An analysis of the evidence for the existence of savings-based poverty traps in South America

(1)

University of Amsterdam Faculty of Economics and Business

An Analysis of the Evidence for the Existence of Savings-Based Poverty Traps in South America

Dominic Stone 10391371

Bachelor Thesis Economics BSc

Supervisor: Prof. dr. M.P. Pradhan 29th June 2016

(2)

1

Statement of Originality

This document is written by Dominic Stone who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

2

Abstract

Recent theoretical work has focused on the different mechanisms conducive to forming a self-reinforcing poverty trap. This paper reviews previous literature on this subject before

reproducing empirical research, used to assess the existence of savings-based poverty traps for the South American region. In order to do so, this paper will show the results of a regression that was conducted to explain the relationship between savings rates and capital per capita and to test for nonlinearities. Due to the significance of the results, the empirical research conducted does little to support the theory of savings-based poverty traps in South America and therefore questions proposals for large upscaling of aid to the region to act as the “big push” needed to drive the population above a theoretical poverty threshold level.

(4)

3

With at least one-fifth of the world’s population suffering from extreme poverty, achieving the objectives outlined in the Millennium Development Goals (MDG) requires a strategic focus on how to drive self –reinforcing growth among the poor (Barrett, Swallow, 2006). Furthermore, Barrett and Swallow (2006) found that to increase the income of over one billion people living in extreme poverty by just $1 per day would require an additional $450 billion per year for these people. The observation of such persistent and chronic poverty beggars the question what keeps populations poor and who will likely remain poor in the future? (Barrett and Carter, 2006).

Research on poverty has evolved over the past several decades with growing focus on how poverty changes through time and why certain populations remain entrenched in extreme poverty for sustained periods of time and others manage to climb out of the poverty (Barnett, Barrett and Skees, (2008). As research continued, the existence of populations fixed in poverty became known as a “poverty trap” (Barnett et al, 2008). Kraay and Mckenzie (2014) offer a definition for poverty traps as a “set of self-reinforcing mechanisms whereby

countries start poor and remain poor: poverty begets poverty, so that current poverty is itself a direct cause of poverty in the future.”

Easterly (2006) postulated that in the classic Solow model all countries have the same steady state which means they all converge to a high level of income. Yet, if saving is low at low levels of income, in comparison to high income, or if population growth is excessively high at a low income level then a poverty trap could occur at low levels of income which results in the equilibrium level being forced down to low or zero capital (Easterly, 2006). This results in a threshold capital stock above which populations could escape the poverty trap (Easterly, 2006). It is worth noting that the potential existence of multiple equilibria differs from the standard economic theory of convergence, toward a unique dynamic equilibrium, in the sense that a population can move toward either of two or more stable dynamic equilibria with the initial resources that they are endowed with affecting the one towards which they are attracted to (Barrett and Swallow, 2006).

This paper will focus empirically on poverty traps resulting from low savings rates at low levels of development – the savings based poverty trap. The savings based poverty traps were among the first to be formulated into articulated models whilst the simple idea behind the

(6)

5 theory is relatively basic: if a region or population are too poor to save, they are unable to accumulate capital and consequently their incomes can only grow at a rate of total factor productivity growth (Kraay and Mckenzie, 2014). Moreover, Kraay and Mckenzie (2014) stress that if this is the case, and if productivity growth is low, then income will be stagnant. In order to test the existence of the savings based poverty trap this paper uses a calibrated Solow Growth model with an exogenous savings rate. In terms of the Solow model variables, generating a poverty trap relies on the savings rate to be a steep S-shaped function of capital per capita (Kraay and Raddatz, 2007). An essential implication of this nonlinearity is the presence of multiple, stable dynamic equilibria, implying the existence of at least one unstable dynamic equilibrium which acts as a threshold point at which a populations’ path dynamics change (Lybbert, Barrett, Desta and Coppock, 2004). The dynamics of the multiple equilibria mechanism is critical in terms of growth theory and the economics of development (Lybbert et al. 2004).

