the distributive impact of terms of trade shocks€¦ · the distributive impact of terms of trade...
TRANSCRIPT
Policy Research Working Paper 8016
The Distributive Impact of Terms of Trade ShocksMaurizio BussoloPatrizia Luongo
Europe and Central Asia RegionOffice of the Chief EconomistMarch 2017
WPS8016P
ublic
Dis
clos
ure
Aut
horiz
edP
ublic
Dis
clos
ure
Aut
horiz
edP
ublic
Dis
clos
ure
Aut
horiz
edP
ublic
Dis
clos
ure
Aut
horiz
ed
Produced by the Research Support Team
Abstract
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Policy Research Working Paper 8016
This paper is a product of the Office of the Chief Economist, Europe and Central Asia Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at [email protected].
The halving of oil prices, which happened in a short period between late 2014 and the first months of 2015, has generated major terms of trade losses for oil export-ing countries. Even if the oil producing sector normally employs a small group of workers and oil export revenues tend to be concentrated in a few firms and in government accounts, these relative price changes have economy-wide effects and significant distributive impacts. This paper describes and quantifies the channels of transmission from the drop in oil prices, to changes in welfare distribution at the household level. Using a macro-micro simulation model, the paper assesses how this shock affects poverty, inequality, and shared prosperity for the case of the Rus-sian Federation. The oil price reduction generates a reverse
Dutch disease that impacts sectoral employment, factor returns, and consumption prices. It causes a contraction of employment and wages in more skill-intensive (non-trad-able) sectors, and a reduction in consumption prices that is more pronounced for nonfood than for food goods. When these shifts are mapped to changes in incomes at the micro level, all households are affected. Poverty rates could increase by 1 to 4 percentage points, depending on the poverty line used. At the US$10 a day threshold, for exam-ple, 4.1 million additional people fall into poverty. Along the consumption distribution, richer people are affected more than those in the bottom 40 percent. However, this minor progressive impact may be reversed due to increases in unemployment and cuts in social programs (transfers).
The Distributive Impact of Terms of Trade Shocks
Maurizio Bussolo
Patrizia Luongo1
Keywords: Terms of Trade, Poverty, Inequality, Shared Prosperity, Macro‐Micro Simulation. JEL code: C68; I32; D63
1 Maurizio Bussolo is with the World Bank: [email protected]; Patrizia Luongo is with the UNDP: [email protected].
2
Introduction
The Russian Federation is a major producer and exporter of natural gas and crude oil. A number of authors
(see Oomes and Kalcheva, 2007 and World Bank, 2005 among others) argued that – following the more
than 300 percent increase in the oil price (from about 25 USD to 110 USD per barrel) between 2002 and
2008 – the economy had been showing symptoms of the Dutch disease. Increased revenues from oil
exports pushed up the real exchange rate and hurt the competitiveness of the tradable sectors;
manufacturing and other exportables shrunk, while services and non‐tradables expanded.2
More recently the Russian economy has been facing the opposite situation; between July and December
2014 the oil price halved from around 110 USD to less than 60 USD per barrel (World Bank, 2015). The
consensus is that the oil price will stabilize at the current lower levels (the average price in 2015 was 50
USD per barrel, and 42 USD in 2016) reflecting structural changes in the global oil markets rather than
cyclical and short term adjustments.
The main objective of this paper is to assess the distributional impact of this terms of trade shock. Recent
macroeconomic data suggest that a reverse Dutch Disease may be at work. In 2015, the exchange rate
depreciated by more than 30% and terms of trade losses were large, equivalent to about 7% of GDP.
Assuming that these income losses will be putting pressure on prices of non‐traded goods and services
and will require related adjustments in the labor markets, this paper tries to identify who are the winners
and losers, and to estimate the changes in income distribution and poverty.
Two difficulties arise in the estimation of the distributive impact. The first is that a household survey for
the period after the shock is not yet available. However, and this is even more important, even if survey
data were available, they may not provide an adequate counterfactual given that, simultaneously to the
oil price drop, the Russian economy has been influenced by other shocks, for example, economic
sanctions. For these reasons, the analysis is performed using a macro‐micro simulation model which
allows to generate a counterfactual scenario in which the oil price change is the only shock affecting the
economy.
2 The economic effects of a natural resource boom were modeled by Corden and Neary (1984) who distinguished between a resource movement effect – related to the reallocation of labor and capital from the tradable sectors to the booming natural resource sector – and a spending effect – which brings a real exchange rate appreciation and further de‐industrialization. The real exchange rate appreciation is caused by the fact that higher incomes from natural resources raise demand for both tradable and non‐tradable goods, but while domestic prices of tradables are linked to international prices and thus remain constant, domestic prices of non‐tradables go up.
3
Overall, the results show that the oil price drop reduces the welfare of the bottom 40 percent by 6.5%. It
also generates greater levels of poverty, whose magnitude depends on the poverty line considered, and
a small reduction of inequality, explained by the relatively higher welfare loss for the top 60. These
distributive impacts result from the transmission of the shock through two main channels. Through the
labor market, the terms of trade shock causes a contraction of employment and wages in more skill‐
intensive (non‐tradable) sectors thus reducing welfare proportionally more for individuals in the upper
part of the distribution. And through the goods market, the reduction of consumption prices – more
pronounced for non‐food than for food goods – benefits more the richer households whose share of non‐
food items in total consumption is larger.
This distributional effect can be reversed if unemployment and reduction of government transfers are
considered. Notwithstanding the real depreciation following the oil price drop, labor and other resources
may not move frictionless to the tradable sectors, and thus unemployment may increase. Initial evidence
shows that poorer workers may be more at risk of losing their jobs vis‐à‐vis higher paid ones. Similarly,
the lower parts of the income distribution rely more heavily on government transfers that may be reduced
because of budgetary pressures linked to lower oil royalties. For these reasons, the slightly progressive
impact of the terms of trade shock may be reversed and become regressive.
These model‐based results are convincing because they represent, in reverse, what happened to sectoral
employment and relative prices in the period when the oil price was growing (2003‐2008).
The rest of the paper is organized as follows: after a brief literature review, the following section
introduces the model and data used in the paper; main results are illustrated in the next section, while
conclusions and policy implications are presented in the last section.
Literature Review
Terms of trade and, more in general, the abundance of natural resources, are subjects of a very large
literature. This section briefly summarizes a few strands of this literature that are particularly relevant to
this paper, namely: (i) the importance of natural resources to the overall development (growth) process
in natural resources abundant countries; (ii) the Dutch disease issue and the impact of relative price
changes on income distribution; and (iii) the modeling techniques employed to link terms of trade shocks
and distributional change.
The impact of natural resource abundance on countries’ economic performances has been widely
analyzed in the literature that has identified three channels through which it might affect economic
4
growth. The first one looks at the link between natural resource abundance and the quality and
effectiveness of political institutions. The underlying idea is that abundance of a commodity generates
rents that might favor rent‐seeking behaviors, corruption, conflicts and /or political instability which have
adverse effects on long‐term growth (Duncan, 2006; Guriev et al., 2008; Isham et al., 2003; Lane and
Turnell, 1996, 1999; Manzano and Monaldi, 2008). The second channel focuses on the fact that natural
resource abundance might expose countries to price volatility (Sala‐i‐Martin and Subramanian, 2003).
