Economic Growth in
Seah Shuqi Gabriel (FXXXXXX)
Edwin Brouwer (XXXXXXX)
Macroeconomics: economic growth
Drs. X. XXX
Utrecht School of Economics
Universiteit Utrecht
Contents
1 Introduction 3
2 The Solow
Model 4
2.1Solw model 4
2.2 Modification of the data set 4
2.3
Instruments used 4
2.4 Instruments excluded 6
2.5 The Model 6
3 Procedure
and results 7
3.1 Procedure 7
3.2
Discussion of factors 9
3.3 Average growth rates and residuals 11
5 Conclusion 12
References 13
1 Introduction
The central
challenge facing African economies is to reduce poverty through higher levels
of economic growth. Long term, broadbased economic growth is essential for
SubSaharan
2 The Solow model
2.1 Solow model
In this paper we will try to explain economic growth in
Our regressions have the form:
gy = a + b0 · ln(y0) + b1 · I1 + b2 · I2 + . . .
Where gy is the growth of real percapita GDP during some period, ln(y0) is the natural logarithm of the initial level of percapita GDP, and Ix are the instruments which will be explained in a later section.
2.2 Modification
of the data set
Due to a lack of data, we had to
exclude some countries from our regressions. No data for 1960 GDP/capita was
available for
2.3 Instruments
used
In our model we used different
instrument to explain the growth of real percapita GDP in
Absolute latitude: We theorized that absolute latitude – distance
from the equator – would help explain differences in countries’ economic
growth, since a country’s latitude determines its climate. We also theorized
that countries’ proximity to the
Log(GDP/capita in 1960):According to the Solow model, ceteris paribus, countries which are further from their steady states will grow more quickly than those nearer to their steady states – beta convergence.
Log(Physical Capital per Capita in 1960):According to the Solow model, ceteris paribus, countries which have less physical capital per worker will accumulate capital at a faster rate than those with more physical capital. The former will thus grow more quickly than the later. If all differences between countries are explained by differences in capital per worker, we will have beta convergence as above.
Growth of Physical
Capital per Capita: According to the basic formulation of the Solow model,
differences in physical capital per capita explain countries’ different levels
of GDP/capita. If workers have more capital to work with, their average output
will naturally be higher.
Malaria Ecology: Malaria is a disease caused by a mosquito parasite
that affects many inhabitants of developing countries. It causes “high fevers,
shaking chills, and flulike illness”^{[4]}.
As might be expected from the symptoms, people afflicted with Malaria would
find it hard to work and study, if they were able to do so at all (either being
too sick to do so or losing their lives to the disease), and this would
severely impact economic growth.
In a 2004 paper[5], Sachs et al. came up with a Malaria Ecology Index based on a country’s geography and mosquito biology and behavior to measure how susceptible a country naturally is to Malaria. Ceteris paribus, a country with a higher Malaria Ecology Index would suffer from a slower rate of GDP/capita growth due to its people suffering from the effects of the disease.
Population growth: In the theory of the Solow growth model, population growth acts in the same way as capital dilution – each worker has less capital to work with and the country moves proportionately further from its steady state.
In reality, the
profile of the people a country gains through population growth is never
identical to the prior composition of its population. In developed countries,
rapid population growth resulting from more babies being born would dampen economic
growth in the years before the children join the workforce as they are educated
and their human capital is built up. In
Also, population
growth resulting from refugee flows would reduce economic growth more than
would be expected by capital dilution, since the sort of refugees who would
migrate to other African countries would be destitute; the well off would
likely migrate to safer, richer countries outside
2.4 Instruments excluded
For various reasons, we found some instruments to be unhelpful in our data analysis.
Malaria exposure (1994) had too many countries with values of 0 and 1, making us suspicious of the data. Data from one year was also, we felt, insufficient to capture the impact of malaria over 40 years of economic development. Malaria ecology, a constant value, was far more suitable as a regressor, since it would not change significantly in the time period studied.
Carbon Dioxide (CO2) emissions was another factor we decided not to use since it would be a proxy for development rather than a factor affecting it – more developed countries emit more carbon dioxide due to bigger manufacturing sectors, higher electricity consumption, higher motor vehicle use and other reasons.
It might have been informative to regress against labor's share of national income (1α) in the Solow Growth Model, but we were missing data for almost all countries.
National saving and average investment rates were also unused, since they would determine physical capital accumulation, which was what we actually regressed against.
2.5 The model
Explained the used instruments our regression will have the following form:
Growth of real percapita GDP (19602000) = a + b0 · Log(GDP/capita in 1960) + b1· Log (Physical Capital per Capita in 1960) + b2 · Growth of Physical Capital per Capita + b3 · Malaria Ecology + b4 · Population growth + b5 · Absolute latitude + u
Where ‘a’ is the meaningless constant term and ‘u’ are the unobservables.
In the next chapter we will evaluate the procedure and the results of our regression.
3 Procedure and
Results
3.1 Procedure
Setting the test size at 10%, we carried out our regressions. First, we used the dataset including data on physical capital accumulation. Regressing growth of GDP/capita from 19602000 on the logarithm of GDP/capita in 1960, absolute latitude, the growth of physical capital/capita, malaria ecology, population growth between 19602000 and the logarithm of Physical Capital/capita in 1960, we obtained:
Table 1: Solow model Africa
Dependent Variable:
GDP_GROWTH_60_00_BASIC 


