ÖREBRO UNIVERSITY
Department of Business, Economics, Statistics and Informatics Economics C-thesis
Instructor: Jörgen Levin Autumn 2006
Aid required to halving the poverty in
Tanzania until 2015
Anders Johansson Joakim Lindberg
Abstract
Halving poverty until 2015 is something that all member states of United Nation have agreed up on by adopting the Millennium Development Goals MDG in 1991. The question asked in this paper is how much aid is needed in Tanzania, to reduce poverty by half. The method we use links estimated annual economic growth rates to the required amount of aid needed to halve poverty.
This study finds that during 2002-2015 Tanzania should receive between $37 to $43 (2002 US) dollars per year and capita depending on the underlying assumptions. Between 1994 and 2002 Tanzania received 36 dollar per year and capita so the amount of aid must at least be kept on the same level as preceding years and perhaps be increased to reach the goal of halving poverty until 2015.
Moreover, this study only calculates the cost of reaching the first MDG and not the cost of reaching the remaining goals stated in the United Nations Millennium Declaration.
Table of content
1. Introduction ...1 2. Theory ...3 2.1. Measuring poverty ...3 2.3. Growth-poverty elasticity ...5 2.4. Growth theory ...72.4.1. Authors augmented growth theory ...9
3. The procedure for calculating the aid needed to halve poverty ...11
3.1. Calculating the households consumption 2002-2015...11
3.2. Calculating the growth-poverty elasticity ...12
3.3. Calculating the annual rate of decline in the headcount index and the yearly economic growth rates ...13
3.4. Calculating the investments as a share of GDP ...14
4. Data ...15
4.1. Execution and result of the Tanzanian Household Budget Survey 2000/2001 ...15
4.2. Data required for calculating the investment ratio ...16
4.2.1. Productivity of capital ...16
4.3. Other data required ...17
4.3.1. Average savings rate...17
4.3.2. Initial value of GDP per capita ...17
4.3.3. Private and public investments as a share of total investments ...17
5. Results ...18
5.1. Effects of crowding out ...20
6. Discussion ...20
7. Conclusions ...24
References...25
1. Introduction
How costly would it be to halve poverty in Tanzania? That is the question that this study will try to answer. On a global level the question of halving poverty originate from a meeting held on the eighth of September 2000 where all of the United Nations member countries adopted the United Nation Millennium Declaration. The United Nation Millennium Declaration can be described as a vision of a better world.
To specify what was intended by improving the living conditions in the new millennium a group of international institutions like for instance the World Bank, International Monetary Fund and the United Nations Secretariat discussed how to harmonize the development goals in the Millennium Declaration and the international development goals. This resulted in the Millennium Development Goals (MDG:s), which are eight different goals with 18 individual targets and 48 indicators. The MDG:s concern several objects like for instance poverty, health, HIV/Aids and global partnership. The member states of the United Nation have agreed to achieve the MDG:s in 2015. (United Nations General Assembly, 2001)
This study will concentrate on the first MDG, which is to eradicate poverty and hunger until 2015. The targets for this goal is to reduce the proportion of people living on less than a dollar per day by half and reduce the proportion of people who suffer from hunger by half. Globally more than 1.2 billion people suffered from extreme poverty in 1990, which was about 30 percent of the people living in the developing world. In 2002 the proportion had declined to about 20 percent but the decline was not distributed equally around the developing world. The number of people living on less than one dollar per day dropped, in absolute terms, fastest in East Asia while the absolute number of poor people living in sub-Saharan Africa increased by 140 million during the same period.
Despite the slow decline in poverty in sub-Saharan Africa many countries in that region show signs of increasing growth rates that could speed up the rate of poverty extermination (United Nations, 2006). One of the countries that have seen increasing growth rates during the last years is Tanzania, which is the country that this study will be concentrating on.
Since 1985, when Tanzania introduced a market-oriented reform program, real gross domestic product (GDP) growth has been positive most of the time. In the beginning of the 1990:s economic growth slowed down because of weakened macro-economic policies but after Tanzania returned to it’s reform course the economic growth picked up speed. The macro-economic stability was not the only reason why the growth rate resumed; structural
reforms were also introduced like for instance privatization of public enterprises, efforts to improve the business climate and privatization of agriculture land. The reform efforts have been supported by large inflows of foreign aid, which also has contributed to economic growth by stimulating the domestic demand. After 1995 GDP growth per capita accelerated and reached 4 percent in 2004 (World Bank, 2005a).
Increased GDP growth has, however, not reduced the percentage of people living below the poverty line by any substantial amount. A household budget survey (HBS) from 1991/1992 indicated that the percentage of people living below the poverty line was 38.6 percent while the latest HBS from 2000/2001 showed that the percentage of people living below the poverty line just had decreased by 3.2 percentage points to 35.4 percent. Due to sampling issues in the two surveys the change in poverty is insignificant and the poverty may not have decreased at all. The two surveys show however, that the severity of poverty has declined (World Bank, 2005a). This raises the question if economic growth is the most necessary mean for poverty extermination or if the degree of inequality also plays an important role.
Heltberg (2002) summarises the discussion about the relationship between growth and poverty. Two views can be identified in the discussion. The growth-optimists believe that growth in average income automatically benefits the poor by the “trickle-down” effect. The opposite group emphasizes that the distribution of income and wealth is the key to understand reduction in poverty and that poverty reduction cannot be achieved without reduction in inequality. Heltberg (2002) concludes that there is no constant relationship between growth and change in poverty. The change in poverty depends on the initial value of inequality and how the poverty line is situated relative to the mean income and that in most cases economic growth is the most important mean to reduce poverty but that inequality also matters.
This study applies a methodology developed by Kakwani and Son (2006) to calculate the amount of aid needed to halve poverty in Tanzania1. In addition, this study takes into account how poverty reduction in recent years effects the amount of aid needed and how crowding-out of private investment due to public investment effects the amount of aid needed.
The outline of the paper is as follows: First we briefly summarise theories about poverty, growth-poverty elasticity and growth. Second, the procedure for estimating the required aid needed to reach the first MDG is presented. In the fourth chapter data are
1
Kakwani and Son (2006) estimate the amount of aid between 2005 and 2015 needed to achieve the MDG of halving the poverty until 2015 for 15 African countries.
described. The fifth chapter presents the results followed by an analysis and discussion of the results in the sixth chapter. In the sixth chapter we also summarise some recent findings regarding growth and aid adequate for this study. The last chapter concludes.
2. Theory
This chapter connects the concept of poverty and poverty measures to theories about poverty reduction such as growth-poverty elasticity, which states how much an economic growth rate of one percent reduces poverty. The simple growth model used by Kakwani and Son (2006) is derived and augmented, incorporating crowding-out effects.