Commodity price volatility could lead to instability of public spending and of export income as well as
under‐saving of natural resource revenues (Sinnott et al., 2010). These in turn can lower economic growth
by increasing the instability of aggregate output and demand in the short run. Finally, commodity
dependence might affect a country’s economic performance as it increases countries’ risk of the so‐called
Dutch Disease, i.e. an appreciation of the exchange rate in response to positive shocks, which brings to a
contraction of the tradable sectors with potential negative effect on growth. The de‐industrialization
caused by the Dutch Disease, in fact, might hamper a country’s growth perspective because it hurts mainly
the manufacturing sector, which tends to be more competitive and innovative than the non‐tradable ones
(Oomes and Kalcheva, 2000) or because it reduces the ability of the economy to absorb shocks via labor
mobility (Hausmann and Rigobon, 2003).
Besides influencing economic growth perspectives, income distribution is also affected, via similar
channels3 by natural resource abundance (UNCTAD Annual Report on Trade and Development, 2012).
Warr (2008), for example, uses a general equilibrium model and finds that the increase of staple food
prices between 2003 and 2008 increased the incidence of poverty in Thailand, an exporter of agricultural
goods. He shows that this happened because the main gainers of higher food prices were the landowners
and not the poor; the expansion of the agricultural sector, in fact, increased the wage of unskilled workers
but this was outbalanced by the negative effect higher staple food prices had on the poor. Essama‐Nssah
et al. (2008) use a macro‐micro simulation approach to analyze the effect of the rising oil price on an oil
importer country, South Africa, between 2003 and 2006. At the macro level, higher oil prices reduced the
real GDP, caused a depreciation of the exchange rate and affected the labor market differently according
to the sector considered, with fuel intensive sectors experiencing a bigger reduction in employment. The
3 Changes in relative prices, related to the resource effect and the spending effect (Corden and Neary, 1984) of the Dutch disease are the main channels through which distribution is affected and the focus of much of this paper. For an oil exporter country the two effects can be described as follows. Higher oil prices will lead to higher returns to labor and capital in the oil sector and this will cause a movement of the productive factors from the less remunerative manufacturing industries to the expanding oil activities. The higher wages and/or profits generated in the latter lead to a higher aggregate demand that will cause an increase in the prices of non‐tradables while the price of tradables, being determined above, remains unchanged. The combination of the two effects causes a de‐industrialization and an appreciation of the exchange rate.
5
distributive analysis, at micro‐level, shows that the oil price shock worsens both poverty and inequality.
Moreover, they find that, even if the effect was not particularly strong on average, the impact of the shock
varied along the distribution of income: wages and employment declined more across the poorest
segment of the formal labor market and low skilled individuals were hit hard by the shock while the high‐
skilled, on average, benefitted from it.
When analyzing the impact of oil price changes there is an additional component to consider, as it causes
a variation in energy prices and, through these, on the price of food and other commodities. There are
several channels through which energy prices influence the price of other goods as energy is used directly
or indirectly as an input factor in the production of several commodities. Recent empirical evidence shows
that the prices of many commodities respond strongly to energy price variation and that the elasticity of
non‐energy commodity prices with respect to energy price has strengthened over time (Baffes, 2007;
Baffes and Haniotis, 2010; Borensztein and Reinhart, 1994; Gilbert, 1989).
The effect of an oil price change on food price in an important channel through which oil price variation
influences the distribution of welfare and the level of poverty and inequality. Food is an important
component of the consumption basket, especially for low‐income families, so a change in food price will
affect the household’s welfare. Moreover, by influencing returns to agricultural activities, food prices
influence the welfare of households active in this sector as landowners or wage earners. This implies that
an increase in food prices, for example, might have a positive or a negative effect on poor households
depending if they are net producers or net consumers (Deaton, 1999). However, as shown in the
literature, in most cases the majority of poor households are net food buyers (Ravallion, 1990) even where
agriculture is the main economic activity (Christiaensen and Demery, 2007), so that an increase in food
prices worsens the welfare of the poor. The anti‐poor effect of rising food prices has been found at the
country level (Hoelman and Olarreaga, 2007; Wodon and Zaman, 2008) as well as when poor countries
are considered as a group (Ataman Aksoy and Hoekman, 2010). Oil and food price effects are jointly
examined by De Souza Ferreira Filho (2008) and Estrades and Terra (2012). The first analyzed the effect
of the increase in food and oil prices that occurred between 2004 and 2008 on poverty and inequality in
Brazil, a food exporter and oil importer country. His results show that the negative impact on poverty of
higher oil prices outweighs the positive impact of the expansion of the agricultural sector. Estrades and
Terra (2012) consider both the economy‐wide and the distributive effect of terms of trade change
between 2006 and 2008 in Uruguay, an exporter of agricultural goods and a fuel importer. Their results
show that the Uruguayan economy has been positively affected by the increase in food prices but this has
6
been partly offset by the simultaneous increase in the oil price. As a result, poverty increased especially
for extremely poor people. The reason, the author suggested, is that the increase of the price of the
households’ consumption basket overcomes the income rise experienced by the poorest households.
By simulating the impact of the oil price reduction at both the macro and micro levels, this paper will allow
us to map precisely the impact of a terms of trade shock – in this case the drop of the oil price in an oil
exporter country – on income distribution and thus on poverty, inequality and shared prosperity.
Methodology
Evaluating the distributive impact of the oil price shock requires the use of a simulation model for at least
two reasons. First, being a recent phenomenon, there are still no micro data to compare the distributions
of welfare across households before and after the reduction in oil price. Second, even if data were
available, they might include other shocks4; hence, to identify the distributive effect of the terms of trade
loss, net of (possible) additional shocks, a counterfactual welfare distribution needs to be generated.5
To obtain the counterfactual distribution the effect of the shock needs to be analyzed under two different
perspectives: the macro‐economic perspective, which provides the impact of the terms of trade variation
at the aggregate level; and the micro‐economic perspective, which is required to take into account the
heterogeneous effect across households. The macro‐micro simulation model used here follows a top‐
down (from macro to micro‐level) approach, where the macro and micro part of the model are handled
separately and then linked through changes in prices and quantity computed with the CGE (Lokshin and
Ravallion, 2008; Chen and Ravallion, 2004). The CGE (macro) and the microsimulation modules are briefly
described below.
The macro CGE model
The CGE model used in this exercise is fairly standard and only a summary of its main features is described
here, as its full detailed documentation can be found in van der Mensbrugghe (2014). Production is
modeled using nested CES (Constant Elasticity of Substitution) functions that combine at various levels,
with different substitution elasticities, intermediates and primary factors. Households’ consumption
demand is derived from maximization of household utility producing a constant‐differences‐in‐elasticity
4 The labor market performances and the income distributions of Russian households, for example, are likely to be affected also by the economic sanctions. 5 This section relies mostly on previous works from Bourguignon and Bussolo (2013) and Bussolo, De Hoyos and Medvedev (2013) and Bussolo et al. (2013).
7
(CDE) demand function.6 International trade is modeled assuming imperfect substitution among goods
originating in different geographical areas. Imports demand results from a CES aggregation function of
domestic and imported goods. Export supply is symmetrically modeled as a Constant Elasticity of
Transformation (CET) function. Producers decide to allocate their output to domestic or foreign markets
responding to relative prices. The assumptions of imperfect substitution and imperfect transformability
grant a certain degree of autonomy of domestic prices with respect to foreign prices and prevent the
model from generating corner solutions.
The labor market specification is an important driver of the distributional results, therefore its
specification calls for some clarification and justification. Two types of labor are distinguished, skilled and
unskilled. These categories are considered imperfectly substitutable inputs in the production process.
Moreover, some degree of factor market segmentation is assumed: skilled workers are perfectly mobile
across sectors, whereas the labor market for the unskilled is divided into agriculture and non‐agriculture
segments.