Method: Least Squares 



Date: 03/30/06 Time: 14:34 



Sample (adjusted): 1
32 



Included observations:
32 after adjustments 












Variable 
Coefficient 
Std. Error 
tStatistic 
Prob. 










LOG(GDP_CAP__1960_BASIC) 
1.480231 
0.620415 
2.385873 
0.0249 
ABS_LAT 
0.077207 
0.037906 
2.036813 
0.0524 
GROWTH_OF_PHYS__CAP__POP 
0.027823 
0.040513 
0.686769 
0.4985 
MALARIA_ECOLOGY_GEOGRAPH 
0.031767 
0.029053 
1.093423 
0.2846 
POP_GROWTH_YR 
0.245797 
0.671848 
0.365853 
0.7176 
LOG(PHYS__CAP__POP__1960_FAC) 
0.475230 
0.248371 
1.913388 
0.0672 
C 
10.81991 
5.115542 
2.115105 
0.0446 










Rsquared 
0.377936 
Adjusted Rsquared 
0.228641 
Population growth was extremely insignificant (0,7176 > our test size of 10%), so excluding this you get:
Table 2: adapted Solow model Africa, without population growth
Dependent Variable:
GDP_GROWTH_60_00_BASIC 


Method: Least Squares 



Date: 03/30/06 Time: 14:39 



Sample (adjusted): 1
32 



Included observations:
32 after adjustments 












Variable 
Coefficient 
Std. Error 
tStatistic 
Prob. 










LOG(GDP_CAP__1960_BASIC) 
1.427070 
0.593027 
2.406418 
0.0235 
ABS_LAT 
0.080684 
0.036078 
2.236394 
0.0341 
GROWTH_OF_PHYS__CAP__POP 
0.025987 
0.039526 
0.657463 
0.5167 
MALARIA_ECOLOGY_GEOGRAPH 
0.032153 
0.028546 
1.126382 
0.2703 
LOG(PHYS__CAP__POP__1960_FAC) 
0.458742 
0.240145 
1.910270 
0.0672 
C 
9.782413 
4.186045 
2.336911 
0.0274 










Rsquared 
0.374606 
Adjusted Rsquared 
0.254338 






From this data, we can see that the
rate of accumulation of physical capital/capita is only significant at a 52%
test size. Thus, physical capital accumulation is not a significant factor in
explaining economic growth in
Excluding the rate of accumulation of physical capital/capita, we obtained:
Table 3: adapted Solow model Africa, without physical capital/capita
Dependent Variable:
GDP_GROWTH_60_00_BASIC 


Method: Least Squares 



Date: 03/30/06 Time: 14:43 



Sample (adjusted): 1
32 



Included observations:
32 after adjustments 












Variable 
Coefficient 
Std. Error 
tStatistic 
Prob. 