Examining poverty traps in this way has significant implications for international policy. Kraay and Mckenzie (2014) explore this issue, in the sense that if multiple equilibria are possible then a one-time policy measure, such as foreign aid, may be able to push the population in question above the threshold capital stock level and enable them to escape the poverty trap. Additionally, this view has been emphasized by recent calls for further across the board debt relief and a large expansion of aid in order to support low income countries in achieving the MDG (Kraay and Raddatz, 2007). Nonetheless, although acknowledging that poverty would be relatively uncomplicated to eliminate if it only endured through the mechanisms captured in threshold models, Bowles, Durlauf and Hoff (2006) found that poverty programs and foreign aid donors have, in many cases, failed to reduce poverty. Hence, it can be said that for aid to prove effective at tackling poverty, the existence of multiple equilibria is fundamental, which is why this paper aims to examine the existence of multiple equilibria poverty traps, particularly those located in South America.

As Barrett and Swallow (2006) highlight, although it’s most prevalent in Africa as a share of population, extreme poverty has also gripped a significant portion of the population in South America. For example, figure 1 draws attention to the large levels of income

inequality in this region with no country managing above a 2.5% share of income held by the lowest 10% whilst Bolivia’s lowest 10% shared as little as 0.82% of income as recent as

(7)

6 0 0.5 1 1.5 2 2.5 1987* 2000** 2012 Income share in % Time period

Income share held by lowest 10%

Argentina Bolivia (*1990) Brazil (**1999) Chile Peru Paraguay (*1990, **1999) Venezuela (**1999) Ecuador Uruguay Figure 1

2012. Furthermore, the share of the population living in extreme poverty1, in the respective countries, can be seen in figure two. It is apparent that these percentages were extremely high in as recently as 1999, with four of the countries on which data was collected all having a share of over 30%. Although this proportion has decreased somewhat over time, in 2012 the data reveals that four countries still have over 10% of their population living in extreme poverty. Consequently, by aiming to substantiate whether poverty traps exist in South America, this paper will be able to conclude whether or not foreign aid will be able to act as the large scale shock needed for a population to grow out of the poverty trap. And

subsequently, with the aid of Solow Growth Model, answer the research question to what extent do savings-based poverty traps exist in South America.

To assess the evidence on whether savings based poverty traps exist in South America, this paper will proceed as follows: section 2 reviews existing literature on the subject of poverty traps including different models and empirical evidence already found; sections 3 follows with the empirical section of the paper which incorporates an outline of the dataset used, an explanation of the Solow Growth model and uses a regression to assess and analyse the empirical evidence for the existence of savings based poverty traps; lastly section 4 presents concluding remarks and poses questions for further research.

1_{Defined as living with less than $3.10 per day in 2011 international prices according to data from the world}

(8)

7

2. Literature Review

Economic theory has long since identified multiple sorts of poverty traps associated with different mechanisms by which they might emerge (Barnett, Barrett and Skees, 2008). Furthermore, these mechanisms have also been widely used to explain why similar aggregate shocks can experience different results (Ravallion and Jalan, 2001). Therefore in this section of the paper existing literature, on three different poverty trap mechanisms, will be reviewed. Namely, financial poverty traps, geographic poverty traps and nutritional poverty traps. From there, this paper will then examine in more detail the savings-based poverty trap which the empirical nature of this paper is based on.

2.1 Financial poverty traps

A portion of models surrounding poverty traps are focused on an interaction between capital market failures, which restrict the amount an individual can borrow (Kraay and Mckenzie, 2014). Market failures are essential to the existence of a poverty trap because if firms with low asset stocks could borrow freely, they would do so in order to reach and pass the critical poverty level thresholds and converge to the high-level equilibrium (Barrett and Swallow, 2006). When there is such financial constraints on multiple levels, for example

0 5 10 15 20 25 30 35 40 45 1987 1999 2012 % of population Time period

Poverty headcount ratio

ARG BOL (*1990) BRA CHL (**2000) COL PER PRY (*1990) VEN ECU SUR Figure 2

(9)

8 from individuals and households to national governments, they often become interdependent leading which can contribute to countrywide poverty traps (Barrett and Swallow, 2006).

