Munich Personal RePEc Archive
Does Government Expenditure on
Education Promote Economic Growth?
An Econometric Analysis
Abhijeet, Chandra
Jamia Millia Islamia (Central University), New Delhi
August 2010
Online at https://mpra.ub.uni-muenchen.de/25480/
MPRA Paper No. 25480, posted 28 Sep 2010 20:12 UTC
Abhijeet Chandra1
Education being an important component of human capital has always attracted the interests
of economists, researchers and policy makers. Governments across the globe in general and
in India in particular are trying to improve the human capital by pumping more investments
in education. But the issue that whether improved level of education resulting from more
education spending can promote economic growth is still controversial. Some economists and
researchers have supported the bi-directional relation between these two variables, while it
has also been suggested that it is the economic growth that stimulates governments spend
more on education, not the other way. Considering this research issue, the present paper uses
linear and non-linear Granger Causality methods to determine the causal relationship between
education spending and economic growth in India for the period 1951-2009. The findings of
this paper indicate that economic growth affects the level of government spending on
education irrespective of any lag effects, but investments in education also tend to influence
economic growth after some time-lag. The results are particularly useful in theoretical and
empirical research by economists, regulators and policy makers.
Keywords: Education expenditure, Economic growth, Indian economy, Granger Causality,
Non-linearity.
JEL Classification Codes: C22, E62, H52, I21.
The author is extremely grateful to Dr. Ravinder Kumar (JMI, New Delhi) and Prof. R. P.
Pradhan (IIT Kharagpur) for introducing him to the Econometric methodology of economic & finance research.
He also gratefully acknowledges the financial support extended by Jamia Millia Islamia (Central University),
New Delhi during his doctoral studies at the university.
1
Doctoral Candidate – Finance, Department of Commerce & Business Studies, Jamia Millia Islamia (Central
University), New Delhi-110 025. E: [email protected].
It is well known and widely accepted that investment in education is critical for
economic growth and social cohesiveness of society. Many of the potential payoffs to society
from various types of public investment in education are not immediately apparent but are
nevertheless important. One of the best examples relevant to the Indian context is the much
hyped software boom that itself reflects at least partly the earlier public investment in Indian
Institutes of Technology (IITs). Further, there are huge advantages to society in improving
the general level of education, not only because the quality of workforce improves, but
because various other aspects such as health, nutrition and sanitation are positively affected,
and also because educated citizens can be more effective participants in a democratic civil
society.
Expenditure on education is supposed to bring into the economic system the
externalities and other indirect effects such as higher education attainment and achievement
of children, better health and lower mortality of children, better individual health and lower
number of birth which subsequently cause higher productivity in terms of increased earnings,
more participation in the labour force i.e. increased labour force; all these coupled with lower
population growth and better health of population tend to positively influence higher
economic growth (Michaelowa, 2000). The relationship between economic growth and
various macroeconomic factors has attracted the interest of many economists and policy
makers since long ago. The history of the issue led back to the era of the classical economist
Adam Smith, followed by neoclassical economists such as Alfred Marshal and Henry Schultz
(Tilak, 2005). The macroeconomists has concentrated on the effects of several government
policies on the sustainable economic growth. This emphasis can be attributed to the
recognition of the fact that the difference between prosperity and poverty in a country
depends on how quickly it grows over the long term. Although all the standard
macroeconomic policies are important for economic growth, understanding their individual
impact on the economic growth is even more significant.
Economics offers a variety of theories and models relating education to economic
growth. Education increases an individual’s earning potential, but also produces a ‘ripple
effect” throughout the economy by way of series of positive externalities. Katharina
Michaelowa (2000) of the Hamburg Institute for International Economics diagrams the
impact of education on both micro and macro level as follows (refer Figure 1):
Figure 1: Micro and Macro Level Effects of Education on Economic Growth
(Source: Michaelowa, Katharina (2000), “Returns to Education in Low Income Countries: Evidence for Africa)
This paper is an attempt to understand the relationship between government
expenditure on education and economic growth in Indian context. Using time series data from
1951 through 2009 on education expenditure and economic growth in India, the researcher
examines the causation between these two factors. The empirical methodology adopted for
this purpose includes the Granger Causality test within an error-correction framework. The
findings suggest strong evidence for a unidirectional causality from economic growth to
education expenditure, but moderate evidence of causality from education expenditure to
economic growth is observed.