The labor market segmentation by skill level has become a standard assumption in CGE modeling. The
imperfect substitution in the production process for workers with different skills is likely to persist for the
medium‐term time horizon, as unskilled workers cannot be ‘transformed’ into skilled ones, even with
increased on‐the‐job training.
The assumption that the market for unskilled labor is further segmented into agricultural and non‐
agricultural activities is more controversial. However, econometric analysis indicates that a gap in
remunerations between these two segments remains even after controlling for education, gender,
experience and other variables including cost of living differentials (between rural areas, where
agricultural activities are predominantly located, and urban ones). Some barrier to mobility – land
ownership providing economic security to farmers, specificity to human capital acquired in agriculture, or
others – must exist and hinder equalization of wages across the two segments. In the model, this
segmentation is implemented with some flexibility. Using a Harris‐Todaro specification, a certain number
of unskilled workers migrate from one segment to the other in response to changes of the wage
differentials across the segments.
This rich set‐up allows to capture changes in wages for workers of different education level and employed
in different segments. And since skilled‐unskilled and rural‐urban (or, more precisely, agriculture‐non
6 This is the standard demand function in GTAP model, see www.gtap.agecon.purdue.edu/models/current.asp.
8
agriculture) wage gaps represent important drivers of inequality, this set‐up allows to explain changes in
distributions.
For the goods markets, equilibrium for domestically produced goods sold domestically is assumed through
market clearing prices. And the small country assumption is assumed for export and import prices and
thus they are exogenous, i.e. export levels do not influence international prices and import demand does
not influence (CIF) import prices. For the factor markets, wages equate demand and supply of the various
segments (agriculture and non‐agriculture) of labor markets with the assumption of full employment, i.e.
a vertical labor supply. Capital supply is assumed to be fixed and mobile across sectors and a market
clearing rental rate is calculated by the model. Finally, a sector‐specific factor, representing natural
resources, is employed exclusively in production in the oil and gas sector; in the current version its supply
is sensitive to the international price of oil, therefore its reduction triggers a reduction of the supply of
the natural resource and some contraction of the output of the oil sector.
The version of the CGE model used here has a 2011 base year and relies on the Social Accounting Matrix
for the Russian Federation and on bilateral trade flows from the Global Trade Analysis Project (GTAP) 9
database to calibrate initial parameters.
The scenario of the oil shock is implemented by reducing the exogenous international price of oil by 50
percent and a new equilibrium is calculated.7 The results of the model, in terms of changes between the
initial equilibrium and the oil shock one of: (i) prices (for food and non‐food items), (ii) wages (for the four
labor market segments), (iii) unskilled labor migration from agriculture to non‐agricultural activities, and
(iv) per capita consumption are passed to the household survey data and, in the microsimulation, a new
hypothetical global income distribution is generated for the oil shock equilibrium.
The microsimulation model
The ultimate focus of analysis is household welfare, and household real per capita income is here assumed
as its indicator:
, ≅ (1)
7 The model is solved in a comparative statics approach. This means that the model assumes no changes in the demographic structure of the population nor in its skill endowment. Likewise, the supply of capital and land is assumed fixed. The only variable factor of production is natural resources that, as mentioned in the main text, responds to changes in the international price of oil. The economy is also assumed to operate in full employment.
9
Household per capita income (Yh) can, in turn, be modeled as a sum of household members’ labor
endowments ( , ) rewarded by the market wages ( ), and an exogenous income ( 0) as follows:
∑ , (2)
And the household‐specific price index is for simplicity assumed to depend on the economy‐wide prices
of food ( ) and non‐food ( ) items, weighted by the household consumption shares ( , ) of these
consumption items:
, 1 , (3)
For each household, welfare effects can be approximated by the following expression:
,
, (4)
Equation (4) determines changes in welfare as changes in household income and the household‐specific
price index. In the simulations, the budget shares , are kept fixed, and thus changes in the household
price index depend only on changes of the food and nonfood economy‐wide price indexes. Changes in
household income are solely determined by changes in labor incomes, and these in turn are allowed to
vary as a result of changes in the allocation of workers in the different labor market segments (agriculture
or non‐agriculture sector of occupation), and the returns to skilled and unskilled labor in the different
labor market segments ( ). A new household welfare aggregate is computed by adding the exogenous
household income to the sum of simulated labor incomes for each member of the household (given his
or her skill endowments, and sector of employment) and deflating the new total household income by
the new household‐specific price index.
In terms of welfare distribution, the initial distribution for year t for a population of N households can be
written as:
, … , , , , … , , , (5)
The microsimulation consists of using new values for , , , and – that are calculated by the CGE,
(see end of the sub‐section above)– and equation (4) to compute new households’ real incomes and to
generate a simulated new distribution:
, … , , , , … , , , (5’)
10
Note that all the variables linking the CGE and the microsimulations are at the aggregate levels apart from
the household specific employment bundle , . Given that the model employed here is of a static nature,
the total amounts of skilled and unskilled workers do not change, and neither do the endowments of
these two labor types at the household level. However, because of the shock, employment by sector
changes. The CGE model produces the economy‐wide new allocation of workers across sectors, and the
microsimulation is used to determine which specific worker moves from one sector to the other. More
specifically, at the micro level, workers are reallocated among the agriculture and non‐agriculture sectors
by means of a probit model, where the probability of switching sectors is estimated as a function of several
personal and household characteristics. Workers are allowed to switch between labor market segments
until the CGE‐estimated differences in labor allocations between the benchmark and oil price scenarios
are achieved. For workers who switch, a labor income is imputed on the basis of observable characteristics
and the return of them prevailing in the receiving labor market segment. For example, if a worker joins a
new sector, that worker will be imputed a wage based on his or her observable characteristics such as
age, gender, and education.
The impact of the oil shock can be assessed by comparing the standard inequality and poverty indicators
estimated for both and .
Results
Macro‐simulation: A reverse Dutch Disease
Oil revenues are concentrated in a few companies and in the government accounts and a small percentage
of the total work force; less than 2 percent in the case of the Russian Federation, is employed in the oil
sector. However, the fall of the international price of oil deeply affects the structure of the Russian
economy and has a pervasive impact on all households through several channels.8
The first channel is represented by the terms of trade loss. Because imports have become more expensive,
the cost of living rises and reductions of real income and consumption are much larger than contraction
of production (GDP) and employment. The income loss and reduction of demand triggers a series of
additional effects in factors and goods markets in line with a reverse Dutch disease. Economic activity
8 It actually has an impact far beyond its borders as it is affecting neighboring countries through a reduction in the value of the remittances they are receiving. See for example World Bank (2016) where the case of the Kyrgyz Republic, a remittance dependent country with most of its migrant workers employed in Russia, is examined.
11
shifts out of non‐tradables. This can create unemployment, and even if new jobs are created in tradable
sectors, relative wages will likely change. Asset prices, including real estate prices, will likely drop sharply.
Due to the overall reduction and shift of final demand, relative prices of consumption goods will also change:
prices of non‐tradables (for example of services) will contract more severely than prices of tradables. Finally,
a fiscal channel may also be at work as transfers from the government, which is facing shrinking oil
revenues, may be under pressure.
In the CGE model, the terms of trade shock is implemented as a 50 percent reduction of international oil
prices9 and, as shown in Table 1, this causes a reduction of consumption per capita of about 7 percent.