Ravallion and Jalan (2002) conducted a test for nonlinearity in income and expenditure dynamics in rural China using a household-level data set spanning six years in four different provinces and discovered some evidence of nonlinearity in the dynamics. However, they did not find any evidence of low–level non-convexities therefore the data is not consistent with the existence of an unstable equilibrium for the poor (Ravallion and Jalan, 2002). On the other hand, research undertaken by Adato, Carter and May (2006) into the existence of poverty traps in South Africa, did identify a dynamic asset poverty threshold that indicates that many South Africans are trapped in poverty without a pathway out. What’s more, Santos and Barrett (2011) presented a discussion of the repercussions of bifurcated expected wealth dynamics for patterns of informal credit and unearthed support for the hypothesis that informal credit arrangements conform to this model. Other research has also found evidence of nonlinear expected wealth dynamics among a poor population in Ethiopia with two equilibrium levels - one above and the other below a threshold point (Lybbert, Barrett, Desta and Coppock, 2004). However, one must not forget that, regardless of the amount of

financing available, nations, individuals and regions with poor skills facing other constraints may still be stuck within a poverty trap (Kraay and Mckenzie, 2014).

2.2 Geographic Poverty Traps

Another example of a poverty trap is the geographical poverty trap. Jalan and Ravallion (2002) define this as characteristics of a household’s area of residence that result in the household’s consumption failing to rise over time, while an otherwise identical household living in a better-endowed area enjoys a rising standard of living. Data on the evidence of geographical poverty traps has thus far highlighted that location is vital in order to raise prospects of climbing out of poverty (Ravallion and Jalan, 2001). Furthermore, this is supported by research undertaken by Bloom, Canning and Sevilla (2003) who found that countries with favourable geography have relatively high income in the low-level equilibrium whilst hot, landlocked countries have very low levels of income at the low-level steady state. Additionally, it is difficult for these countries with unfavourable geography to break out of the geographic poverty trap and reach the high-level equilibrium (Bloom et al. 2003). Kraay and Mckenzie (2014) found that the evidence most consistent with this type of poverty trap

(10)

9 was observed when examining poor households in remote rural regions. This is a result of the seclusion which decreases the number of available production technologies which then adversely affects the choice between lower-income and higher-income outcomes (Kraay and Mckenzie, 2014).

Moreover, using a six-year panel of farm-household data for rural areas of southern China, Jalan and Ravallion (2002) obtained substantial evidence to support the analysis that living in a poor area lowers the productivity of a farm-household’s own investments resulting in a reduction of the growth rate of consumption, given restrictions on capital mobility. The results imply that, for the areas investigated in the given time period, there were areas so poor that consumption was actually falling. Whilst in almost identical households, in different more favourable geographic areas, a rise in consumption was experienced (Jalan and Ravallion, 2002). Therefore, it can be said that geography is significant and that the simple model of geographic determinism can be rejected in favour of a multiple equilibrium mode (Bloom et al. 2003).

2.3 Nutritional Poverty Traps

One of the original illustrations of multiple equilibria, S-shaped poverty trap is centred around nutrition levels (Kraay and Mckenzie, 2014). Barret and Swallow (2006) describe the nutritional poverty trap as one of the most extreme on a micro scale that simplistically comes around due to irreversible human capital accumulation as a result of infant undernutrition, illness, and a lack of education. In this model, poverty is self-reinforcing ensuing from a vicious circle of malnutrition. If individuals are too malnourished to be able to perform productive work then they won’t be able to earn sufficiently to buy food or produce enough to relieve this malnourishment (Kraay and Mckenzie, 2014). Furthermore, Dasgupta (1997) argued that poverty, as a result of the link between nutrition and productive capability, can be successional, for instance once a household descends into this poverty trap, it can be

particularly difficult for the resulting children to climb out of it. On a collective scale, this intuitive link between poor nutrition and productivity affects the ability of firms and societies to profit from human capital spillovers which in turn negatively affects aggregate

productivity and consumption growth (Barrett and Swallow, 2006). The crucial idea is that this link is nonlinear with increasing returns when one starts from a low consumption level (Kraay and Mckenzie, 2014). However, Kraay and Mckenzie (2014) also found that in reality nutritional poverty traps are questionable due to the fact that calories are too cheap in a large