2.1. Measuring poverty
The first issue in trying to measure poverty is to identify the poor among the total population. In order to distinguish between the poor and the non poor we have to discriminate between those two groups by using a poverty line2. The second problem is how to construct a poverty index that describes how poor an individual, household or a country is (Sen, 1997).
There are several ways of defining a poverty line. Generally, the poverty line is a critical threshold of income, consumption, or access to goods and services. If individuals fall below the poverty line they are regarded as poor (Ray, 1998).
Using food as a critical threshold gives the food poverty line, which is based on the fact that we need a minimum amount of calories, vitamins and minerals a day in order to live a healthy life. When determining the food poverty line, a food basket necessary to obtain the required energy intake must be composed. The cost of purchasing the food basket is the food poverty line. In the aggregate these amounts are expressed in adult equivalent since most data is in household form. It is assumed that for instance a child needs 50 percent less calories than a full-grown human being and is there for given a different weight in the household. Since most data is at the household level it is then possible to say how many calories a household
2 This study uses a welfarist approach, which uses income or consumption as a proxy for “well-being” or utility. An individual might be defined as poor even though the individual have the capability to have higher income or consume more but deliberately chooses not to and i.e. falls below the poverty line. One can argue that the dimensions of poverty cannot be comprised to a single dimension and that it is multidimensional in nature. See Duclos (2002) for a short description of welfarist versus non-welfarist approach or Sen (1997) for the concept of functionings.
needs and then calculate the income needed to purchase that amount of calories. One can argue that poverty is broader matter than just having enough calories to survive. It is hard to argue against that having shelter or having a decent set of clothing is a basic fundamental human necessity.
Using a basic needs poverty line is maybe a better approach since it allows for expenditures on necessary commodities. It is based on the food poverty line and a non-food component. The non-food component is often derived from the share of food in total expenditures. Dividing the food poverty line with the share of food in total expenditures yields the basic needs poverty line (Duclos, 2002).
Amartya Sen proposed a set of axioms so that the poverty measure describes the poverty adequate (Sen, 1976):
Axiom 1: Monotonicity. Given other things, a reduction in income of a person below the poverty line must increase the poverty measure.
Axiom 2: Transfer. Given other things, a pure transfer of income from a person below the poverty line to anyone who is richer must increase the poverty measure.
However, these two axioms are not enough to explain how poor the poor is and another axiom is therefore needed in order to take this into account (Kakwani, 1980):
Axiom 3: Transfer sensitivity. If a transfer t > 0 of income takes place from a poor household with income xi to a poor household with income xi + h (h > 0), then the magnitude of the
increase in poverty must be smaller for larger xi.
There are several different poverty measures which all describe different aspects of poverty. Foster, Greer and Thorbecke (1987) developed a decomposable poverty measure that is able to fulfill the above axioms. Further, this will be referred to as FGT-poverty index.
Let x = x
(
1, x2,..., xn)
be a vector of household incomes or expenditures per adult equivalent in increasing order, z > 0 is the poverty line, q is the number of poor households, n is the number of households and is a poverty weighting parameter. It is then possible to generalize to the following class of poverty index:P =1 n z xi z i=1 q
, > 0 (1) If =0 equation (1) becomes: P0= q n = H (2)This only shows the number of poor people as a fraction of the total population and does not account for the extent of shortfall from the poverty line. This is also referred to as the headcount ratio, H.
If we set =1, then one monetary unit gained by the very poor has the same impact on the poverty index as if the same amount was given to an individual close to the poverty line. That is, the index shows the average income distance from the poverty line, i.e. the income gap. However the measure is problematic since it will be easier to target the poor just below the poverty line when trying to reduce the poverty. For the measure to account for the severity of poverty we can set =2. Then greater weights are given to the poor far from the poverty line than those rather close to it3. The method we are using is applicable on different poverty indexes but we choose to use the headcount ratio in this study. This is justified because the headcount index has a straightforward interpretation showing the share of individuals below a given poverty line. Other poverty indexes can sometimes be harder to interpret and therefore more confusing.
2.3. Growth-poverty elasticity4
The growth elasticity of poverty is a measure that describes how a one percent change in economic growth changes the percentage of poor. Kakwani (1993) derives the growth-poverty elasticity under the assumption that poverty measures are functions of three variables; the poverty line, z , average per capita income, μ, and the inequality of income generally described by a Lorenz curve, L.
3 If
=0 then none of the axioms are fulfilled, =1 then axiom 1 and 2 are satisfied and for =2 all three axioms are satisfied. For derivation see Foster, Greer and Thorbecke (1987).
When the Lorenz curve is used to describe the income inequality any changes in the Lorenz curve will represent changes in the distribution of income. If the Lorenz curve is characterized by v number of parameters, m1 m2...mv, the Lorenz curve will shift due to
changes in these parameters. The poverty index, , can therefore be described by the function: )) ( , , (z L m f μ = (3)
If the poverty line is constant, changes in the poverty index can be expressed as:
= + = v i i i dm m d d 1 μ μ (4)According to this expression the change in poverty index depends on two effects. The first component on the right hand side is “the pure growth effect”. The pure growth effect is obtained by leaving the income distribution, i.e. the Lorenz curve, unaffected. The second component on the right hand side is the “inequality effect”, which is obtained by leaving the total income unchanged.
The pure growth effect can be estimated by assuming that individual income, x, is a random variable, where the distribution function of individual income is written as F(x) and the density function is expressed like f(x).
If z is the poverty line, the proportion of the individuals below the poverty line can be written as F(z), which is the headcount ratio. This can be generalized for a class of poverty measures, including the FGT-measures, as:
= z dx x f x z P 0 ) ( ) , ( (5) 5The elasticity of with respect to mean income per capita, while holding the distribution constant, is expressed as:
5 P(z,x)
is a homogenous function of degree zero in z and x since it holds that <0
x P and 2 0 2 x P
For the FGT-poverty measure we can write equation (5) as:
P = z x z 0 z
f (x)dx (7)The elasticity can then be written as:
P = P μ μ P = (P1 P) P (8)
When =0, which is the case when the headcount index, H, of poverty is of interest, the elasticity of headcount with respect to mean income becomes:
H = H μ μ H = zf (z) H < 0 (9)
This expression shows the percentage of poor who will cross the poverty line when average income grows with one percent, holding the distribution of income constant (Kakwani, 1993).