Rest of the paper is organised as follows: next section is dedicated on the concentrated
review of concerned literature on the issue. Following section consists of detailed
methodological issues and data description, followed by analysis of test results and
discussions and interpretation of the findings. The last section concludes with summarising
remarks.
!"
In the empirical literature, relation between public spending and economic growth has
found much attention of economists and researchers in public economics and finance. O’Neill
(1995) reports in his findings that convergence in education levels have resulted in a
reduction in income dispersion. He further states that for the world as a whole, income has
diverged despite substantial convergence in education levels. This is a result of increases in
the return to education that favour the developed countries at the expense of the less
developed countries. Sylwester (2000) explores the transition mechanism that might link the
income inequality and economic growth. He found that public education expenditures are
positively associated with future economic growth, although the contemporaneous effect
upon growth is negative. Barro (2001) examines a panel data of around 100 countries
observed from 1965 to 1995 and finds that growth is positively related to the starting level of
average years of school attainment of adult males at the secondary and higher levels. Growth
is insignificantly related to school attainment of females at the secondary and higher levels,
and also to male schooling at the primary level.
Blankenau et al (2005) carried out an empirical study on expenditure–growth
relationship in the context of an endogenous growth model. They found that the response of
growth to public education expenditure may be non-monotonic over the relevant range. The
relationship depends on the level of government spending, the tax structure and the
parameters of production technologies. Review of extensive literature in this respect is
beyond the scope of the present paper, the researcher, therefore, focuses on only most
relevant and contemporary studies on the relationship of education expenditure and economic
growth. The literature has focused on the link between level of public expenditure on
education and economic growth; majority of the studies deal with endogenously generated
economic growth and stress on the role of human capital accumulation in economic growth
(Chakraborty, 2005), and that an investment in education is very beneficial to the society,
both at the micro level as well as macro level and affects the economic growth both directly
and indirectly (Dahlin, 2005).
In their attempt to determine the causation between education expenditure and
economic growth, most of the researchers applied linear Granger Causality tests, which have
proved to be essential for identifying the predictive ability of the time series models
(Alexakis and Siriopoulos, 1999). More specifically, all studies on causal relationships rely
exclusively on traditional linear Granger Causality tests with error correction models, though
their proxy for education varied from public expenditure on education to school enrolments,
to school attainment age and so on. A survey of the literature reveals that there is much
controversy as far as the nature and the direction of causality between education spending and
growth is concerned.
The uniqueness of the present study lies in the fact that it uses the actual government
spending on education and gross domestic product at current prices at its two variables and
run the linear and non-linear Granger Causality tests in order to understand the patterns of
relationship between these variables. Detailed methodology is discussed in the following
section.
#
In this section, details of the linear and the non-linear Granger Causality tests are
discussed followed by the data variables and their respective sources are presented. Mainly
the Granger Causality tests (Granger, 1969), and the statistical technique – developed by
Baek and Brock (1992) and modified by Hiemstra and Jones (1994) – are used to test for
linear and non-linear Granger Causality relationships respectively.
The empirical investigation has been carried out in the case on Indian economy with a
dataset of the period 1950-51 to 2008-09. The present study uses the secondary data which
have been collected from the National Accounts Statistics, Central Statistical Organisation,
Ministry of Statistics and Plan Implementation, Government of India, New Delhi
(http://www.mospi.gov.in) and the Budget Expenditure on Education, Department of
Secondary and Higher Education, Government of India, New Delhi
(http://www.education.nic.in/secondary.htm). The reliability of the data for empirical
research can be attributed to the fact that all the data sources used in this study are
government sources and thus, data is very much reliable and perfect for policy research.
The data variables used in the present study are government expenditure on education
(henceforth referred to as EDEX) and gross domestic product at current prices (GDP). The
researcher attempts to test the direction of causation between these two variables. All data
used are expressed in current prices. The use of current prices is incumbent in the case of
non-linear causality tests. This is because the transformation in constant prices filters time-
series and the results are distorted. For reasons of mathematical consistency, filtering should
be avoided. The cross correlation between the two variables are given in the Table (3).
The test is to reject the null hypotheses of “GDP does not Granger cause EDEX” and
“EDEX does not Granger cause GDP” against alternative hypotheses of bidirectional
Granger causality between these two variable. Statistical description of the relationship is
described in the following sub-section.