(11)

10 part of the world and that, whilst nutrition does affect productivity; the shape of the

relationship tends not to follow the S-shaped pattern required for poverty trap dynamics.

2.4 Savings based poverty trap

There are a great deal of nonlinear dynamic models with multiple equilibria that have been used to explain the persistence of poverty and how otherwise matching aggregate shocks can have dissimilar outcomes (Ravallion and Jalan, 2001). Therefore, this section of the paper outlines some of the mechanisms and assesses the results that were found.

One such mechanism is the savings-based poverty trap. Easterly (2006) succinctly explains the mechanism behind this model in that poor people do not save enough resulting in physical capital accumulation failing to keep up with depreciation and population growth. Furthermore, he concedes that the prospect of low savings at low income levels is definitely credible (Easterly, 2006). However, although the theoretical argument has been welcomed for a considerable period of time, empirical evidence concerning this idea is more recent (Kraay and Mckenzie, 2014). Because this paper aims to replicate their empirical research it is worth thoroughly examining the work conducted by Kraay and Raddatz (2007).

Their paper aims to test the empirical relevance of the poverty trap view of underdevelopment which, in order to do so, they calibrate an aggregate growth model in which poverty traps can arise due to low savings (Kraay and Raddatz, 2007). In order to do so, the authors use a simple Solow growth model in which saving increases exogenously with the level of development and test this on Sub-Saharan African countries. As already

highlighted in this paper, a crucial condition for the existence of a savings based poverty trap is that savings rates are an S-shaped function of the level of development. However, Kraay and Raddatz (2007) found little evidence of this relationship. Theoretically, savings rates should start out flat when countries are poor then increase significantly over some immediate range before levelling off again (Kraay and Mckenzie, 2014). Whereas the empirical

relationship found that saving rates seem to be increasing at low levels of capital per worker, flat at intermediate levels and then increasing once more at high levels consequently meaning that the stable low-level equilibrium associated with poverty traps does not materialize (Kraay and Raddatz, 2007).

Therefore, these results question the argumentation that increases in aid will bring about sharp and sustained increases in growth which is why this research will be replicated in this

(12)

11 paper for the region of South America in order to be able to be able to justify increases in aid to the area.

3. Empirical research

This section examines the dataset used throughout the empirical research before

introducing the mechanism behind the savings-based poverty trap, the Solow growth model, and the variables used during the research. It then presents the empirical results and analyses them.

3.1 Data set

During the empirical research, data was gathered from two independent sources, namely the World Bank and the Penn World Table (PWT) 8.1, on eleven of the twelve South American countries2. The World Bank was used to gather the gross savings rates3 for each country in South America whilst the PWT was used to collect data on real Gross Domestic Product4 (GDP), population5, depreciation rate6 and capital stock7.

Data from the PWT covered the period from 1950 until 2011, with the exception of data from Suriname which started in 1970. However, figures on the gross savings rates for each country from the World Bank are slightly more recent with most countries registering data from around 1976 until 2011, with the notable exceptions of Colombia (1968-2011), Paraguay (1995-2011), Suriname (1977-2005) and Venezuela (1970-2011). Furthermore, some countries do not have fully complete data sets which mean the dataset is slightly distorted.

3.2 Economic theory and methodology

In order to test for the existence of savings-based poverty traps in South America, this paper will first use the Solow growth model to illustrate such a trap. The Solow growth

2

South American counties: Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Guyana Paraguay, Peru, Suriname, Uruguay, Venezuela. There was insufficient data available for Guyana so it hasn’t been included in the research.