This is a one‐off large welfare reduction that materializes even if production and employment do not
contract. In fact, the model simulates the impact of the oil shock as moving to a new long term equilibrium
with full employment of labor and physical capital. Gross domestic production marginally contracts
because of a reduction in the supply of natural resource in the oil sector, but employment does not go
down. It may be possible that factor markets do not adjust to this new equilibrium in the short run and
unemployment can rise with additional negative impacts for GDP. However, this exercise aims at modeling
the impact of the shock as if it were permanent and after the economy has fully adjusted to the new
relative prices.10
9 The 50% reduction was the one estimated at the end of 2014. During 2015 the price of oil dropped even more, reaching 36 USD per barrel in December 2015 (World Bank Commodity price data). 10 Further below the assumption of full employment is (partially) relaxed and the distributional consequences of an increase in unemployment are assessed.
12
Table 1: The aggregate impact of the oil shock
Percent change with respect to
no shock
Consumption per capita ‐6.9
GDP per capita ‐1.1
Labor demand Unskilled (Agriculture) 6.9
Unskilled (No‐Agriculture) ‐0.6
Wages
Unskilled (Agriculture) ‐3.9
Unskilled (No‐ Agriculture) ‐10.6
Skilled ‐11.5
Skill Premium ‐1.5
Urban Premium (for unskilled) ‐7.0
Consumer Price Index Food ‐7.8
Other goods ‐12.7 Source: GTAP and ROSSTAT data, and CGE simulation results
Following the reduction of demand, prices of non‐tradables decrease making tradables relatively more
expensive. This real exchange rate depreciation induces shifts in economic activity and employment that
are quite different across sectors. Table 2 shows that, in terms of output contraction, besides the oil
sector, the hardest hit sectors are non‐tradables, namely construction, transport and communication,
other business services and public services. While import competing industries, and export oriented
sectors, benefitting from the real depreciation, experience output expansions. The important caveat of
the model operating as if the economy adjusts to its new long term equilibrium without frictions needs to
be reiterated here. The gains of the expanding sectors are genuine opportunities, but they may not
necessarily be realized quickly in the real economy. The analytical framework used here to model the oil
shock is signaling what could happen if the markets were fully flexible and resources were allowed to
respond to the new price signals; whereas, in reality, it may not be possible to reallocate resources as
markets are just imperfectly flexible. For example, closure of firms in urban shrinking sectors such as in
construction may be costly, or banks and other financial intermediaries may not easily start lending to
new firms in expanding sectors. Nevertheless, if the relative prices change triggered by the oil shock is, as
it is believed, not a temporary fluctuation but a long‐term shift, not adjusting to these sectorial
reallocations – by, for example, imposing price controls or other restrictions – will be futile and
counterproductive.
13
In terms of employment, heavy job losses are recorded for the construction and transport‐communication
service sectors, about 0.9 and 1.2 percent of total employment, respectively (see Table 2, third column).
These, translated in actual levels of employment, are close to 500,000 and 700,000 jobs, a considerable
loss. However, new opportunities arise in import competing and export oriented sectors. A potential large
gain is highlighted in the manufacturing sector which may create close to 800,000 jobs.
Table 2: Oil shock triggers sectoral adjustments in favor of tradables and against non‐tradables
Production (% change with
respect to no shock)
Employment (% change with
respect to no shock)
Employment (change with respect to no shock, as % of
tot employment)
Factor Intensity (number of skilled per 100 unskilled)
Oil ‐13.4 ‐18.9 ‐0.4 211 Construction ‐5.3 ‐9.7 ‐0.9 182 Transport and Trade Services ‐1.3 ‐6.0 ‐1.2 139 Business Serv. & Publ. Admin. ‐0.4 ‐0.8 ‐0.4 182 Food products 2.3 3.6 0.1 145 Other agriculture 3.4 6.3 0.3 89 Other mining 5.1 20.4 1.2 145 Other manufacturing 8.2 8.6 0.3 89 Export agriculture 9.5 15.6 0.2 211 Export manufacturing 12.7 23.2 0.7 145
Source: GTAP and ROSSTAT data, and CGE simulation results
Sectors use different inputs in their production, and specifically employ skilled and unskilled11 workers in
different proportions; these reallocations produce imbalances in labor markets. In particular, as on
average skilled workers are employed more intensively in shrinking sectors – such as Business Services
and Public Services, Oil, and even in Construction (where for every 100 unskilled workers, 182 skilled
workers are employed, see rightmost column in Table 2) – job losses affect more severely skilled workers
than unskilled ones. Similar imbalances influence the urban premium, defined here as the ratio of non‐
agriculture activities (essentially urban) wages over agriculture (mainly rural12) wages. These imbalances
in the labor markets put downward pressure on the skill and urban premia which, as shown in Table 1,
decrease by a few percentage points.
Another relevant aggregate result estimated by the CGE model is the change in relative prices of goods
and services. The oil price reduction and the real exchange rate depreciation make energy cheaper,
11 The partition of the population into skilled and skilled workers is based on the highest level of education attained; specifically, unskilled are all those who have completed, at most, secondary education while all the others are considered skilled (even if they did not completed tertiary education). Based on this definition, unskilled workers account for 40 percent of the population and skilled for 60 percent. Among the unskilled, 20 percent has only primary education and 17 percent has junior education; 20 percent have been enrolled in junior vocational education and the highest share (38%) has secondary education. 12 Agricultural activities are mainly (76 percent) located in rural areas.
14
importables more expensive, and non‐tradables less expensive. Energy is an important input in the
production of many final goods, and imports can be, in part, substituted by domestic varieties. The model
results take into account these general equilibrium effects and, even if production of food tends to be
energy intensive, the reduction of prices of other goods is expected to be larger (see last two rows of
Table 1).
The distributive effects of the oil price shock
The oil price shock will affect all households, but the final impact will likely be larger for some groups than
others. In terms of per capita consumption, the poorest bottom 40 percent loses 6.2 percent of its welfare
and the top 60 percent suffers a loss of 7.0 percent (see Table 3). This minor progressive impact, i.e. the
fact that richer people experience welfare losses slightly larger than those for poorer people, amounts to
a marginal change of the Gini coefficient, a standard indicator of inequality, which decreases from 41.06
to 40.86. Worryingly, poverty rates rise for all the usual poverty lines. For example, when measured at the
low 2.5 $ a day line, the oil shock increases poverty by dragging about 100,000 additional people below
that line. At higher poverty lines, such as 5 and 10 $ a day, 900,000 and 4.1 million additional people fall
behind and poverty rates increase to almost 4 and 22 percent, respectively.
Table 3: Welfare, poverty and distributional impact of the oil shock in the Russian Federation
Without Oil
Shock With Oil Shock Change
Per capita Consumption USD in constant 2011 PPP %
All Population 8,971 8,354 ‐6.88
Bottom 40 3,619 3,395 ‐6.20
Top 60 12,539 11,660 ‐7.01
Poverty headcount Headcount ratios (%) % Points
2.5 $ a day line 0.25 0.33 0.07
5 $ a day line 3.14 3.77 0.64
10 $ a day line 19.06 21.98 2.92
Gini coefficient 41.06 40.86 ‐0.20 Source: HBS data and micro‐simulation results. Note: per capita consumption is expressed in 2011 US PPP $; poverty headcounts and Gini in percentages
The impact of the oil shock along the full welfare distribution of the Russian population can be seen in the
growth incidence curve (GIC) of Figure 1. For all percentiles, per capita consumption is measured with and
without the oil price shock and the difference between these two situations is plotted in the graph. The
welfare distribution obtained from the household survey of 2011 is assumed to represent the situation
without the oil shock. The microsimulation model is used to generate the counterfactual distribution with
15
the oil shock. The negative slope of the curve denotes the progressive impact of the shock. However, the
poorest percentiles still lose significantly so that the equalizing effect of this result should not be
overstated.