The Linear Granger Causality Test: A time-series xt causes another time series yt in the
Granger sense if the present value of y can be predicted better using past values of x than by
not doing so, considering also other relevant information, including, past values of y. in
mathematical terms, x is said to cause y, provided some βj is non-zero in the full regression
equation (1):
r s
yt = δ 0 + ∑ ai yt −i + ∑ β j xt − j + ε t (1)
i =1 j −1
The relevance of x is indicated when comparing the error in equation (1) to that of the
reduced equation (2):
r
y t = δ 0 + ∑ a i y t −i + ε (2)
i =1
The error terms are compared formally with the F-statistics.
The Non-linear Granger Causality Test: A non-parametric statistical model is proposed by
Baek and Brock (1992) for detecting non-linear causal relations that is beyond the scope of
standard linear tests. They follow an approach that employs the correlation integral, which
provides an estimate of spatial dependence across time. For instance, consider two stationary
and weekly dependent time-series {Xt} and {Yt}, t = 1, 2, 3, …., n. Let the m-length lead
vector Xt be designated by X tm , and the Lx-length and the Ly-length lag vectors of Xt and Yt be
Ly
designated by X tLx
− Lx and Yt − Ly , respectively.
For given values of m, Lx, and Ly ≥ 1 and for e >0, Y does not strictly Granger Cause X if:
Pr( X tm − X sm < e X tLx− Lx − X sLx− Lx < e, Yt −LyLy − YsLy
− Ly < e)
= Pr( X tm − X sm < e X tLx− Lx − X sLx− Lx < e) (3)
where Pr (.) denotes probability and . denotes the maximum norm (the maximum norm for
Z ≡ (Z1, Z2, …, ZK) is defined as max (Zi), i = 1, 2, 3, …, K).
The probability on the left-hand side of the above equation is the conditional
probability that the two arbitrary m-length lead vector {Xt} are within a distance e of each
other, given that the corresponding Lx-length lag vectors of {Xt} and Ly-length lag vectors of
{Yt} are within e of each other. The probability on the right-hand side of the equation is the
conditional probability that two arbitrary m-length lead vectors of {Xt} are within a distance e
of each other, given that their corresponding Lx-length lag vectors are within a distance e of
each other.
For testing of non-linear Granger causality, first it requires to remove the linear
dependence. For this reason, a Vector Autoregression (VAR) model is applied and the
estimate residuals are used to test for non-linear causality. Following VAR model is
estimated for our sample dataset, where εi,t is the innovation at time t and p the lag length:
p p
GDPi ,t = ∑ δ i ,k GDPi ,t − k + ∑ β i , k EDEX i ,t − k + ε i ,t (4a)
k =1 k =1
p p
EDEX i ,t = ∑ χ i ,k GDPi ,t − k + ∑ γ i ,k EDEX i ,t − k + u i ,t (4b)
k =1 k =1
A significantly positive test statistics of the coefficients in the above equations
suggest that lagged values of EDEX help to predict GDP and also lagged values of GDP help
to predict EDEX, whereas a significant negative value of the coefficients suggests that
knowledge of the lagged values of EDEX and GDP confounds the prediction of GDP and
EDEX respectively. For this reason, Hiemstra and Jones (1994) argue that the test statistics in
the above equation should be evaluated with right-tailed critical values when testing for the
presence of granger causality. In order to test for non-linear Granger causality the above test
is applied to the two estimated residual series from the VAR models.
The trend of India’s economic growth (expressed as GDP) and public spending on
education (EDEX) during the sample period of 1950-51 and 2008-09 is represented in Figure
(2) below. It reflects that there exists a gap in the linearity in the relationship between the two
variables.
Figure 2: Trend in India’s GDP and Education Expenditure
The foremost step of the Granger Causality test is to perform stationarity tests for
each of the relevant variables. There have been a variety of proposed methods for
implementing stationairity tests and each has been widely used in the applied economics
literature. Tests of stationarity that have become popular over the past several years is the
Unit Root test and Dickey-Fuller test. However, there is now a growing consensus that the
stationarity test procedure due to the Dickey and Fuller (1979) has superior small properties
compared to each alternative. The present study, therefore, employs the augmented Dickey-
Fuller (ADF) test procedure for implementing stationarity tests. The ADF test consists of
estimating the following regression:
m
∆Yt = β 1 + β 2 t + δYt −1 + ∑ α i ∆Yt −i +ε t (5)
i =1
where εt is a pure white noise error term and where ∆Yt-1 = (Yt-1 – Yt-2), ∆Yt-2 = (Yt-2 – Yt-3), etc.