2.4. Growth theory6
Kakwani and Son (2006) use a simplified growth model in which they calculate the gross investment share of output needed to halve poverty until 2015. Since the household budget survey contains information regarding household consumption a relationship between the growth rate of household consumption and the gross investment rate must be derived. When there exists no direct relationship between the growth rate of household consumption and the gross investment rate Kakwani and Son assumes that the household consumption growth rate,
on average, is the same as the growth rate of GDP per capita. Using the assumption of equal growth rates of household consumption and GDP per capita makes it possible to estimate the investments required by using a growth model that links capital accumulation to economic growth. The growth model used by Kakwani and Son takes only physical capital into account, though there are other growth models that take human capital into account as well (Romer, 2001).
Kakwani and Son use the following production function to derive an expression for economic growth: ) ( ) ( ) (t a K t a K t Y = r r + g g (10)
where Y is output, Kr is the stock of private capital, Kg is the stock of public capital, ar and ag
is the productivity of the private and public capital stocks, respectively, and t is a time subscript. Differentiating the production function with respect to t shows how output is effected by investments in private (I ) and public capital (r Ig) and the depreciation rate, . With dropped time subscript the differentiated production function is written as:
dt dK a dt dK a dt dY g g r r + = (11) where r r r K I dt dK = and g g g K I dt dK =
Inserting this in equation (11) gives the following expression:
) ( ) ( ) ( r r g g g r r g g r r g g r I K a I K a I a I a K a K a dt dY + + = + = (12)
By using equations (10) and (12) the growth rate of output, g, can be expressed as:
(
)
(
+ +)
= + + = r r g g r r g g r r g g g g r r i a i a K a K a I a I a K a K a dt dY Y g (1/ )( / ) 1 (13)where Y I i r r = and Y I
ig = g are the shares of private and public investments in output, respectively.
According to equation (13) the growth rate of the economy depends on the productivity of private and public capital, the share of private and public investment in output, and the depreciation rate. When taking the effect from population growth, gpop, into account, equation (13) can be rewritten as:
g = arir+ agig ( + gpop) (14)
According to equation (14) the growth rate of the economy depends positively on the productivity of private and public capital and the share of private and public capital of output, and negatively on the population growth rate and the depreciation rate for fixed capital.
2.4.1. Authors augmented growth theory
To incorporate how crowding-out effects the investment ratio, when assuming the same productivity for both private and public investments, the previously presented growth model by Kakwani and Son (2006) need to be extended. The following section is therefore a modification made by the authors of this paper.
Let 0< <1 be the crowding-out ratio, which is measured by the fraction of how much public investment reduces private investments and thus effects the value of the total investments. One in the case when public investments offset the private investments by 100 percent and zero in the case when public investments not offset private investments at all. Equation (14) will then be written like:
) ( ) ( r g g g pop r i i a i g a g = + + (15)
Two new variables must be added in the model so the model can take crowding-out into consideration: Let irg denote total investment as a share of output. The shares of private and
public investments of total investment share in output will then be expressed as r rg r i i i ˆ = and g rg g i i i ˆ =
Equation (15) can then be rewritten as:
) ( ) ˆ ˆ ˆ ( ) ( ) ˆ ) ˆ ˆ ( (ar ir ig agig irg gpop arir ar ig agig irg gpop g = + + = + + (16)
The growth rate of the economy will as earlier depend negative on the population growth rate and the depreciation rate for fixed capital. The growth rate will also depend on the shares of public and private investments of total investments and the degree of crowding-out. From equation (16) the aggregated investments as a share of GDP can be expressed like:
) ˆ ˆ ˆ ( ) ( g g g r r r pop rg i a i a i a g g i + + + = (17)
Assuming that there is no crowding-out of private investments due to public investments, 0
=
, iˆr and iˆ will sum to one. Under the assumption that the productivity of private and g public capital is the same, ar =ag =a, equation (17) can be written as:
) ( 1 pop rg g g a i = + + (18)
Assuming that there is crowding-out of private investments due to public investments, 1
0< , and that the productivity of private and public capital is the same, ar =ag =a, equation (17) can be written as:
) ˆ ˆ ˆ ( ) ( 1 g g r pop rg i i i g g a i + + + = (19)
Due to crowding-out, the value inside the parenthesis in the denominator will be less than one and the investments share of GDP will be larger, ceteris paribus. It might be seen as a paradox that the investments share of output will be larger when there is crowding-out but that depends on the fact that the model uses a fixed economic growth rate and when there is crowding-out effects, the total investments share in output must be larger to offset the effect from crowding-out.
3. The procedure for calculating the aid needed to halve poverty
7In this part of the study the procedure for calculating the amount of aid needed to reach the MDG is presented. The chapter starts by explaining how the growth-poverty elasticity is calculated by using the data from HBS 2000/2001. Second the economic growth rate required to reach the MDG is derived. In the last part of the chapter the methodology to attain the aid needed to halve poverty is explained.
3.1. Calculating the households consumption 2002-2015
Economic growth may be distributed differently within a population. This leads to the effect that some will receive greater (smaller) benefits from economic growth than others. Economic growth can therefore be divided into pro-poor growth, anti-poor growth and distribution neutral growth depending on how growth effects the distribution of income. Growth is pro-poor if it is accompanied by a decrease in inequality, anti-pro-poor if the growth is accompanied by an increase in inequality and distribution neutral if there is no change in equality. To take the distribution of growth into consideration, the parameter k indicates how the growth is distributed. Let inequality be measured by the Gini-index, then if mean consumption changes by r percent, the Gini-index will change with kr percent. With pro-poor growth the change in the Gini-index is negative, implying a negative value of k and indicating a more equal distribution, while with anti-poor growth the change in the Gini-index is positive implying a positive value of k and indicating a more unequal distribution. The growth is distribution neutral if the Gini-index does not change implying that the value of k is zero. To be able to
calculate the change of each household’s consumption given different growth scenarios the following formula is used:
(
x kr x)(
r)
xit = it1+ ( it1 μt1) 1+ (20)
where xit is the consumption of the i:th household in period t, r is the growth rate of mean consumption, μ is the average consumption and kr reflects the distribution pattern of economic growth. To be able to calculate each household’s consumption given the different growth scenarios, k takes the value of zero if the growth is distribution neutral, -0.5 if the growth is pro-poor and 0.5 if the growth is anti-poor. Given the different values of k and a growth rate of mean consumption of one percent per year, each household’s consumption can be calculated according to (20) for the years between 2002 and 2015.
3.2. Calculating the growth-poverty elasticity
The growth-poverty elasticity for each year is possible to calculate for the headcount measure of poverty by first calculating the headcount index, Ht. The headcount index is the proportion of the population below any given poverty line. The headcount index can be calculated by defining a dummy variable that takes the value one if the household falls below the poverty line, z, and zero otherwise. The dummy variable has the formal expression:
= 0 1 it if if z x z x it it < (21)
The headcount index for period t can then be estimated by:
= = n i i it t w H 1 (22)where wi is the population weight attached to the i:th household used in the HBS to make the sample of households representative for mainland Tanzania. The growth-poverty elasticity can then be calculated for the time period 2002 to 2015 by using equation (9)8.