LOG(GDP_CAP__1960_BASIC) 
1.494514 
0.577913 
2.586052 
0.0154 
ABS_LAT 
0.081200 
0.035688 
2.275267 
0.0310 
MALARIA_ECOLOGY_GEOGRAPH 
0.029620 
0.027985 
1.058406 
0.2992 
LOG(PHYS__CAP__POP__1960_FAC) 
0.440382 
0.235995 
1.866066 
0.0729 
C 
10.68258 
3.913992 
2.729332 
0.0110 










Rsquared 
0.364208 
Adjusted Rsquared 
0.270017 
Since Malaria Ecology is only significant at 30%, we dropped it and ran the regression again:
Table 4: adapted Solow model Africa, without malaria ecology
Dependent Variable:
GDP_GROWTH_60_00_BASIC 


Method: Least Squares 



Date: 03/30/06 Time: 15:46 



Sample (adjusted): 1
32 



Included observations:
32 after adjustments 












Variable 
Coefficient 
Std. Error 
tStatistic 
Prob. 










LOG(GDP_CAP__1960_BASIC) 
1.600410 
0.570407 
2.805733 
0.0090 
ABS_LAT 
0.093512 
0.033812 
2.765678 
0.0099 
LOG(PHYS__CAP__POP__1960_FAC) 
0.464814 
0.235367 
1.974849 
0.0582 
C 
10.94374 
3.914584 
2.795633 
0.0093 










Rsquared 
0.337830 
Adjusted Rsquared 
0.266883 
The logarithm of physical capital per capita in 1960 is significant at a test
size of 6%, but the logarithm of per capita GDP in 1960 and absolute latitude
were significant at 1%.
The value for Rsquared shows that the variables used explain 33.8% of the variation in the growth of per capita GDP between the countries, the average value for which is 0.7694% per year. There is also strong evidence for beta convergence (which will be explained in a later in this chapter).
We suspected that using a value for Latitude10 might improve the accuracy of the regression, so we ran a new regression and got:
Table 5: adapted Solow model Africa, with Latitude –
10
Dependent Variable:
GDP_GROWTH_60_00_BASIC 


Method: Least Squares 



Date: 03/30/06 Time: 15:50 



Sample (adjusted): 1
32 



Included observations:
32 after adjustments 












Variable 
Coefficient 
Std. Error 
tStatistic 
Prob. 










LOG(GDP_CAP__1960_BASIC) 
1.101626 
0.546954 
2.014113 
0.0537 
LOG(PHYS__CAP__POP__1960_FAC) 
0.423489 
0.253906 
1.667900 
0.1065 
_LATITUDE_10_ 
0.056906 
0.026735 
2.128489 
0.0422 
C 
7.880784 
3.892344 
2.024688 
0.0525 










Rsquared 
0.274351 
Adjusted Rsquared 
0.196603 
Apparently, absolute latitude
explains economic growth in
3.2 Discussion of factors
Absolute latitude: We found that absolute latitude was the only instrument
that explained the steady state of a country in
Contrary to our predictions, regressing on
Latitude10 yielded a worse result than regressing on absolute latitude.
African countries further away from the Equator do better, as opposed to those
further away from the
Thus, it is not absolute latitude per se that explains economic growth in countries further from the equator, but the fact that these countries are more politically stable and enjoy better governance. Thought of in this way, the fact that absolute latitude affects a country’s steady state makes sense.
Log(GDP/capita in 1960): We found strong evidence for conditional beta convergence, since countries with a lower initial level of GDP/capita grew more quickly than those with higher initial levels, even when we only corrected for absolute latitude and malaria ecology; many other factors which we did not correct for affect countries’ steady states.
With a pvalue of 0.0090 which is way below our test size of 10%, there is very strong evidence for beta convergence among the countries; for every increase of 1 in the value of log(GDP/Capita in 1960), a country had slower growth to the tune of 1.6% per year from 19602000.
Log(Physical Capital per Capita in 1960): It
is interesting that the logarithm of the 1960 level of physical capital per
capita was a significant regressor – the more physical capital per capita a
country had in 1960, the faster it grew in the next 40 years. This result is
hard to explain, because Solow’s theory tells us the opposite. According to his
model, ceteris paribus, countries which
have less physical capital per worker will accumulate capital at a faster rate
than those with more physical capital. The former will thus grow more quickly
than the later. If all differences between countries are explained by
differences in capital per worker, we will have beta convergence as is the case
with Log(GDP/capita in 1960). In this case there is divergence between the
countries which we can not explain.
Growth of Physical
Capital per Capita: Was a relatively insignificant variable. This is probably
because economic activity in
Malaria Ecology: Malaria Ecology was only significant at the 30% test size – more significant than the other factors we dropped, but less so than those we kept. This is probably because the effect of malaria is ameliorated by international aid, and by the fact that countries with a naturally more serious malaria problem would do more to combat it.
Population growth (theory vs practice): The results showed that
population growth is not a significant factor affecting the growth of per
capita GDP. This shows that the composition of refugee inflows and newly born
generations is not radically different from that for the population as a whole.
3.3 Average growth rates and residuals
The average growth rate of GDP
per capita between 1960 and 2000 is 0,7694 percent.
The countries that perform better than expected are the countries that show a positive residual value (country 1 – country 13 in table 6). While the countries that are performing less than expected are those countries witch show a negative residual value (country 14 – country 32 in Table 6). So there are more countries that are performing below our expectations.
Table 6: Deviation of the different countries in
Africa
No 
Country 
Deviation 
No 
Country 
Deviation 
1 