3

Gross savings (%GDP) defined by the World Bank as gross national income less total consumption, plus net transfers.

4_{Real GDP at constant 2005 national prices (in mil. 2005US$)} 5

Population (in millions)

6_{Average depreciation rate of the capital stock} 7_{Capital stock at current PPPs (in mil. 2005US$)}

(13)

12 model uses a neoclassical aggregate production function in the form of a Cobb-Douglas function which expresses capital stock per effective worker as:

𝑘̇ = 𝑠(𝑘)𝐴𝑘∝− (𝑛 + 𝛿)𝑘

Where k is the capital stock per capita, A is an exogenous level of technology, α is the output elasticity of capital, and n and δ are the population growth rate and depreciation rates

respectively. The essential part of the model is the exogenous and constant savings rate as this assumption allows one to theoretically show the existence of a savings based poverty trap. In the model the savings rate is shown to be an exogenous function of the capital stock per capita. This assumption means that the savings rate is constant and low, 𝑘𝑙_{, until a}

threshold level of capital stock, 𝑘∗, is reached. Once this threshold is reached, the savings rate accelerates to a constant higher rate, 𝑘ℎ. For instance:

𝑠(𝑘) = {𝑠𝑙, 𝑘 < 𝑘

∗

𝑠_ℎ, 𝑘 ≥ 𝑘 ∗

Figure 3 shows the Solow Growth model for an economy suffering from a savings based poverty trap. This explanation allows for two steady states, 𝑘𝑙_{and 𝑘}ℎ_{, representing low}

and high rates of savings. The steady state at 𝑘𝑙_{can be seen as a poverty trap because if a}

country starts out below this point its marginal product of capital is high for low levels of capital so it grows until it reaches the steady state. Additionally, if a country starts out above the steady state its savings rate is too low meaning they will be unable to accumulate enough capital to climb out of the trap thus the resulting capital investment is lower than the

breakeven level needed to maintain a level capital stock and so capital stock per capita

decreases until it reaches 𝑘𝑙. On the other hand, if a country starts already above the threshold level but below 𝑘ℎ_{its savings rate is high enough such that investment is greater than the}

breakeven level of investment meaning capital stock per capita will increase until it reaches the higher steady state and vice versa. Intuitively, at low levels of income, and therefore low savings rates, high levels of investment cannot be reached without some sort of aggregate shock, for example foreign aid. As a result, a country is reduced to the lower steady state with low levels of capital stock per capita.

(14)

13 However, Kraay and Raddatz (2007) note that the low poverty trap equilibrium exists only by assumption. For example, suppose it transpires that the threshold point at which saving jumps to a higher rate occurs to left of what is currently presented as the poverty trap equilibrium. This would mean there would not be a poverty trap and a country starting out with a low capital stock per capita would grow until it reaches the high steady state (Kraay and Raddatz, 2007). Additionally, if the depreciation rate and/or population growth rates were lower this would result in the breakeven line rotating downwards and the low equilibrium potentially vanishing leaving only the high stable equilibrium (Kraay and Raddatz, 2007).

Now that the savings based poverty trap has been thoroughly explained this paper will perform a regression of capital per capita on a third order polynomial level for savings rates to see if and how it relates the two variables. After which, investment per effective labour is plotted against breakeven investment for each country with the available data to see if multiple equilibria exist and if they show signs of an existing poverty trap.

3.3 Empirical results and analysis

In this sub-section, this paper will analyse the empirical results obtained when investigating the relationship between capital per capita and the savings rate before

Figure 3 - The Solow growth model with poverty traps. (Adapted from Kraay and Raddatz (2007).

= Investment per effective labour

(15)

14 examining data corresponding to the Solow Growth model for the countries in question to assess whether multiple equilibria exist therefore showing signs of an existing poverty traps in South America.