Figure 1: Growth incidence curve due to the oil shock
Source: HBS data and micro‐simulation results. Note: the points represent the difference in per capita consumption measured with and without the oil shock; the line is obtained as a polynomial fit of the points.
What explains this heterogeneous impact of the shock? Table 1 offers some clues. The skill and urban
premia are decreasing and one may expect skilled workers and (among the unskilled) urban ones to be
towards the richer tail of the distribution. This table also highlights that among the workers displaced by
the shock, about 100,000 unskilled ones will have to move from better paid urban occupations to rural
lower paid jobs. However, the aggregate results of this table do not inform from which part of the
distribution these displaced unskilled workers are more likely to come. Finally, the lower reduction of food
prices vis‐à‐vis prices of other consumption items suggests a potential regressive impact given that poorer
households spend a larger share on food than richer households.13
Other channels of transmission of the shock can also contribute to its uneven incidence. For example,
rents in oil producing and closely related sectors (such as refineries) will be directly hit. Given the large
capital stocks of the oil sector, capital returns may also suffer larger losses vis‐à‐vis labor returns. This shift
13 Note that if rich consumers purchase a larger share of imported goods, they may be impacted more severely than poorer ones by the increase in import prices. Some authors [Broda and Romalis 2008] have studied this issue for the case of the US. They found that lower import prices of goods originating in China have actually helped the poorer consumers in the US. However, the current household data for Russia do not allow to investigate this channel.
Overall Change
‐9
‐8
‐7
‐6
‐5
‐4
0 10 20 30 40 50 60 70 80 90
Chan
ge in
welfare (% chan
ge in
per capita
consumption)
Population percentiles
16
in factorial distribution may translate into a reduction of personal income disparities (and thus a
negatively sloped GIC), as capital incomes may be more concentrated among the rich groups of the
country.14 Asset prices, including real estate prices, will likely drop sharply. Moreover, those who have
borrowed in foreign currency suddenly experience a rise in real debt. Finally, transfers from the
government are not distributionally neutral as some groups, for example retirees, receive proportionally
more than others. A change in transfer amounts, and coverage, caused by budgetary difficulties can thus
change the distribution.
Using the microsimulation model, it is possible to untangle these multiple effects and study their
individual separate impacts. Consider first the adjustments in the labor market. Unskilled workers
displaced from jobs in non‐tradable services are among those who lose the most: their welfare is almost
halved; before the crisis the average consumption for a displaced worker was around 9,700 USD but with
the oil shock it declined to around 6,000 USD.15 Although the impact is clearly large for individual displaced
workers, the consequences for the overall distribution depend on the number of displaced workers and
whether they were initially concentrated in a specific part of the distribution. About 100,000 unskilled
workers, around 0.1 percent of total employed workers, move back to agriculture sectors because of the
shock. These are not randomly selected, as some will have a higher probability of moving than others.
Nevertheless, as shown in Figure 2, they are coming from all parts of the distribution. They tend to be
aged above 40 and, in most cases, males with at most junior secondary or secondary education, but they
are not necessarily poorer.16 For these reasons, the impact on the overall distribution is muted as shown
by the GIC 1, the dashed line in Figure 3.17
14 Note, however, that this effect cannot be currently assessed given the data limitations of the available household survey where income from capital is not accurately measured. 15 Because of the segmentation, two separate wages are clearing the rural (agriculture) and urban (non‐agriculture) segments. The urban premium, i.e. the ratio of the wage of unskilled workers in non‐agriculture sectors over that of unskilled workers in agriculture is initially equal to 1.3. A labor migration function allows for some imperfect mobility across segments and this, together with sectoral demand for labor, determines the shift in the urban premium shown in Table 1. 16 Moving workers are identified using a probit estimation, as explained in the methodological section, see the Appendix for more details. 17 The line ‘GIC 1’ in the figure almost overlaps with the horizontal straight line that intercepts the vertical axis at ‐6.9. This horizontal line represents the average percentage welfare loss, i.e. the loss that everyone would incur were the shock uniformly distributed. The case represented by GIC 1 takes into account the losses of displaced unskilled workers, but the impact is minor, visible only because of the small blips on the GIC 1 line.
17
Figure 2: Unskilled workers displaced by the shock come from all parts of the distribution
Source: HBS 2011 and authors calculations using microsimulations
In addition to displacement of workers, the oil shock indirectly but more significantly affects labor markets
by changing relative returns to human capital and to sector specific skills. While the whole population will
be negatively affected by the oil shock, in terms of incidence those depending on the returns to skills will
lose about 1.5 percent more than the others, and, among the unskilled, those in urban activities will lose
7 percent more than those in expanding rural sectors. Figure 4 shows that the human capital is not
uniformly distributed. Skilled employed workers (mainly in non‐agriculture sectors) represent about 31
percent of the population and the majority of them, about 18 percent, are found in the upper 60 percent
of the distribution. And even within the bottom 40 and the top 60 the share of skilled workers is not
uniform, as shown by the upward sloping line in the figure.18 Conversely, the unskilled workers are
overrepresented at the bottom of the distribution, especially those employed in the agriculture sectors.
In light of the shares shown in Figure 4, the reason for a progressive impact of the reduction of the skill
and the urban premia – illustrated by the downward sloping GIC 2 line in Figure 3 – becomes apparent.
18 On average for every 100 individuals, 9 are skilled among the bottom 40, while 21 are skilled among the top 60. However, these proportions are 5 and 26 for the bottom 10 and top 10 respectively.
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
20,000
2,048 3,262 4,152 5,014 5,986 7,115 8,568 10,686 14,552 28,369
Number of displaced workers
Deciles (identified with their per capita consumption in 2011 rubles)
18
Figure 3: Decomposition of the incidence of the oil shock
Source: HBS 2011 and authors calculations using microsimulations
The progressive effect of the change in wage premia is reduced by another round of unequal adjustments:
the change in relative prices of consumption goods. Not all households consume the same proportions of
food items. As shown in Figure 4, the share of total expenditure devoted to food is much higher for poorer
households: the average Russian household spends 33 percent of its total expenditures on food, while a
household in the bottom 40 percent spends 43 percent, and one in the top 60 only 16 percent. Therefore,
richer households benefit disproportionally more from the stronger reduction of prices of non‐food items,
including non‐tradables. This is reflected by a counterclockwise tilting of the GIC (from GIC 2 to GIC tot in
Figure 3) which makes the overall incidence less progressive.
Robustness and validation
The incidence results discussed up to this point rely on two assumptions: the first is that the shock affects
the allocation of labor across sectors but without generating unemployment; the second is that the
reduction in oil royalties does not influence the level of government expenditure and its household
transfers. The implications of the two assumptions are related, as it can be inferred by looking at the
employment share along the welfare distribution depicted in Figure 4. The intensity of the use of labor,
the main source of income for most households, is not uniformly distributed. Households at the bottom
40 percent record an average employment rate of 45 percent against an average of 53 percent for
GIC tot: GIC 1 + GIC 2 + Δ Goods Prices
GIC 2: GIC 1 + Δ in Returns to Skills
GIC 1: Labor Movement across sectors
‐9
‐8
‐7
‐6
‐5
‐4
0 20 40 60 80
Chan
ge in
welfare (% chan
ge in
per capita
consumption)
Population percentiles
19
households at the top 60 percent. An increase in unemployment and a reduction of social government
programs may hit poorer households harder, given their more intense dependency on transfers and
higher risk of becoming unemployed (or lower employability).
Figure 4: Unequal Distribution of Employment and Assets, such as human capital, and of consumption of food across the Russian Population, 2011
Source: HBS data (2011)
To get some insights on the implications of these assumption we carried out two additional simulations.