3.3. Calculating the annual rate of decline in the headcount index and the yearly economic growth rates
According to the MDG:s the headcount ratio of poverty should be halved between 1990 and 2015. The annual decline in the headcount ratio can be calculated by using the equation below and solving for m. The equation holds for the assumption that the headcount decline rate is constant each year:
25 1990
2015 P (1 m)
P = + (23)
where m is the annual headcount ratio decline rate between 1990 and 2015, P2015 is the headcount rate in 2015 which can be generalized to 0.5 since the poverty should be halved until 2015 and P1990 is the headcount rate in 1990 which can be generalized to 1 since that is
the initial headcount ratio that should be halved.
In this study the same technique as above is used for calculating the annual decline rate of the headcount ratio but with some changes. Since there is information regarding the headcount ratio in 1991/92 and 2000/01 this makes it possible to calculate the annual decline rate between 2002 and 2015 needed to halve the poverty when taking into consideration how much the headcount ratio had declined between 1991/92 and 2000/01. The equation for calculating the annual decline becomes:
14 1991 / 1990 02 / 2001 2015 (1 m) P P P = + (24)
where P2015 is the headcount ratio that shall be reached according to the MDG:s, P1990/91 is the headcount ratio in 1991, P2000/01 is the headcount ratio in 2001 and m is the annual decline in
the headcount ratio between 2002 and 2015.
The yearly growth rate of consumption needed to reach the goal of halving poverty will be attained by the following formula:
t k t k m g , , = (25)
where gk,t is the growth rate of consumption needed to reach the MDG in period t given growth scenario, k, m is the annual rate of decline in the headcount ratio needed to reach the MDG, and is the growth-poverty elasticity in period k ,t t given the different growth scenarios, k.
3.4. Calculating the investments as a share of GDP
The investment share of GDP per capita can be calculated for different growth scenarios, different time periods and different crowding-out ratios by using formula (19). By subtracting the calculated investment share with the rate of gross national savings share, the savings gap, which is needed to halve poverty, is attained. To estimate the investment gap in US dollars per capita, the per capita GDP is calculated for the different time periods and growth scenarios. The value of GDP per capita for each year can be estimated by using the formula:
) 1 ( , 1 , ,t kt kt k GDP g GDP = + (26)
where GDPk,t is the GDP per capita in period t given different growth scenarios k, GDPk,t1 is
the GDP per capita in period t-1 and given different growth scenarios k, and gk,t is the growth rate in period t given different growth scenario k. The additional per capita investments needed to achieve the MDG are calculated by multiplying the yearly savings gap as a share of GDP with the estimated GDP per capita for each year. The investment gap9 will be expressed in 2002 US dollars.
9
It should be noticed that countries can finance their investment gap from different sources. The most obvious source is domestic savings. In this paper the assumption is that foreign aid covers the investment gap.
4. Data
In this section the data material required to calculate the amount of aid needed to halve poverty is presented. The section starts with a description of how the Tanzanian Household Budget Survey was executed and the results that stem from the survey. Then the required data for calculating the necessary amount of aid are presented.
4.1. Execution and result of the Tanzanian Household Budget Survey 2000/2001
The Tanzanian National Bureau of Statistics implemented the work with the Household Budget Survey (HBS) 2000/2001 towards the end of 1999. The survey started during the months of May and June in 2000 and was completed in June 2001.
The sample of households used in the HBS 2000/2001 was larger than the sample used in previous HBS, because the aim of the HBS 2000/2001 was to provide estimates of poverty measures for each of the 20 regions of the Tanzanian mainland. The sample of households in HBS 2000/2001 was selected in two steps. In the first step 1161 Primary Sampling Units (PSU) were selected from the 20 regions of mainland Tanzania. In the second step 24 households were selected in each PSU. The survey is representative for the total population of mainland Tanzania and the final number of households participating in the HBS 2000/2001 become, after adjustments, 22 178, which is about 98 percent of the final sample.
In the HBS 2000/2001 two poverty lines are defined, the food poverty line and the basic needs poverty line. The food poverty line corresponds to the minimal expenditure necessary to attain 2200 calories per day, which is the minimum amount of calories necessary for survival. The basic needs poverty line allows for some expenditure on essential non-food goods and is therefore higher than the food poverty line.
Consumption expenditure was used as a measure of the household’s welfare instead of the household’s income. The reason behind that is that in developing countries household income is more volatile than the value of expenditure and is reported less precisely than consumption expenditure (United Republic of Tanzania, National Bureau of Statistics, 2002). The HBS performed 1991/1992 indicated that the percentage of people living below the basic needs poverty line was 38.6 percent while the latest HBS from 2000/2001 showed that the percentage of people living below the basic needs poverty line had decreased by only 3.2 percentage points to 35.4 percent. Due to different sampling techniques in the two surveys the
change in poverty is insignificant and poverty may not have decreased at all. However, the two studies showed that the severity of poverty has declined. The basic needs poverty line used in these two surveys was 7253 Tanzanian shillings which is below the international one dollar per day poverty line and lower than poverty lines used by other countries in the region (World Bank, 2005a).
4.2. Data required for calculating the investment ratio
According to equation (17) the investment share of GDP depends on the productivity of capital, the growth rate of consumption per capita, the population growth rate, the depreciation rate for fixed capital, and the crowding-out ratio.
4.2.1. Productivity of capital
The productivity of capital, a , which can be measured as output per unit of capital, is in Kakwani and Son (2006) constant and takes a value of 1/3. Kahn and Kumar (1993) investigated how long-run economic growth per capita is contributed by public and private investments during 1970-1990 for 46 African countries including Tanzania. They found that a one percent increase in public or private investments increases economic growth per capita by 0.32 percent. This can be interpreted that the elasticity of investments with respect to growth is 0.32 for both private and public capital. This result seems to support the assumption that there is no difference between productivity of private and public capital and that productivity is around 1/3.
4.2.2. Population growth rate
Average annual population growth rate between 2002 and 2015 is estimated to be 1.7 percent (World Bank, 2004). In this study the same population growth rate is used for all years. This approach differs from the method used in Kakwani and Son (2006), who use a trend regression model to calculate population growth.
4.2.3. Depreciation rate for fixed capital
The average depreciation rate for fixed capital is in the study by Kakwani and Son (2006) 3.1 percent for the 15 studied countries. The same depreciation rate for fixed capital is used in this study.