To begin with, I do a cross-sectional data analysis to investigate the relationship between capital per capita and savings rate in South America. In order to see this potential non-linearity that might be present in the data, I ran a regression8 of the average savings rate per country on capital stock per capita on a third-order polynomial level. Crucially, remember that for a savings trap to exist we need savings rates to be an S-shaped function of capital per capita and therefore I expect to find low savings rates at low levels of capital per capita before a sharp increase over some transitional range before levelling out once again but at a higher rate. The graph in figure 4 plots the relationship found by the research, on the Y-axis is the relative savings rate averaged over all the available years from 1968 until 2011 and on the X-axis is capital stock per capita in 2011. What’s more, the third-order polynomial relationship has been added in order to draw attention to any nonlinearity that might be present in the data.

From figure 4 it is evident that the savings rate does not resemble the S-shaped curve that expected for the theoretical existence of a savings based poverty trap. Savings rates do increase somewhat over lower levels of capital per capital, however, after flattening out for levels of capital per capita between $32,000 and $34,000 savings actually start to decrease sharply. In the case of the countries examined in South America though, over the given time period available, we do not find statistically significant evidence of nonlinearities. Both the quadratic term and the cubic term fail to appear significantly. This may be due to the relatively small sample size with only eleven countries investigated, additionally, with respect to Suriname and Paraguay, both had little data available for their respective savings rates.

As a result, due to the shape of the curve and the statistical significance of the results, I can assume that in the case of South America there is not enough evidence for a strong positive non-linear relationship between savings rate and capital per capita. Consequently, in terms of the theory of savings based poverty traps the results are unfavourable. Bearing in mind for the theory to be correct savings rates need to be found to be low at low levels of capital per capita before sharply increasing at some threshold level and then levelling out at a

(16)

15 higher rate of capital per capita. Yet, the results found for South America contradict this theory and hence call into question large scale increases in aid to the region.

The next analysis I do is to see if, corresponding to the Solow Growth model already discussed, multiple equilibria exist that show signs of an existing poverty trap in each country. In order to do this I used the data from the PWT 8.1 and the World Bank and computed investment per effective labour9 and breakeven investment10. The results of these calculations can be seen in figures 5 and 6.

For a savings based poverty trap to exist we would have to find at least two equilibria in the data. In all of the results found we do find evidence of this, however, in each country we continuously find different levels of equilibria as capital per capita increases. This is harder to interpret in terms of the existence of savings based poverty traps within South America and correspondingly the aid implications. As such, if the idea behind foreign aid policies is to shock a country out of the savings based poverty trap towards a maintainable, higher level of capital per capita, the evidence does not provide sufficient reassurance that a sustainable level of capital per capita will be found. As a result, although capital aid might lead to an increase in capital per capita, the empirical evidence does not support that there is a recognisable reliable threshold of capital per capita that can be reached by aid to drive South America out of its current poverty condition.

9_{Investment per effective labour is defined as the savings rate multiplied by real GDP per capita.}

10_{Breakeven investment is defined as population growth plus depreciation multiplied by capital per capita.}

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 0 10000 20000 30000 40000 50000 60000 Sav in gs rate (% )

Capital per capita (k)

South America

_All

Argentina Bolivia Brazil Chile Colombia Ecuador Paraguay Peru Suriname Uruguay Venezuela Figure 4

(17)

16 0 500 1000 1500 2000 2500 3000 3500 66 54.39 80 76.77 96 63.83 10 843.9 3 13 737.3 2 17 629.9 7 27 83 6.6 2 44 497.8 9 42 847.8 9 39 196.9 6 44 939.7 9 49 030.4 1 k(t)

Argentina

sr*y (n+d)*k 0 200 400 600 800 1000 1200 38 35.77 41 05.29 35 56.02 35 42.53 37 55 .18 37 00.24 38 18.47 43 38.25 53 11.88 65 28.59 81 02.61 83 44.74 k(t)