In the first one we assume that all those who lose their jobs in the contracting sectors – up to 3.1 percent
of total employment – become unemployed. In the second one we analyze the implication of a cut in
government transfers.
Abstracting from multiplier effects, 19 what would be the distributional impact of the increase in
unemployment? The first step of the simulation was to identify those who are more likely to become
unemployed.20 Not surprisingly, the results of the econometric specification show that less qualified,
19 Clearly, if employment goes down (or equivalently if unemployment increases), GDP will also go down and that would trigger a reduction of incomes, then consumption will decrease with another round of reductions. This negative multiplier effect is not accounted here. 20 Specifically, workers more likely to be unemployed are selected through a multinomial probit estimation that computes the probability of either being employed in tradable sectors, employed in nontradable sectors, or being unemployed according to individual and household characteristics. These characteristics include gender, education level, age, marital status, household size, living in urban or rural areas, and headship.
Unskilled in Non Agri
Total Occupation
Unskilled in Agri
Skilled in No Agri
Skilled in Agri
Share of food in total consumption (right axis)
0
0.1
0.2
0.3
0.4
0.5
0.6
0
0.1
0.2
0.3
0.4
0.5
0.6
0 10 20 30 40 50 60 70 80 90
Share of food
Percent share of total population
Population percentiles
20
younger and female workers are more likely to lose their job. However, exposure to unemployment is not
concentrated in particular income groups, as these individuals are found across the whole welfare
distribution (see Figure 5).
Figure 5: Unemployed workers by decile of the welfare distribution
This result suggests that the ultimate distributive impact of the unemployment increase would depend
heavily on the social protection programs available to those who lost their job and, of course, to the whole
population. However, it is unlikely that these programs and, more generally, public transfers and
expenditure will not suffer from the shrinking oil royalties following the drop in oil price. The oil price
shock, indeed, puts government expenditure under pressure and its reduction is likely to have affected
differently individuals located in different parts of the welfare distribution.
Taking advantage of the availability of data on public transfers in a different survey, the RLMS (Russian
Longitudinal Monitoring Survey), our second simulation estimates the effect of a 10 percent cut in public
transfers. The impact on poverty and shared prosperity depends on the incidence of the transfers along
the welfare distribution: if poor households depend relatively more than the richer ones on public
transfers, the final effect of the oil shock could be reversed once the cut of transfers is taken into account.
Most of the beneficiaries of public transfers21 and pensions are indeed located in the first decile of the
consumption distribution (see Figure 6) while the share is much smaller for households located in the top
30 percent of the distribution.
21 Transfers are defined as the sum of children, rent, utilities, fuel and unemployment benefits.
0
2
4
6
8
10
12
14
1 2 3 4 5 6 7 8 9 10
New
ly unem
ployed as a share of total
population
Household income decile
21
Figure 6: Share of transfers by decile of the consumption distribution
Source: RLMS (2011)
Due to the different incidence of public transfers for households located in bottom 40 and top 60 of the
income distribution, the “progressive” effect of the oil price shock shown above is reversed as poor
families lose more (‐7.13) than the top 60 of the income distribution (‐6.82). (See Figure 7.)
0.00
5.00
10.00
15.00
20.00
25.00
1 2 3 4 5 6 7 8 9 10
Transfer as share of
consumption
Consumption decile
22
Figure 7: Growth incidence curve due to the oil shock and the reduction of public transfers
Source: RLMS (2011) and author calculations using microsimulations
Validation
A full backward study of the Russian economy during the oil price boom is beyond the scope of this paper
but a preliminary descriptive analysis of the labor market trends over that period helps to validate the
results of our simulation. In fact, if our inferences on the effect of the oil price drop are correct, we expect
to observe the opposite effect when the oil prices are rising: increases of employment in non‐tradable
sectors, increasing wage premia and, possibly, larger economic gains in the upper part of the income
distribution which would imply higher inequality.
By using the ROSSTAT data on national accounts and the information on consumption available in the
HBS, below we look at the changes that occurred in employment, value added and the skill premia
between 2003 and 2008. This period was characterized by a substantial increase in the oil price, which
jumped from around 35 USD per barrel in 2003 to slightly more than 140 USD per barrel in 2008 (see
Figure 8). The rising oil price was accompanied by an appreciation of the exchange rate (Figure 8) and a
reduction of the contribution of tradable sectors to economic growth (Figure 9).
23
Figure 8: Changes in oil price and exchange rate, 2002 ‐ 2008
Source: World Bank
Looking at employment and added value by sector of activity, we find a first confirmation of the expected
effects: employment and value added decreased in tradable sectors and increased in non‐tradable sectors
(fig. 10). Between 2003 and 2008 the increase in employment was particularly high in the construction
sector (+13.4%) while an almost equal reduction affected the agricultural one (‐13.8%). Similar results
emerge from the analysis of changes in the share of value added, which grew especially in construction
(+23.5%) and services (+16.2%) (Figure 9).
Figure 9: Changes in employment and added value by sector, 2003 ‐ 2008
Source: ROSSTAT
24
During the same time period wages grew in the whole economy but the increase was significantly higher
in non‐agricultural sectors than in the agricultural one, where it increased by the 41%. As a result, the
wage premia, computed as the ratio between the compensation per worker in non‐agricultural and
agricultural sectors, rose from 3.6 in 2003 to 4.7 in 2008 (Figure 10).
Figure 10: Changes in wage premia, 2003 ‐ 2008
Source: ROSSTAT
Testing the validity of our simulations for the skill premia is a bit more difficult, as during the 2003‐2008
period supply and not only demand for these two types of workers were changing. On the demand side,
the growth of non‐tradable sectors could have pushed up the skill premia as these sectors are more skill
intensive (see table 1). However, on the supply side two things might have outweighed this effect. First,
between 2003 and 2008 the share of skilled workers grew from 21% to 30%. Second, unskilled workers
might have moved to better paid sectors as a consequence of the contraction in the agricultural sector.
The small increase in the skill premia depicted in Figure 11 can be interpreted as a sign that, despite these
supply shifts, which would have pushed the skill premia downwards, the demand side effect dominated.
25
Figure 11: Changes in skill premia, 2003 ‐ 2008
Source: HBS
Conclusions
The oil price shock will affect the inequality dynamics between and within countries. For oil importing
countries lower oil prices are likely to affect positively households’ and corporates’ incomes, while the
opposite holds for oil exporter countries. This could cause a shift in the income distribution from the latter
to the former, affecting inequality between countries, but also a reallocation of capital and labor returns
whose net distributive impacts are not easily predictable. Moreover, even if overall this phenomenon
could have a positive impact, by increasing the global GDP and reducing the global inflation (Baffes et al.,
2015), these results will be effective in the medium‐long run, while the effects on the oil exporter
countries tend to occur in the short run.
Considering that the strong growth experienced by the Russian economy during the early 2000s was
accompanied by all the symptoms of the so‐called Dutch Disease, and that the current shock is likely to
produce a structural economic change, we wondered if the effect of current oil price shock would be the
reversal of the one observed during the oil price boom and what would be its effect in terms of poverty,
inequality and shared prosperity.
The overall effect of the oil price shack is “decomposed” in two parts, the distributive effect ‐ which
operates through the variation in consumption prices and a restructuring of the labor market – and the
“level” effect – that is the reduction in average consumption generated by the loss in terms of trade. The
two components are estimated by using a macro‐micro simulation model and, following the literature,
26
the overall effect on households’ welfare is computed by the money metric variation of welfare due to
the prices and wage changes.