3,83606 
17 

0,21882 
2 

3,09565 
18 

0,22415 
3 

1,7373 
19 

0,26167 
4 

1,69891 
20 

0,38075 
5 

1,22518 
21 

0,47207 
6 

1,19002 
22 

0,56832 
7 

0,69237 
23 

0,59631 
8 

0,68841 
24 

0,86276 
9 

0,66988 
25 

1,02551 
10 

0,63771 
26 

1,08612 
11 

0,21942 
27 

1,15139 
12 

0,19085 
28 

1,23524 
13 

0,02851 
29 

1,24394 
14 

0,14013 
30 

1,86078 
15 

0,15078 
31 

2,1021 
16 

0,19836 
32 

2,13109 
4
Conclusion
In this Paper we tried to explain economic growth in African countries by using Solow’s optimal growth model. We used his model because the model can tell us if there is any convergence between the countries, and which factors can explain the steady state of a country.
In our adapted Solow model we used different instrument to explain the growth of real percapita GDP in Africa. We included the natural logarithm of the initial level of percapita GDP in the model to find out if there where any beta convergence. According to the Solow model, ceteris paribus, countries which have less physical capital per worker will accumulate capital at a faster rate than those with more physical capital. The former will thus grow more quickly than the later. With our data we found there is very strong evidence for beta convergence among the countries.
But on the other hand including the natural logarithm of Physical Capital per Capita in 1960 leaded to divergence of African countries. The more physical capital per capita a country had in 1960, the faster it grew in the next 40 years.
Future more we included other variables as growth of physical capital per capita, malaria ecology, population growth and absolute latitude to see which instruments can change the steady state in African countries. We found that absolute latitude was the only significant instrument that explained the steady state of a country in Africa. And we think this is caused by the fact that countries further from the equator are more politically stable and enjoy better governance. If countries want to influence the level of there steady state they therefore have to stabilize their political situation (less civil wars) and improve their governance.
References
Fafchamps, Teal & Toye (2001), Towards a Growth Strategy for Africa.
Keller
& Poutvaara (2003), Do the Augmented
Solow Models Rule? A Contribution to the
Empirics of Human
Capital, R&D, and Economic Growt.
Kiszewski, Mellinger,
Spielman, Malaney and Sachs (2004), A
Global Index of the
Stability of Malaria Transmission. American
Journal of Tropical Medicine and
Hygiene, 70(5),
May, pp. 486498.
Nunn
(2005), Slavery, Institutional
Development, and LongRun Growth in Africa, 1400–
2000, pp. 37
WebPages:
http://www.cdc.gov/malaria/faq.htm
http://www.ppu.org.uk/war/countries/africa/africa_index.html
[1] Nunn. 2005, Slavery,
Institutional Development, and LongRun Growth in
[2]
Fafchamps, Teal & Toye. 2001, Towards a Growth Strategy for
[3] http://www.cebr.dk/upload/cebr_dp_2003_13_final.pdf
[4] http://www.cdc.gov/malaria/faq.htm
[5] "A Global Index of the Stability of Malaria Transmission," with Anthony Kiszewski, Andrew Mellinger, Andrew Spielman, Pia Malaney, and Sonia Ehrlich Sachs. American Journal of Tropical Medicine and Hygiene, 70(5), May 2004, pp. 486498.
[6] http://www.ppu.org.uk/war/countries/africa/africa_index.html
[7] Depending on how strict one is with test sizes, Malaria could also influence a country’s steady state, but we chose to drop it.