Bolivia

sr*y (n+d)*k 0 500 1000 1500 2000 2500 3000 82 04 .70 98 92.18 11 508.8 5 13 543.9 0 15 067.0 5 17 404.0 8 21 065.1 8 24 814.1 8 31 006.6 0 33 015.0 1 k(t)

Brazil

sr*y (n+d)*k 0 500 1000 1500 2000 2500 3000 3500 10 426.3 5 11 392.6 5 12 510.8 2 12 40 6.6 6 14 407.2 6 18 870.5 9 22 561.2 4 24 752.9 0 33 803.7 1 41 393.1 3 k(t)

Chile

sr*y (n+d)*k 0 200 400 600 800 1000 1200 1400 1600 1800 11 15 5.1 3 13 620.1 9 14 159.1 2 14 251.5 4 14 526.5 7 14 513.8 2 14 469.0 0 17 195.9 2 17 998.4 7 20 102.1 0 25 869.9 0 k(t)

Colombia

sr*y (n+d)*k 0 500 1000 1500 2000 2500 78 64.27 94 37.51 10 852.7 7 10 397.0 9 10 94 8.8 9 11 431.7 4 12 041.7 7 13 293.7 9 15 576.6 5 19 494.9 7 26 006.4 1 27 105.0 6 k(t)

Ecuador

sr*y (n+d)*k Figure 5

(18)

17 0 200 400 600 800 1000 1200 k(t)

Paraguay

sr*y (n+d)*k 500₀ 1000 1500 2000 2500 3000 3500 58 79.80 66 62.61 75 21.66 80 85.99 86 34.73 10 54 9.9 2 13 862.0 9 16 562.4 4 18 143.4 2 20 685.8 2 25 451.0 4 28 629.8 2 k(t)

Peru

sr*y (n+d)*k -1000 -500 0 500 1000 1500 2000 2500 10 587.1 5 12 681.3 4 13 618.5 8 13 408.9 9 12 658.6 1 11 880.4 5 13 126.0 7 13 776.7 7 17 789.4 8 21 008.1 8 k(t)

Suriname

sr*y (n+d)*k ₀ 500 1000 1500 2000 2500 17 085.9 8 25 181.1 2 23 771.4 0 23 344.4 4 24 313.1 0 25 486.3 8 28 439.3 2 29 98 8.4 3 29 651.1 8 31 982.5 6 36 878.8 4 39 708.7 0 k(t)

Uruguay

sr*y (n+d)*k 0 1000 2000 3000 4000 5000 6000 19 652.3 5 24 105.7 0 32 325.8 9 35 162.3 0 29 51 2.2 5 26 253.7 7 24 518.8 2 24 765.4 2 25 715.9 9 29 766.0 2 31 200.5 4 k(t)

Venezuela

sr*y (n+d)*k Figure 6

(19)

18 3.4 Limitations

However, the empirical analysis that has been conducted does have important limitations. In the Solow Growth model used, saving rates are assumed to be exogenous which Kraay and Mckenzie (2007) describe as unreasonable and can result in counterfactual predictions

regarding the relationship between capital per capita and the savings rate. In addition to this, the analysis was restricted by the amount of available data, for example both Paraguay and Suriname had significantly less data on their savings rates thus resulting in unreliable results. Easterly (2006) explains another potential problem with assessing poverty traps in that poor countries may be poor due to bad policies and institutions which can complicate the

mechanism used to evaluate poverty traps. Another fundamental problem with poverty traps that rely on low capital per capita is that they imply extremely high returns to capital in poor countries which is not necessarily always the case (Easterly, 2006).

4. Conclusion

As this paper has addressed, there is large selection of literature providing theoretical models of poverty traps depicting a wide variety of self-reinforcing mechanisms through which poor countries will possibly stay poor. However, the literature is somewhat lacking in its empirical proof of such models, which could have serious repercussions given the greater emphasis on foreign aid policy, although Kraay and Raddatz (2007) did generate an empirical test to assess the existence of savings based poverty traps in sub-Saharan Africa.

Consequently, this paper attempted to replicate such methods and apply it the impoverished region of South America.