The distributive analysis shows that the two channels (the consumption prices and the labor market
changes) work in the opposite direction. The food prices decrease less than non‐food prices, increasing
more the welfare of better‐off households. On the other side, a lower skill premia and a reallocation of
the labor force across sectors generates a bigger loss, in terms of wage premia, for the skilled workers
usually located in the upper part of the welfare distribution. Together with the GDP contraction, the loss
in terms of trade will cause a considerable decline in the average consumption per capita but, as suggested
by the results on inequality and shared prosperity, individuals in the upper part of the distribution are
those who will be more strongly hit by these negative changes.
However, this distributional effect can be reversed if one takes into account employment losses and
reduction of government transfers. The real exchange rate depreciation generates the incentives for labor
and other resources to move from non‐tradables to tradables sectors, but inter‐sectoral resource
movements may not be as smooth. There may be frictions (due to imperfections in financial markets, for
example) and other adjustment costs (a worker in a non‐tradable sector may have sector specific skills
and thus not able to be immediately productive in a tradable sector) which may create unemployment.
Or to decrease the pressure on its budget, the government may reduce its transfers. These two additional
effects would impact the lower parts of the income distribution – which rely more heavily on labor
incomes or transfers – more severely than the richer deciles.
27
References
Aksoy, A. M.; A. Isik‐Dikmelik. (2008). Are Low Food Prices Pro‐Poor? Net Food Buyers and Sellers in Low‐Income
Countries. World Bank Working Paper No. 4642
Ataman Aksoy, M., and Hoekman, B.M. (2010) Food prices and rural poverty. Washington, DC: The World Bank.
Baffes, J. (2007). Oil spills on other commodities. Resources Policy 32, 126‐134
Baffes, J., Haniotis, T. (2010). Placing the recent commodity boom into perspective. In Ataman Aksoy, M., and
Hoekman, B.M. Food prices and rural poverty. Washington, DC: The World Bank.
Baffes, J.; Ayahan Kose, M.; Ohnsorge, F.; Stocker, M. (2015), “The Great Plunge in Oil Prices: Causes, Consequences
and Policy Responses”, World Bank Policy Research Note n.1
Barnes, D.; Floor W. (1996), “Rural energy in developing countries: a challenge for economic development.” Annual
Review of Energy and the Environment, 21, pp. 497–530
Benjamin N. C.; Devarajan, S.; Weiner, R. J. (1989), “The ‘Dutch’ disease in a developing country: Oil reserves in
Cameroon”, Journal of Development Economics, 30(1), 71‐92
Borenzstein, E., Reinhart, C.M. (1994) The macroeconomic determinant of commodity prices. IMF Staff Papers 42,
236‐261.
Bourguignon, François (2004): “The Poverty‐Growth‐Inequality Triangle”, Indian Council for Research on
International Economic Relations Working Paper #125.
Bourguignon F, Bussolo M and Pereira da Silva L (Eds.) (2008) The Impact of Macroeconomic Policies on Poverty and
Income Distribution ‐ Macro‐Micro Evaluation Techniques and Tools, Washington DC: The World Bank and Palgrave
Macmillan.
Bourguignon, F., Bussolo, M., 2013. Income Distribution in Computable General Equilibrium Modeling. In: Dixon,
P.B., Jorgenson, D.W. (Eds.), Handbook of Computable General Equilibrium Modeling. North Holland, Elsevier B.V.,
pp. 1383–1437.
Bussolo, M.; De Hoyos, R. E.; Medveded, D. (2010), “Economic Growth and Income Distribution: Linking
Macroeconomic Models with Household Survey Data at the Global Level”, Internationa Journal of
Microsimulation, (3)1, 92‐103
Bussolo M, De Hoyos R and Medvedev D (2010) Global Poverty and Distributional Impacts: The GIDD Model‘, in
Anderson K, Cockburn J and Martin W (Eds.) Agricultural Price Distortions, Inequality, and Poverty, Washington
D.C.: The World Bank, 87‐119.
Chen, S., Ravallion, M., 2004. Welfare impacts of China's accession to the world trade organization. In: Bhattasali,
D., Li, S., Martin, W. (Eds.), China and the WTO: Accession, Policy Reform, and Poverty Reduction Strategies, Oxford
University Press and the World Bank, Oxford and Washington , DC.
Christiaensen, L., Dmery, L. (2007). Down to earth. Agriculture and poverty reduction in Africa. Washington, DC:
World Bank.
Corden, W. M.; Neary, J. P. (1982), “Booming sector and De‐industrialization in a small open economy”, Economic
Journal 92, 825 – 848
28
Corden, W.M. (1984), “Booming sector and Dutch Disease economics: survey and consolidation”, Oxford Economic
Papers, 36(3), 359‐380
Gylfason, T. and Zoega. G. (2003), “Inequality and economic growth: Do natural resources matter?” In Inequality and
Growth: Theory and Policy Implications, Eicher, T. and Turnovsky, S. (eds.), MIT Press, Cambridge (MA).
Deaton, A. (1999) Commodity prices and growth in Africa. Journal of Economic Perspective 13, 23‐40.
De Souza Ferreira Filho, J.B. (2008), “The World Food Price Increase and Brazil: Opportunity for All?” Paper presented
at the II Regional Meeting on Computable General Equilibrium (CGE) Modeling: Contributions to Economic Policy
in Latin America and the Caribbean. San José, Costa Rica. November, 2008.
Estrades, C., Terra, M. I. (2012), “Commodity prices, trade, and poverty in Uruguay”, Food Policy 37, pp. 58–66.
Gilbert, C.L. (1989). The impact of exchange rates and developing country debt on commodity prices. Economic
Journal 99, 773‐783.
Hoekman, B., Olarreaga, M. (2007). Global trade and poor nations: the poverty impact and policy implications of
liberalization. Washington DC: Brookings Institution.
Isham, J., L. Pritchett, M. Woolcock, and G. Busby (2003), “The Varieties of the Resource Experience: How Natural
Resource Export Structures Affect the Political Economy of Economic Growth,” mimeo, World Bank, Washington
D.C.
Jakob, M. (2010), “Dutch Disease or Botswana’s Blessing? Natural Resources and Economic Growth – A Channel
Approach”, PIK Potsdam, WP
Leamer, E., Maul, H., Rodriguez, S., Schott, P. K. (1999), “Does natural resource abundance increase Latin American
income inequality?”, Journal of Development Economics Vol. 59, 3–42.
Pachauri, S.; Spreng, D. (2004), “Energy Use and Energy Access in Relation to Poverty.” Economic and Political
Weekly, Vol. 39, No. 3, pp. 271‐278
Ravallion, M. (1990) Rural welfare effects of food price change under induced wage responses: theory and evidence
for Bangladesh. Oxfrod Economic Papers 42(3), 574‐585.
Ravallion, M., Lokshin, M., 2005. Winners and Losers from Trade Reform in Morocco, Mimeo, World Bank.
Ross, M. (2007), “How Can Mineral Rich States Reduce Inequality?”, in Sachs, J.D., Stiglitz, J.E. and Humphreys, M.
(Eds.), Escaping the Resource Curse, Columbia University Press, New York.
Ruggeri Laderchi, C.; Olivier, A.; Trimble, o.: (2013) “Balancing Act. Cutting Energy Subsidies While Protecting
Affordability”. Washington, DC: World Bank. doi: 10.1596/978‐0‐8213‐9803‐6 License: Creative Commons
Attribution CC BY 3.0
UNCTAD (TDR 2012), “Trade and Development Report, 2012. Policies for Inclusive and Balanced Growth.” United Nations publication, Sales No. E.12.II.D.6, New York and Geneva
Vagliasindi, M. (2012) “Implementing Energy Subsidy Reforms: An Overview of the Key Issues.” Policy Research
Working Paper 6122, World Bank, Washington, DC
Warr, P. (2008), “World food prices and poverty incidence in a food exporting country: a multihousehold general
equilibrium analysis for Thailand.” Agricultural Economics 39 (Supplement), 525–537.