4.3. Other data required
To complete the task of calculating the investment gap between yearly investments and the required investments needed to reach the MDG of halving poverty, the average savings rate in Tanzania must be calculated and an initial value of GDP per capita must be set. In the case where we assume crowding-out we need to know the shares of private and public investments of total investments.
4.3.1. Average savings rate
The gross national savings rate as a share of GDP for Tanzania between the years 1994 and 2002, can be seen in Table A.1 in the Appendix. The savings rate has increased significantly during the last years in the sample. Therefore, the average savings rate of the last three observations was used to get a more adequate average savings rate. The average savings rate as a share of GDP used in this study is 8.6 percent (World Bank, 2005b).
4.3.2. Initial value of GDP per capita
Since the period of interest in this paper is 2002-2015, the starting value of GDP per capita is the value of GDP per capita in 2001. The GDP per capita 2001 expressed in 2002 USD was 278.2 USD (International Monetary Fund, 2004).
4.3.3. Private and public investments as a share of total investments
Using International Money Fund (2004), public and private investment as a share of total investment was calculated. The average shares of private and public investment between 1997 and 2002 were used in this study. Public investment as a share of total investment is 28.7
percent and private investment as a share of total investment is 71.3 percent. Figures are presented in Table A.2 in the Appendix.
5. Results
In order to reach the first MDG Tanzania has to reduce the headcount ratio from 38.6 percent to 19.3 percent until 2015. Using the same approach as Kakwani and Son (2006) implies that the headcount ratio needs to decline at an annual rate of 2.7 percent. Considering how much Tanzania has reduced poverty until 2001, the headcount ratio has to decline at an annual rate of 4.3 percent.
The growth-poverty elasticity, assuming that growth is neutral, yields an average reduction of poverty by 2.0 percent for a one percent growth in mean consumption. However, if growth is pro-poor the reduction of poverty is larger, on average 2.3 percent, and if growth is anti-poor the reduction of poverty is lower, on average 1.8 percent. The growth-poverty elasticity, annual economic growth, investments as a share of GDP, savings-gap as a share of GDP and aid per capita between 2002 and 2015 can be seen in the Appendix Table A.4 to Table A.14.
Knowing the growth-poverty elasticity and how much poverty that needs to be reduced on a yearly basis, the annual per capita growth rate required to achieve that reduction in headcount ratio can be calculated. Under the assumption that poverty has to decline by 2.7 percent each year results in an average growth rate of 1.4 percent (Table 1), if growth is distribution neutral. However, if growth is pro-poor an average growth rate of 1.2 percent is needed and if growth is anti-poor an average growth rate of 1.5 percent is needed.
Accounting for how much Tanzania has reduced poverty since 1990 implies higher growth rates (Table 2). The minimum growth rate required for halving poverty by 2015, when growth is neutral, is on average 2.1 percent. As expected, a lower growth rate is required if growth is pro-poor, a rate of 1.9 percent is then needed. In the anti-poor case a growth rate of 2.4 percent is required to meet the goal.
Table 1, Average values for different growth scenarios when m=2.7
Pro-poor Neutral Anti-poor
Growth-Poverty Elasticity -2.3 -2.0 -1.8
Growth per capita 1.2% 1.4% 1.5%
Investments (in % of GDP) 18.0% 18.5% 19.0%
Savings gap (in % of GDP) 9.4% 9.8% 10.3%
Aid per capita $ 28.9 $ 30.5 $ 32.4
(Source: authors calculations)
Table 2, Average values for different growth scenarios when m=4.3
Pro-poor Neutral Anti-poor
Growth-Poverty Elasticity -2.3 -2.0 -1.8
Growth per capita 1.9% 2.1% 2.4%
Investments (in % of GDP) 20.1% 20.8% 21.6%
Savings gap (in % of GDP) 11.5% 12.1% 12.9%
Aid per capita $ 37.2 $ 40.0 $ 43.3
(Source: authors calculations)
Knowing the productivity of capital, the annual growth rates, the population growth rate and the depreciation rate for fixed capita it is possible to calculate the yearly investments as a share of GDP needed to halve poverty until 2015. Using m=2.7, under the neutral growth scenario the average investments needed to halve poverty as a share of GDP is on average 18.5 percent, for the pro-poor growth scenario 18.0 percent and for the anti-poor growth scenario 19.0 percent (Table 1). Assuming m=4.3 increases the investments needed by between 1.5 and 2 percentage points (Table 2).
Assuming that the national gross savings is constant over the time period (and using average values over three years) yields a national savings rate of 8.6 percent per year as a share of GDP. The difference between investments required and national savings rate yields the savings-gap. Neutral growth and m=2.7 results in a savings-gap of 9.8 percent of GDP, 9.4 percent under pro-poor growth and 10.3 percent under anti-poor growth (Table 1). As with investments, assuming m=4.3 increases the savings-gap proportionally (Table 2).
Finally it is possible to calculate how much aid that is needed under the different growth scenarios. Using the same method as Kakwani and Son (m=2.7) yields the results that when growth is distribution neutral the average amount of aid needed per capita in 2002 USD is $30.5, under pro-poor growth $28.9 and under anti-poor growth $32.4. Considering how much poverty already has been reduced (m=4.3), higher amounts of aid per capita are needed in order to achieve the MDG goal as shown in Table 2.
5.1. Effects of crowding out
The crowding-out parameter reflects how public investments offset private investments. Assuming different values on the crowding-out parameter we can see whether it is of importance for further analysis. The assumed values of the crowding-out parameter have no empirical foundation and are pure ad hoc. However, they allow us to see how the results change. Table 3 shows how much more investments that are needed under different crowding-out scenarios.
Table 3, Average needed investments as a share of GDP assuming different values on the
crowding-out parameter and different growth scenarios, (m=4.3)
Pro-poor Neutral Anti-poor
0.0 20.1% 20.8% 21.6%
0.2 21.3% 22.0% 22.9%
0.4 22.7% 23.5% 24.4%
0.6 24.3% 25.1% 26.0%
(Source: authors calculations)
Since required investments increase so does aid needed. The impact on aid is shown in Table 4.
Table 4, Average aid needed per capita to reach MDG goal in 2002 US dollar, assuming
different values on the crowding-out parameter and different growth scenarios.
Pro-poor Neutral Anti-poor
0.0 $36.9 $39.7 $43.0
0.2 $40.9 $43.9 $47.4
0.4 $45.3 $48.5 $52.3
0.6 $50.4 $53.9 $57.9
(Source: authors calculations)
As can be seen, a higher value of the crowding-out parameter increases the investments needed to reach the MDG. Due to that the amount of aid needed to reach the MDG increases as well.