Within the empirical data analysis, the regression that was conducted in this paper does not exhibit the nonlinearities expected within the framework of multiple-equilibria savings based poverty traps leading to the result that there is no statistically significant evidence to support these mechanisms in South America.

What is more, when analysed through the Solow Growth model, multiple equilibria were found but it was also found that they do not increase in the correct non-linear way required to generate a stable low level equilibria and correspondingly a saving based poverty trap in the countries this paper examined. Consequently, this raises serious questions about the popular argument that large-scale increases in aid will have disproportionate effects on

(20)

19 economic growth in low income countries such as the one’s used in the empirical research. That is not to say foreign aid is not important for development, but more precisely we do not find any indication of the threshold levels needed in South America whereby large levels of aid can be justified as the “shock” needed to push a country above this level so they can experience sustainable growth.

For further research, additional empirical work could be carried out for more of the self-reinforcing mechanisms to assess whether they are empirically significant as

explanations of poverty traps. Likewise, a regression could be undertaken using several variables from the different mechanisms which may find then find evidence of the existence of a poverty trap which might in turn lead to implications in the way foreign aid is used.

(21)

20

References:

Adato, M., Carter, M. R., & May, J. (2006). Exploring poverty traps and social exclusion in South Africa using qualitative and quantitative data. Journal of Development Studies, 42(2), 226-247.

Barnett, B. J., Barrett, C. B., & Skees, J. R. (2008). Poverty Traps and Index Based Risk Transfer Products. World development, 36(10), 1766-1785.

Barrett, C., Carter, M. (2006). The Economics of Poverty Traps and Persistent Poverty: An Asset-Based Approach. Journal of Development Studies, 42(2), 178-199.

Barrett, C. B., & Swallow, B. M. (2006). Fractal poverty traps. World Development, 34(1), 1-15.

Bloom, D., Canning, D, & Sevilla J. (2003). Geography and poverty traps. Journal of Economic Growth, 8, 355-378.

Bloom, D. E., Sachs, J. D., Collier, P., & Udry, C. (1998). Geography, Demography, and Economic Growth in Africa. Brookings Papers on Economic Activity, 1998(2), 207. Bowles, S., Durlauf, S. N., & Hoff, K. (2006). Poverty traps. New York: Russell Sage

Foundation.

Dasgupta, P. (1997). Nutritional status, the capacity for work, and poverty traps. Journal of Econometrics, 77(1), 5-37.

Easterly, W. (2006). Reliving the 1950s: the big push, poverty traps, and takeoffs in economic development. Journal of Economic Growth, 289-318.

Jalan, J., & Ravallion, M. (2002). Geographic poverty traps? A micro model of consumption growth in rural China. J. Appl. Econ. Journal of Applied Econometrics, 17(4), 329-346.

Kraay, A., & Raddatz, C. (2007). Poverty traps, aid, and growth. Journal of Development

Economics, 82(2), 315-347.

Kraay, A., & Mckenzie, D. (2014). Do Poverty Traps Exist? Assessing the Evidence †.

Journal of Economic Perspectives, 28(3), 127-148.

Lokshin, M., & Ravallion, M. (2004). Household Income Dynamics in Two Transition Economies. Studies in Nonlinear Dynamics & Econometrics, 8(3).

Lybbert, T. J., Barrett, C. B., Desta, S., & Coppock, D. L. (2004). Stochastic wealth dynamics and risk management among a poor population*. The Economic Journal, 114(498), 750-777.

(22)

21 Ravallion, M., & Jalan, J. (2001). Household Income Dynamics in Rural China. Policy

Research Working Papers.

PWT Releases. (n.d.). Retrieved June 15, 2016, from http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1

Santos, P., & Barrett, C. B. (2011). Persistent poverty and informal credit. Journal of Development Economics, 96(2), 337-347.

World Development Indicators| World DataBank. (n.d.). Retrieved from http://databank.worldbank.org/data/reports.aspx?source=2

(23)

22