The number of lagged difference terms to include is often determined empirically, the idea
being to include enough terms so that the error term in the above equation (5) is serially
uncorrelated. In ADF, we test whether δ = 0. The ADF statistics in the present case suggest
that all variables are integrated of order one, I(1), whereas the first difference is integrated of
order zero, I(0). In the null hypothesis, the examined variable has a unit root which means
that it is non-stationary. Following the above procedure, the series have been proved to be
stationary in the first differences.
The Granger Causality test is applied in order to test for the causal flow between
government expenditure on education and economic development, expressed by the Indian
GDP. The null hypothesis states that no Granger causality exists; thus no linear relationship
between education expenditure and GDP is observed in India from 1951 to 2009. On the
other hand, the alternative hypothesis suggests that linear Granger causality exists. A
bidirectional flow of causality indicates that as the public spending on education grows there
is a tendency for economic growth to further improve.
To test whether the Granger causality exists or not, we need to compare the
probability that the null hypothesis exists with the critical value. If the critical value is greater
than the probability, the null hypothesis stands to be rejected and the alternative hypothesis
gets accepted. In case of the probability value is greater than the critical value, the null
hypothesis is considered as significant and we accept it as the true case.
Table 1: Granger Causality Test Statistics
$ % &
EDEX does not Granger Cause GDP GDP does not Granger Cause EDEX
Lags Obs F-Stats P-value Obs F-Stats P-value
1 58 5.10987 0.0278 58 38.185 8.00E-08
2 57 1.13655 0.3288 57 51.1819 5.00E-11
3 56 1.85007 0.1504 56 34.165 5.00E-12
4 55 1.4438 0.2348 55 22.9531 2.00E-10
5 54 1.00368 0.4272 54 26.288 5.00E-12
6 53 6.49722 8.00E-05 53 23.0422 1.00E-11
7 52 10.9997 2.00E-07 52 22.1279 2.00E-11
8 51 8.90142 2.00E-06 51 20.4525 7.00E-11
Obs: No. of Observations included in the test.
As can be inferred from the test statistics in the table (1) above, there are a significant
indication of a linear unidirectional causal relationship running from economic development
(expressed as GDP) to the education spending (EDEX). Thus, in India, GDP determines
public spending on education in a linear way. But at the same time, it is also observed that
with a lag value of 6 or more, this causal relationship between public education expenditure
and economic growth becomes bidirectional; in other word, the causation runs from GDP to
EDEX and also from EDEX to GDP, as the P-values in both cases are less than the assumed
critical values. It can be, therefore, said that public expenditure on education in past years is
translated into and affects to some extent the economic growth.
In the light of the reported empirical results, it may be assumed that the growth of
education expenditure by the government in India is dependent on and determined by
economic growth to great extent, but the impact of education expenditure of past years on the
economic growth also exists in our test results. Thus, there exists a bidirectional relationship
between these two variables. The results of this study are comparable to those of other
researchers. The direction of causation running from economic growth to education
expenditure is identified by Blankenau and Simpson (2003), Bose, Haque and Osbon (2003),
Basu and Bhattarai (2009) and Pradhan (2009), while the findings of causality running from
education spending to economic growth are in lines with those of Al-Yousif (2005),
Jiranyakul (2007), and Parmani (2009).
The model suggested by Baek and Brock (1992) and revised by Hiemstra and Jones
(1994) is employed in order to test the non-linearity in the causal relationship between
education expenditure and economic growth. It is important to mention that implementing
that model requires a choice of values for various parameters such as the lead length, m, the
lag lengths Lx and Ly, and the scale parameter, e. In this study, we have adapted the approach
as followed by Karagianni and Pempetzoglu (2007) and set the lead length at m = 1 and Lx =
Ly and a common lag length of 1 to 8 lags (the choice of the specific lag length is proposed
by Kyrtsou and Labys, 2006). A common scale parameter of e = 1.5σ, where σ denotes the
standard deviation of the standardised time series. The test statistics are presented in the
Table (2) below:
Table 2: Non-linear Granger Causality Test Statistics
$ % &
EDEX →GDP GDP →EDEX
Lx = Ly Cs T-value Cs T-value
1 0.0241 1.0278 0.0201 1.134
2 0.0197 1.0288 0.0182 1.732*
3 0.0148 1.1074 0.0116 1.996*
4 0.0113 1.2348 0.0092 2.008*
5 0.0096 1.5272* 0.0076 2.102*
6 0.0093 2.0134* 0.0076 2.102*
7 0.0093 2.0134* 0.0076 2.102*
8 0.0093 2.0134* 0.0076 2.102*
Note: * statistically significant at 5% level.
