different scenarios. The first is Base, which shows the
in many of these countries. The productivity effects dominate, poverty rate in the absence of productivity growth but allowing
and the gains are reduced or losses increased (for example, from for differential rates of population growth between countries.
+5.6 to –27% between S1 and S2 for South America, excluding The three other scenarios are our S0, S1, and S2 cases. In
Brazil). Only in regions with upgraded real income growth the Base case, the global poverty rate rises from 13% in 2012
projections are farmers better off relative to the baseline (for to 20% in 2030 because population growth is highest in the
example, West and East Africa). In almost all cases, agriculture poorest countries; this makes clear that the alternative to moving
appears to be more resilient to the slowdown, with much lower forward is not standing still but rather sliding backward. Under
losses than those seen in the overall economy. the three growth scenarios, the real income of each household
The third variable presented in Table 2 is the average un- changes over time in response to changes in household business
skilled wage rate, which is a weighted average of the rural incomes (driven by productivity growth), changes in wage rates
and urban wage rates distinguished in the model. Average real for households selling labor, and changes in the cost of living.
wages follow a similar pattern to real incomes, declining sharply Because our long-run scenarios include substantial economic
when productivity falls. In S1, the average unskilled wage rage growth over the period to 2030, the poverty rates under all of
is strongly negatively impacted in West Africa (–11.6% on aver- these scenarios are substantially below current levels. The first
age for WAEMU countries and –3.6% in Ghana), in South Asia bar in each set of three shows the poverty rate, defined as the
(–16.5%, except India), and in Central America (–9.6%), limit- share of the population below $1.90 a day using the 2011 PPP
334 D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338
35%
30%
25%
20% Base
S0
S1
15%
S2
10%
5%
0%
ALL Farmer Rural
Fig. 5. Global poverty headcount under alternative scenarios by 2030 (%). [Color figure can be viewed at wileyonlinelibrary.com]
definition, under our S0 assumptions. As expected, the poverty sents an increase of 8% from the projection under S0—in other
rate among farmers is higher than the rate among the population words, 34 million people not lifted out of extreme poverty. For
at large. Interestingly, however, the poverty rate is higher for farmers, the increase is larger in absolute terms but slightly
the rural population more broadly than among farmers—with smaller as a share of the initial poverty projection. For the ru-
higher poverty rates among groups such as the landless poor ral group, dominated by farmers, the increase is from 7.15%
outweighing lower poverty rates among higher income groups under the baseline to 7.74%. When we turn to S2, the over-
like teachers living in rural areas. all poverty rate goes up slightly at the whole-country level
(38 million fewer people are lifted out of extreme poverty
compared to the baseline) but declines slightly for the farmer-
4.3.1. Net global poverty impacts headed and rural groups. This decline reflects improvements
As shown in Fig. 5, the decline in poverty under the growth in the growth outlook in many African and South Asian coun-
scenario (S0) is very similar to the 3% decline seen in the tries noted in Fig. 3, which contribute to stronger employment
World Bank (2015) report on the prospects for achieving global growth and poverty reduction in these countries, together with
poverty reduction. Our projection based on the 2012 outlook the modest increases in agricultural prices presented in Table 1.
put poverty in 2030 at 4.79%, whereas the World Bank study It also reflects the near elimination of poverty in many countries
(2015, 17) put its median value based on past growth experience that have substantial numbers of poor people today—poverty
at 5.1%. Examination of our individual country-level results in rates reach below 4% in India and as low as 1% in Bangladesh
3 and the World Bank regional results (2015, 42) indicates that under the baseline scenario.
poverty rates are projected to be very low in all regions except An important feature of Fig. 5, under all scenarios, is the
Africa south of the Sahara and South Asia, and only substantial much higher poverty rates seen in rural areas and among farm-
(above 5%) in Africa south of the Sahara. On one level, this ers. While the overall poverty level falls dramatically by 2030,
scenario is enormously encouraging, implying a two-thirds re- the higher poverty rate in rural areas and among farmers per-
duction in the poverty rate from its level in 2011 (World Bank sists, with poverty rates more than 40% higher in rural areas
2015, 41). On another level, this poverty rate is substantially than overall.
above the 3% level that the Bank study uses as a possible,
realistic criterion for success in achieving SDG 1.