6. Discussion
Kakwani and Son (2006) studied how much poverty needs to be reduced on a yearly basis during the whole 25-year period, 1990-2015. However, since the HBS 2001 is available for Tanzania the progress in poverty reduction so far could be accounted for. Since Tanzania has had little or no reduction in poverty between 1991 and 2001, this would make the estimation
more accurate. A critique against this would be that the two surveys are not comparable since they use different household weights. This study shows both ways and according to the result, the savings-gap and aid per capita needed increase substantially when accounting for the small progress in poverty reduction between 1991 and 2002.
As expected, the results showed that if growth is beneficial to the poor, i.e. lowers inequality measured by the Gini-index, the growth-poverty elasticity is higher. This means that under the pro-poor scenario a one percent growth will reduce poverty by a greater extent. The reverse holds for the anti-poor scenario since growth will be accompanied by an increase in inequality, and, hence benefits the rich more than the poor and higher growth is needed to reduce poverty. Our average growth-poverty elasticity ranges from -1.8 to -2.3 depending on the growth scenario. In Kakwani and Son (2006) the average growth-poverty elasticity for different growth scenarios and countries varies between -0.7 and -2.4. Our results vary less between different growth scenarios and time periods. The growth-poverty elasticity depends on initial inequality and therefore one explanation could be that Tanzania has a lower Gini-coefficient than most other countries in Sub-Saharan Africa. Another explanation could be that Tanzania has a more equal distribution of incomes among the poor and shifting the distribution upward holding the poverty line constant yields about the same number of people crossing the poverty line at every shift.
This study used a simplified growth model where overall economic growth depends only on savings rate, depreciation, capital productivity and population growth rate. A number of strong assumptions are being made, such as assuming the same productivity for both public and private investments. Assumptions are also being made about future savings and population growth rates. A different approach could have been to use more sophisticated growth models. However due to lack of data or inferior quality, could have led to making even more ad hoc assumptions. The study uses basic needs as a national poverty line and headcount as poverty index, which does not fully take into consideration the severity and depth of the poverty. However, the method is applicable to other poverty lines and indexes.
This study showed that Tanzania needs approximately $40 in aid per capita per year if growth is neutral, $37 if growth is pro-poor or $43 if growth is anti-poor. Applying the same annual decline in poverty as Kakwani and Son (2006), the aid needed ranges from between $29 to $32 depending on how growth effects inequality. Tanzania lies somewhere in the middle among other Sub-Saharan African countries when it comes to the calculated aid per capita needed. Between 1994 and 2003, Tanzania has received on average $36 per capita in
aid annually. This indicates that aid revived by Tanzania is almost sufficient in order to reach the first MDG. However, a slight increase in aid might be necessary to reach the goal.
A lot more need to be said before drawing any conclusions about the level of aid and if it should be sustained at the same level or if it should be scaled up. Aid might have little effect on growth and growth might have little effect on poverty. Decreasing effectiveness of aid is also a potential problem.
The question of whether or not foreign aid spurs economic growth in the developing countries and thereby may reduce poverty has been deeply examined since the mid twentieth century. The research done between 1950 and 1996 regarding this question could be divided into two groups (McGillivray et al., 2005). Either the aid was considered to result in higher savings rate, which according to the Harrod-Domar model increased the growth rates or the aid had no impact on the savings rate and thereby not resulting in higher economic growth rates. A publication made by the World Bank (1998), “Assessing aid: What works, What doesn’t and Why”, developed the discussion on aid. The report stated that aid increases economic growth but only in countries where the government performs good economic policies and builds good institutions. The conclusion of the report was that aid should be given to countries with good institutions and governance.
Collier and Dollar (2002) showed that if aid should be given to countries with these circumstances the number of poor should be reduced by 18 millions more per year compared to the existed aid allocation. The Assessing Aid report initiated a lot of research where other scientists tried to remake the econometrics behind the report but without getting the result that the policy environment should be of special concern when addressing aid. The assumption that policy environment should be important for the linkage between aid and higher growth rates then seems a bit weak. The report was also criticized due to the econometrics method used to support the theory. After the report was published, alternative ideas of explaining the linkage between the effectiveness of aid and economic growth occurred. The main ideas that can be seen in the literature are that: aid has decreasing returns so countries can’t absorb the amount of aid received effectively and large inflows of capital cause the problems of Dutch disease, aid flows are not constant and thereby reducing the effectivity aid, the effectivity of aid depends on climate-, external-, and political conditions and finally that the effectivity of aid depends on the institutional quality within the countries that receives aid. Despite the different views on what determines the effectiveness of aid, it seems like aid increases economic growth one way or another (McGillivray et al., 2005).
investments can crowd out private investments. Incorporating crowding-out scenarios lead to, as expected, that the amount of aid needed increases monotonically. When the crowding-out parameter goes from 0 to 0.6, aid increases from $40 to $54 under neutral growth. The parameter seems to be of vital interest in order to get a good approximation for the amount of aid needed. However, this study assumed the same productivity in both public and private investments and the values used on the crowding-out parameter are purely ad hoc. If productivity was higher in the public sector, crowding-out would have a positive impact. There is also a possibility of crowding-in, where government investments in infrastructure might increase incentives for private investments. In Kakwani and Son (2006) they assume that there is no crowding-out effect present. That assumption is based on a number of articles like for instance an article made by Easterly and Rebelo (1993).
Scaling up aid can have different impact in different countries. Aid can have little effect on growth or poverty and can lead to implausible effects such as Dutch disease. Gupta, Powell, and Yang (2006) argue that an increase in aid must be followed by policies that allow the country to absorb and spend the aid inflows without destabilizing the macroeconomic environment. The typical assumption in scaling-up scenarios is that the country will both absorb and spend most of the aid, which raises the possibility that the real exchange will appreciate. The higher domestic demand raises the prices of nontradables in relation to tradables, hence productive resources will move away from exporting sectors. However, they argue that when ensuring high import content in the increased public spending, investing in infrastructure (which may speed up progress in productivity), and liberalizing trade, the risks of Dutch disease are reduced. Further, Gupta et al. (2006) argue that aid does not generate growth per se. Some types of aid (emergency assistance, humanitarian aid, disaster relief etc.) do not have the intention to spur growth and some types of aid only increase growth in the very long run (environmental, institutional, education etc.). Short to medium run policies to generate growth often focus on infrastructure such as building road, ports, electricity etc. However, the effectiveness of aid can fast diminish in sectors where there is potential occurrence of supply bottlenecks. Good governance and a sound policy environment probably enhance the effectiveness of aid. Gupta et al. (2006) also argue that an exit strategy has to follow with a scale-up of aid, in the situation where the higher amount of aid is not to be sustained.