The critical value for 5% is 1.523.
As can be inferred from the test statistics of non-linear Granger causality test, there
exists a bi-directional causal relationship between the two variables. The results support to a
great extent the findings of linear Granger causality tests. Here also, we can see that
education expenditure is dependent on the country’s economic growth, that is the causation
runs from economic growth to education expenditure during the sample period. But it is also
noteworthy to observe that with a lag value of 5 or more, the direction of causality between
these two variables become dynamic. In simple words, a shock in education expenditure is
expected to affect economic growth, even after some time, in a non-proportional way, due to
the non-linear causality.
The relationship between government’s expenditure on education and country’s
economic growth being non-linear makes it difficult to predict them with accuracy and
precision, and also it is not possible to evaluate the extent and magnitude of the impact
caused by the shocks in one variable on the other variable and vice versa.
&
The present paper made an attempt to explore the causal relationship between
government spending on education and economic growth in respect of India employing a
Granger Causality test with both a linear as well as a non-linear model framework. The
period for which data have been used in this study is 1950-51 through 2008-09. The data
sources used for this purpose are the concerned departments of the Government of India. The
linear framework of Granger causality model proposed by Granger (1969) and non-linear
models of Granger Causality test suggested by Baek and Brock (1992) and revised by
Hiemstra and Jones (1994) are employed in order to test the relationship between the two
macroeconomic variables. This study also relies for the methodological issues to much extent
on the study carried out by Kyrtsou and Labys (2006).
The empirical findings of this study provide with the following conclusions: first, the
time-series data used in the present study are found to be non-stationary at the level data, but
stationary after first differences, indicating that they are integrated of order one (Pradhan,
2009). Second, there is a strong support for the observation that the causation between
education expenditure and economic growth is bi-directional, i.e. the causality runs from
economic growth to education expenditure and vice versa. Third, the results also show that
the direction of causation is from education expenditure to economic growth is not immediate
to take effect, rather it can be said that investment in education is expected to affect economic
growth of a country after some period, 5 or 6 years in present study. Fourth, economic
growth has always remained the major influencing factor as a determinant of education
expenditure made by any government as obvious from the relevant literature. In the case of
present study also, the causality running from economic growth to education expenditure is
persistent irrespective of lead or lag values. Finally, it is also observed that more studies
comprising of cross-countries (especially developing countries) time-series data could
contribute to the better and improved understanding of the relationship of economic growth
and investments in education.
Since, education is an important constituent of the human capital, improved education
scenario certainly influences a country’s economic growth. Governments should feel the need
to focus on increased investments in education which contributes to economic growth, both
directly and indirectly. The findings of the present study may be helpful to future theoretical
and empirical research on the relation between these specific variables, by warning policy
makers to pay attention to the shocks they induce in the economy (Kyrtsou and Labys, 2006).
Al-Yousif, Yousif Khalifa (2005), “Education Expenditure and Economic Growth: Some
Empirical Evidence from the GCC Countries”, United Arab Emirates University
Working Paper.
Alexakis, P. and C. Siripoulos (1999), “The International Stock Market Crisis of 1997 and the
Dynamic Relationships between Asian Stock Markets: Linear and Non-linear Granger
Causality Tests”, Managerial Finance, Vol. 25, No. 8, pp. 22-38.
Baek, E. and W. Brock (1992), “A General Test for Non-linear Granger Causality: Bi-variate
Model” Iowa State University and University of Wisconsin at Madison Working Paper.
Barro, Robert J. (2001), “Human Capital and Growth”, The American Economic Review, Vol.
91, No. 2, pp.12-17.
Basu, Parantap and Keshab Bhattarai (2009), “Does Government Spending on Education
Promote Growth and Schooling Returns?”, Durham University and Hull University
Working Paper.
Bose, Niloy, M. Emranul Haque and Denis R. Osborn (2003), “Public Expenditure and
Economic Growth: A Disaggregated Analysis”, University of Wisconsin Working
Paper.