Wodon Q., Hassan, Z. (2008). Rising food prices in Sub‐Saharan Africa: poverty impact and policy responses. Policy
Research Working Paper 4739, Washington DC: The World Bank
29
World Bank (2005), Russia Economic Report. N. 11, Washington. The World Bank
World Bank (2015), Russia Economic Report. N. 33, Washington. The World Bank
30
Appendix
Table 1: Distribution of initial consumption
Decile All Displaced Non‐ Displaced
Numb. Of obs. Mean Numb. Of obs. Mean Numb. Of obs. Mean
1 14,206,517 1430 16,408 1278 14,190,109 1430
2 14,206,329 2268 9,492 2199 14,196,837 2268
3 14,206,425 2888 3,351 2938 14,203,074 2888
4 14,208,270 3482 6,035 3466 14,202,235 3482
5 14,204,135 4149 1,383 4064 14,202,752 4149
6 14,227,296 4930 3,259 4819 14,224,037 4930
7 14,186,107 5935 2,293 5769 14,183,814 5935
8 14,205,689 7397 4,518 7216 14,201,171 7397
9 14,206,303 9973 1,253 9520 14,205,050 9973
10 14,206,166 19432 2,628 17146 14,203,538 19432
Source: HBS & microsimulation
Table 2: Distribution of consumption before and after the oil shock
Decile Final Distribution (Change in skill premia & change in consumption)
1 2 3 4 5 6 7 8 9 10 Total
Initial D
istribution
1 13,531,
680 674,837 0 0 0 0 0 0 0 0 14,206,
517 95.25 4.75 0 0 0 0 0 0 0 0 10
2 673,414 12,393,
454 1,139,4
61 0 0 0 0 0 0 0 14,206,
329 4.74 87.24 8.02 0 0 0 0 0 0 0 10
3 1,716 1,131,4
88 11,744,
640 1,328,5
81 0 0 0 0 0 0 14,206,
425 0.01 7.96 82.67 9.35 0 0 0 0 0 0 10
4 0 6,057 1,321,5
24 11,498,
668 1,382,0
21 0 0 0 0 0 14,208,
270 0 0.04 9.3 80.93 9.73 0 0 0 0 0 10
5 0 0 3,347 1,373,2
15 11,511,
110 1,316,4
63 0 0 0 0 14,204,
135 0 0 0.02 9.67 81.04 9.27 0 0 0 0 10
6 0 0 0 3,259 1,310,7
18 11,745,
368 1,167,9
51 0 0 0 14,227,
296 0 0 0 0.02 9.21 82.56 8.21 0 0 0 10
7 0 0 0 0 2,293 1,141,9
70 12,093,
049 948,795 0 0 14,186,
107 0 0 0 0 0.02 8.05 85.25 6.69 0 0 10
8 0 0 0 0 309 3,910 942,901 12,623,
029 635,540 0 14,205,
689 0 0 0 0 0 0.03 6.64 88.86 4.47 0 10
9 0 0 0 0 0 0 926 641,484 13,333,
902 229,991 14,206,
303 0 0 0 0 0 0 0.01 4.52 93.86 1.62 10
10 0 0 0 0 0 0 0 0 231,433 13,974,
733 14,206,
166 0 0 0 0 0 0 0 0 1.63 98.37 10 Total
14,206,810
14,205,836
14,208,972
14,203,723
14,206,451
14,207,711
14,204,827
14,213,308
14,200,875
14,204,724
142063237
10 10 10 10 10 10 10 10 10 10 100
Source: HBS & microsimulation
31
Table 3: Poverty rates before and after the oil shock
Distribution of Consumption
Poverty line 2.5 US $ per day
All Non Displaced Displaced Non shrinking Shrinking
Initial 0.554 0.553 3.345 1.401 0.271
Change in Skill Premia 0.642 0.638 10.739 2.142 0.321
Consumption variation 0.713 0.708 13.742 2.342 0.364
Poverty line 5 US $ per day
Initial 5.872 5.866 23.514 13.123 3.893
Change in Skill Premia 5.88 5.866 44.216 13.695 3.893
Consumption variation 6.721 6.706 49.058 16.862 4.553
Poverty line 10 US $ per day
Initial 29.713 29.703 57.503 44.985 23.5
Change in Skill Premia 29.718 29.703 72.128 45.389 23.5
Consumption variation 32.612 32.597 73.42 49.298 26.386
* The macroeconomic shock causes a change in the composition (in terms of total population and individual's characteristics) of the 2 sectors
Source: HBS & microsimulation
Changes in the Labor Market
The results obtained via the macro‐simulation are used as inputs in the micro‐simulation model to obtain
the new allocation of workers across sectors and the new distribution of wages.
The simulation proceeds in subsequent steps. First, unskilled workers are reallocated across sectors
according to the prediction of the CGE. For each individual i, we compute the probability of moving out
the shrinking sector as a function of individual ( ) and household ( characteristics22.
Pr 1 Pr
Individuals are then ranked in a descending way according to the predicted probabilities and moved to
the expanding sectors up to the point where the macro‐prediction on sectoral reallocation of labor is
satisfied.
Once workers are reallocated across sector, the changes in the skill premia23 determine the new
distribution of wage across sectors. The wage each worker receives is computed through a Mincer
22 The probability is computed using a probit model; a list of dependent characteristics is presented in the estimation’s result table in the Appendix. 23 The new skill premia, for skilled and unskilled in non‐farm activities, are computed with respect to the wage of unskilled workers employed in agricultural sector, as the model assumes that the mobility across the agricultural and non‐agricultural segments of the labor market is a function of the changes in the farm‐ and non‐farm wage premia (Bussolo, De Hoyos and Medvedev, 2010).
32
equation that estimates the earnings of in the agricultural and non‐agricultural sector; these wages are
then rescaled to correct for the different distribution of unobservable factors in the two segment of the
labor market (Agricultural and Non‐Agricultural).
, , ∗,
,
The resulting distribution of wage is rescaled to its original mean and then adjusted in order to match the
“level effect” predicted by the CGE:
, , , ,,
, _
With , , representing the individual welfare obtained after all the distributional changes have
taken place and is the percentage change in aggregate average income predicted by the CGE.
Table 4: Results from the estimation of the probability of being employed and probability of being a displaced workers
VARIABLES Displaced Employed
gender 0.0742** 0.482***
(0.0300) (0.0004)
age 0.0651*** ‐0.00187***
(0.00677) (0.0000)
Education Level ‐0.0349* 0.660***
(0.0199) (0.0002)
HH size 0.00841 0.0147***
(0.00999) (0.0002)
Constant ‐3.097*** ‐2.836***
(0.156) (0.00123)
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: HBS & microsimulation
33
Table 5: Returns to endowments in the expanding and shrinking sector
VARIABLES Expanding Shrinking
Household Head 0.0324* 0.1044***
(0.0178) (0.0031)
Gender 0.0906*** 0.0439***
(0.0176) (0.0031)
age ‐0.0249*** ‐0.0207***
(0.0053) (0.0007)
age2 0.0004*** 0.0003***
(0.0001) (0.0000)
Education level 0.1099*** 0.102***
(0.0040) (0.0007)
Constant 7.6232*** 8.0521***
(0.1059) (0.0157)
Observations 8,500 282,517
R‐squared 0.102 0.082
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Source: HBS & microsimulation