Comparing our scenarios, we see that S1 involves a slowdown 4.4. Sensitivity to sectoral TFP growth rates
in the decline in global poverty relative to the sharp decline un-
der S0, with the poverty headcount falling only to 5.21% rather Thus far in the analysis, we have considered the implications
than to the originally projected 4.79%. While the outcome is of productivity growth rate changes that are uniform across
vastly better than that seen in the absence of growth, it repre- sectors. This assumes that the causes of the slowdown (or, in
D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338 335
not directly affected by productivity changes are likely to be farm firms, as well as the real income changes associated with
affected by the changes in commodity prices associated with changes in real wage rates for labor sold by households and
this productivity slowdown. changes in the prices received and paid for food.
To assess the likely consequences of this growth slowdown We find that the changes in poverty rates associated with
for poverty in 2030, we compare the most recent IMF forecasts this growth slowdown are serious but perhaps not as serious as
for economic growth to 2017 with those from an earlier and might have been expected given the extent of the drop in average
more benign environment—in 2012—and assume that the cha- growth rates. Our projection for the global poverty headcount
nges in projected growth rates from the latter part of each of in 2030 is 4.8% under the original, optimistic scenario and rises
the five-year projection periods are continued to 2030. The to 5.2% under our projection based on more recent forecasts.
resulting changes in the level of GDP are very substantial. They For rural people, the increase is from 7.15% to 7.74%.
point to sharply reduced growth prospects for many developing The effects of the growth slowdown are particularly sharp
countries, particularly relative to the period of more rapid eco- in the poorest countries. In these countries, there are also very
nomic growth in developing countries and of income convergen- substantial gross changes in poverty, with over 5% of the pop-
ce between poor and rich countries experienced since the 1990s. ulation falling into poverty relative to the baseline and another
In this article, we assess the implications for the poor— 3% rising out of poverty, for a net increase in poverty of 2% from
particularly the rural poor, who tend to be the poorest of the the original forecast. When we focus on households headed by
poor—of such a sharp deterioration in the economic outlook. farmers, the most striking change is a substantial increase in
We do this by first projecting the implications of the global poverty in the poorest countries, with over 4% of the farm pop-
growth slowdown at a national level, including reductions in ulation no longer lifted out of absolute poverty as a result of
productivity growth rates and the consequential shifts in eco- this growth slowdown. Examination of the detailed results in
nomic balances and changes in relative prices. Then we use the Discussion Paper version of this article (Laborde and Mar-
household models for almost 300,000 households to assess the tin, 2016) suggest that most of the countries with large numbers
implications of these changes for the poor. These models let us of vulnerable people remaining in 2030 are in Africa south of
assess the impacts of productivity changes within households’ the Sahara and South Asia.
Appendix
Table A1
Regional disaggregation in the MIRAGRODEP CGE
Region code Region label GTAP regions
Source: Authors.
D. Laborde Debucquet, W. Martin/Agricultural Economics 49 (2018) 325–338 337
Table A2 Table A3
Sectoral disaggregation in the MIRAGRODEP CGE Continued
Sector code Sector label GTAP sectors ISO Country Poverty rate, farmer population
Laborde, D., Robichaud, V., Tokgoz, S., 2013. MIRAGRODEP 1.0: Documen- OECD, 2017. GDP Long term forecast. Accessed November 15, 2017, available
tation. AGRODEP Technical Note, Washington DC, IFPRI. at https://data.oecd.org/gdp/gdp-long-term-forecast.htm.
Lakatos, C., Maliszewska, M., Osorio-Rodarte, M., Go, D., 2015. China’s Ravallion, M., Datt, G., 1996. How important to India’s poor is the sectoral
Slowdown and Rebalancing: Potential Growth and Poverty Impacts on Sub- composition of output? World Bank Econ. Rev. 10(1), 1–25.
Saharan Africa. Policy Research Working Paper 7666, World Bank, Wash- UN (United Nations), 2015. World population prospects: The 2015
ington, DC. revision. United Nations, New York. Accessed September 23,
Loayza, N., Raddatz, C., 2010. The composition of growth matters for poverty 2016, available at https://esa.un.org/unpd/wpp/Download/Standard/
alleviation. J. Dev. Econ. 93, 137–151. Population/.
OECD (Organization for Economic Cooperation and Development), 2016. World Bank, 2015. A measured approach to ending poverty and boosting shared
GDP long term forecast. Accessed April 17, 2016, available at prosperity: Concepts, data and the twin goals. Policy Research Report, World
https://data.oecd.org/gdp/gdp-long-term-forecast.htm. Bank, Washington, DC.