Both Treichel (2005) and World Bank (2005A) argue that the growth performance in Tanzania since 1995 is a result of several structural reforms such as large-scale privatization, liberalization and macroeconomic stabilization. Continuation of the reform programs will
give Tanzania prospects of sustaining a high growth rate until 2015. However, Tanzania had little reduction in poverty between 1990 and 2000 even though the country showed endorsement for pursuing structural reforms and has had high growth rates since 1995. Further, Treichel (2005) is hopeful regarding Tanzania’s possibility to reach the first MDG. As discussed earlier the effect of aid on growth and poverty reduction is a widely discussed topic. Maybe one can say that among the Sub Saharan African countries, Tanzania is among the countries most likely to reach the first MDG.
7. Conclusions
In this study the cost of reducing poverty by half in Tanzania has been calculated. The aid per capita needed was estimated to an average $37 to $43 (2002 US dollars) annually, depending on whether growth benefits the poor or not. This is bit higher than the average received aid between 1993 and 2003. Though, more should, be said before determining whether a higher amount of aid should be addressed to Tanzania. Scaling up aid might have decreasing effectiveness. An inflow of aid might cause Dutch disease. The effects of private investments offsetting private investments, i.e. crowding-out, were studied as well. This paper concludes that further study of crowding-out or crowding-in effects in Tanzania might be of interest since it seems to effect the investments required to halve poverty a great extent and hence the need of aid.
This paper concludes that an increase in aid has to follow with sound macroeconomic policies and shows that it is less costly if growth benefits the poor. Hence, efforts should be made to reduce inequality as well as increasing growth. Efforts should be made to identify the poor and create programs to spur economic growth in areas and sectors where a lot of poor people live and work.
If one is optimistic and Tanzania continues with the reform efforts made, maybe one can say that of the sub-Saharan African countries Tanzania is one candidate likely to reach the MDG of halving poverty by 2015.
References
Books
Sen, A (1997) On economic inequality. Clarendon Press.
Ray, D (1998) Development Economics. Princeton University Press. Romer, D. (2001) Advanced Macroeconomics. McGraw-Hill/Irwin
Articles
Foster, J. Greer, J. and Thorbecke, E “A class of Decomposable Poverty Measures”. Econometrica, Vol. 52, No. 3, pp. 761-766.
Kakwani, N (1980) “On a class of Poverty Measures”. Econometrica, Vol. 48, No 2, pp 437-446.
Kakwani, N (1993) “Poverty and Economic Growth with Application to Côte D’Ivoire”. Review of Income and Wealth, Series 39, No. 2, pp. 121-139.
Sen, A (1976) “Poverty: An Ordinal Approach to Measurement”. Econometrica, Vol. 44, No 2, pp. 219-231.
Collier, P. and Dollar, D. (2002) “Aid Allocation and Poverty Reduction”. European Economic Review, Vol. 46, No. 8, pp. 1475-1500.
Easterly, W. and Rebelo, S. (1993) “Fiscal Policy and Economic Growth: An Empirical Investigation”. Journal of Monetary Economics, Vol. 32, No. 3, pp 417-458.
Working papers
Treichel, V (2005) ”Tanzania’s Growth Process and Success in Reducing Poverty”, International Monetary Fund Working Paper 05/35.
McGillivray et al., (2005) “It Works; It Doesn’t; It Can, But That Depends...”, UNU-WIDER Research Paper No. 2005/54.
Heltberg, R. (2002) “The Poverty Elasticity of Growth”, UNU-WIDER Discussion Paper No. 2002/21.
Kahn, M. and Kumar, M. (1993) “Public and Private Investment and the Convergence of Per Capita Incomes in Developing Countries”, International Monetary Fund Working Paper 93/51.
Kakwani, N. and Son, H. (2006) “How costly is it to achieve the millennium development goal of halving poverty between 1990 and 2015?”, UNDP Working Paper number 19/2006.
Other references
Duclos, J-Y (2002) “Poverty and Equity: Theory and Estimation”. Département d’économique and CRÉFA, Université Laval.
Gupta, S, Powell, R. and Yang Y. (2006) “Macroeconomic Challenges of Scaling Up Aid in Africa: A Checklist for Practitioners”. International Monetary Fund.
World Bank (2005a) “Sustaining and Sharing Growth in Tanzania”, Country Economic Memorandum and Poverty Assessment (Draft), Washington DC.
World Bank (1998) “Assessing Aid: What Works, What Doesn’t and Why”, Policy Reasearch Report, Washington DC.
United Republic of Tanzania, National Bureau of Statistics (2002), “Household Budget Survey 2000/01”.
United Nations General Assembly (2001) “Road map towards the implementation of the United Nations Millenium Decleration- Report of the Secretary General ”, A/56/326, New York.
United Nations (2006) “The Millenium Development Goals Report – 2006”. Department of Economic and Social Affairs, New York.
Statistical resources
World Bank (2004) “World Development Indicators”. World Bank (2005b) “African Development Indicators”.