Blankenau, William E., Nicole B. Simpson and Marc Tomljanovich (2005), “Public
Education Expenditure, Taxation and Growth: Linking Data to Theory”, Kansas State
University Department of Economics Working Paper.
Chakraborty, B. (2005), “Human Capital, Education Policy and Economic Growth”,
Productivity, Vol. 46, No. 1, pp. 13-20.
Dahlin, B. G. (2005), “The Impact of Education on Economic Growth: Theory Findings and
Policy Implications”, Duke University Working Paper.
Granger, C. W. J. (1969), “Investigating Causal Relations by Econometric Models and Cross-
spectral Methods” Econometrica, Vol. 37, No. 3, pp. 424-438.
Gujarati, D. N. (2003), “Basic Econometrics” McGraw Hill New York.
Hiemstra, C. and J. D. Jones (1994), “Testing for Linear and Non-linear Granger Causality in
Stock Price-Volume Relation”, Journal of Finance, Vol. XLIX, No. 5, pp. 1639-1664.
Jiranyakul, Komain (2007), “The Relationship between Government Expenditure and
Economic Growth in Thailand”, Journal of Economics and Economic Education
Research, Vol. 8, No. 1, pp. 93-102.
Karagiannni, Stella and Maria Pempetzoglu (2009), “Defense Spending and Economic
Growth in Turkey: A Linear and Non-linear Granger Causality Approach” Defense and
Peace Economics, Vol. 20, No. 2, pp. 139-148.
Kyrtsou, C. and W. C. Labys (2006), “Evidence for Chaotic Dependence between US
Inflation and Commodity Prices”, Journal of Macroeconomics, Vol. 28, pp. 256-266.
Michaelowa, K. (2000), “Returns to Education in Low-income Countries: Evidence for
Africa. http://www.hwwwa.de/projects/ResProgrammes/RP/Development
Processes/Vfs EL 2000 Rev2.pdf.
O’Neill, Donal (1995), “Education and Income Growth: Implications for Cross-country
Inequality”, The Journal of Political Economy, Vol. 103, No. 6, pp.1289-1301.
Parmani, Risti (2009), “The Role of Education in Economic Growth in East Asia: A Survey”,
Asia-Pacific Economic Literature.
Pradhan, R. P. (2009), “Education and Economic Growth in India: Using Error Correction
Modelling”, International Research Journal of Finance and Economics, Issue 25, pp.
139-147.
Sylwester, Kevin (2000), “Income Inequality, Education Expenditure and Growth”, Journal
of Development Economics, Vol. 63, pp. 379-398.
Tilak, J. B. G. (2005), “Post-elementary Education, Poverty and Development in India”,
Center of African Studies University of Edinburgh Working Paper No. 6.
% ' $ ($ ) ! * (Source: Author’s
Calculations using E-Views 6)
Sample: 1951 2009
Included observations: 59
Correlations are asymptotically consistent approximations
EDEX,GDP(-i) EDEX,GDP(+i) i lag lead
. |********** . |********** 0 0.9978 0.9978
. |*********| . |*********| 1 0.8598 0.8654
. |******** | . |******** | 2 0.7517 0.7508
. |******* | . |******* | 3 0.6571 0.6561
. |****** | . |****** | 4 0.5762 0.5862
. |***** | . |***** | 5 0.5066 0.5328
. |**** | . |***** | 6 0.4466 0.4833
. |**** | . |**** | 7 0.3937 0.4309
. |*** | . |**** | 8 0.3412 0.3795
. |*** | . |*** | 9 0.2911 0.3141
. |**. | . |**. | 10 0.2427 0.2500
. |**. | . |**. | 11 0.1967 0.1974
. |**. | . |**. | 12 0.1560 0.1587
. |* . | . |* . | 13 0.1174 0.1217
. |* . | . |* . | 14 0.0834 0.0886
. |* . | . |* . | 15 0.0539 0.0600
. | . | . | . | 16 0.0280 0.0346
. | . | . | . | 17 0.0051 0.0110
. | . | . | . | 18 -0.0158 -0.0114
. | . | . | . | 19 -0.0347 -0.0322
. *| . | . *| . | 20 -0.0516 -0.0516
. *| . | . *| . | 21 -0.0669 -0.0683
. *| . | . *| . | 22 -0.0801 -0.0830
. *| . | . *| . | 23 -0.0924 -0.0954
. *| . | . *| . | 24 -0.1041 -0.1076
*****