Appendix
Table A.1, Gross national savings in Tanzania, 1994-2002
Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 Gross national
savings 8,5 -0,2 2,1 2,2 -1,9 0,9 7,5 7,5 10,9
(source: African Development Indicators 2005)
Table A.2, Public/private sector fixed capital formation in Tanzania, 1997-2002
Year 1997 1998 1999 2000 2001 2002 Gross fixed capital formation 14,7 16 15,4 17,4 16,8 18,9 Public sector fixed capital
formation 2,9 3,3 3,1 6 5,6 7,6
Private sector fixed capital
formation 11,8 12,7 12,3 11,4 11,2 11,4
(source: IMF Country Report No. 04/284, Statistical appendix)
Table A.3, Aid/capita received in Tanzania (US dollar)
Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Aid per capita, current
prices $34 $30 $29 $30 $31 $30 $30 $37 $35 $47
Aid per capita, 2002
dollar $41 $35 $33 $34 $34 $32 $31 $38 $35 $46
Table A.4, Annual growth-poverty elasticity for different growth scenarios
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor -1,92 -1,91 -1,90 -1,88 -1,87 -1,85 -1,84 -1,82 -1,81 -1,78 -1,76 -1,74 -1,72 -1,71 -1,69 -1,81
Neutral -1,92 -1,94 -1,96 -1,98 -1,98 -2,01 -2,02 -2,03 -2,04 -2,06 -2,05 -2,06 -2,06 -2,06 -2,08 -2,02
Pro-Poor -1,92 -1,97 -2,02 -2,08 -2,12 -2,16 -2,20 -2,24 -2,28 -2,36 -2,38 -2,42 -2,49 -2,57 -2,66 -2,28
(source: authors calculations)
Table A.5, Annual economic growth rate, m=4,3
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 2,24% 2,25% 2,27% 2,29% 2,30% 2,32% 2,34% 2,36% 2,38% 2,41% 2,44% 2,47% 2,49% 2,52% 2,54% 2,38%
Neutral 2,24% 2,22% 2,20% 2,17% 2,17% 2,14% 2,13% 2,11% 2,11% 2,09% 2,10% 2,08% 2,09% 2,08% 2,06% 2,13%
Pro-Poor 2,24% 2,18% 2,13% 2,07% 2,03% 1,99% 1,96% 1,92% 1,88% 1,82% 1,80% 1,78% 1,73% 1,67% 1,62% 1,90%
(source: authors calculations)
Table A.6, Investments as a share of GDP, m=4,3
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 21,11% 21,15% 21,20% 21,27% 21,31% 21,36% 21,42% 21,48% 21,54% 21,63% 21,71% 21,80% 21,88% 21,96% 22,01% 21,55% Neutral 21,11% 21,05% 20,99% 20,92% 20,91% 20,81% 20,79% 20,74% 20,73% 20,67% 20,69% 20,65% 20,67% 20,65% 20,59% 20,78% Pro-Poor 21,11% 20,94% 20,78% 20,60% 20,48% 20,37% 20,27% 20,16% 20,05% 19,87% 19,81% 19,74% 19,58% 19,42% 19,25% 20,09%
(source: authors calculations)
Table A.7, Savings-investment gap as a share of GDP, m=4,3
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 12,48% 12,52% 12,56% 12,64% 12,68% 12,73% 12,78% 12,85% 12,91% 12,99% 13,07% 13,17% 13,25% 13,33% 13,38% 12,92% Neutral 12,48% 12,41% 12,36% 12,29% 12,28% 12,18% 12,16% 12,11% 12,10% 12,04% 12,06% 12,02% 12,04% 12,02% 11,96% 12,14% Pro-Poor 12,48% 12,31% 12,15% 11,96% 11,84% 11,74% 11,63% 11,53% 11,42% 11,24% 11,18% 11,10% 10,95% 10,79% 10,62% 11,46%
(source: authors calculations)
Table A.8, GDP/capita, m=4,3
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Anti-poor 278,18 284,44 290,89 297,55 304,40 311,47 318,75 326,28 334,05 342,09 350,43 359,07 368,03 377,30 386,88 Neutral 278,18 284,35 290,59 296,91 303,36 309,84 316,44 323,13 329,95 336,84 343,91 351,07 358,41 365,88 373,43 Pro-Poor 278,18 284,25 290,29 296,29 302,29 308,31 314,34 320,37 326,41 332,36 338,35 344,37 350,31 356,18 361,94
Table A.9, Aid/capita in 2002 USD, m=4,3
Aid/capita: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 34,71 35,60 36,55 37,60 38,60 39,65 40,75 41,92 43,12 44,45 45,82 47,29 48,77 50,28 51,75 43,01
Neutral 34,71 35,30 35,90 36,49 37,25 37,74 38,48 39,12 39,91 40,55 41,47 42,19 43,15 43,96 44,66 39,73
Pro-Poor 34,71 34,99 35,26 35,45 35,80 36,18 36,57 36,94 37,27 37,35 37,81 38,24 38,34 38,44 38,43 36,93
(source: authors calculations)
Table A.10, Annual economic growth rate, m=2,735
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 1,42% 1,43% 1,44% 1,46% 1,47% 1,48% 1,49% 1,50% 1,52% 1,53% 1,55% 1,57% 1,59% 1,60% 1,61% 1,52%
Neutral 1,42% 1,41% 1,40% 1,38% 1,38% 1,36% 1,36% 1,34% 1,34% 1,33% 1,33% 1,33% 1,33% 1,33% 1,31% 1,35%
Pro-Poor 1,42% 1,39% 1,35% 1,31% 1,29% 1,27% 1,24% 1,22% 1,20% 1,16% 1,15% 1,13% 1,10% 1,07% 1,03% 1,21%
(source: authors calculations)
Table A.11, Investments as a share of GDP, m=2,735
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 18,67% 18,69% 18,73% 18,77% 18,80% 18,83% 18,86% 18,91% 18,95% 19,00% 19,05% 19,11% 19,16% 19,21% 19,24% 18,95% Neutral 18,67% 18,63% 18,59% 18,55% 18,54% 18,48% 18,47% 18,43% 18,43% 18,39% 18,40% 18,38% 18,39% 18,38% 18,34% 18,46% Pro-Poor 18,67% 18,56% 18,46% 18,34% 18,27% 18,20% 18,13% 18,07% 18,00% 17,88% 17,84% 17,80% 17,70% 17,60% 17,49% 18,02%
(source: authors calculations)
Table A.12, Savings-investment gap as a share of GDP, m=2,735
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 10,04% 10,06% 10,09% 10,14% 10,17% 10,20% 10,23% 10,27% 10,31% 10,37% 10,42% 10,48% 10,53% 10,58% 10,61% 10,32%
Neutral 10,04% 10,00% 9,96% 9,92% 9,91% 9,85% 9,83% 9,80% 9,79% 9,76% 9,77% 9,74% 9,76% 9,74% 9,71% 9,82%
Pro-Poor 10,04% 9,93% 9,83% 9,71% 9,63% 9,57% 9,50% 9,43% 9,36% 9,25% 9,21% 9,16% 9,06% 8,96% 8,85% 9,39%
Table A.13, GDP/capita, m=2,735
k: 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Anti-poor 278,18 282,17 286,23 290,40 294,66 299,02 303,46 308,02 312,69 317,48 322,40 327,47 332,66 338,00 343,45
Neutral 278,18 282,10 286,05 290,01 294,01 298,01 302,05 306,11 310,22 314,35 318,54 322,77 327,06 331,40 335,75
Pro-Poor 278,18 282,04 285,86 289,62 293,35 297,06 300,76 304,44 308,09 311,66 315,23 318,80 322,30 325,74 329,09
(source: authors calculations)
Table A.14, Aid/capita in 2002 USD, m=2,735
K 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Avg:
Anti-poor 27,92 28,39 28,89 29,44 29,95 30,50 31,05 31,64 32,24 32,91 33,58 34,31 35,03 35,75 36,44 32,15
Neutral 27,92 28,20 28,49 28,76 29,14 29,35 29,71 30,00 30,38 30,67 31,12 31,45 31,92 32,29 32,59 30,29
Pro-Poor 27,92 28,01 28,09 28,12 28,26 28,42 28,57 28,72 28,84 28,82 29,03 29,21 29,21 29,20 29,14 28,69
