Macro Determinants of Total Factor Productivity Growth of Agriculture in Pakistan

Asghar Ali*, Khalid Mushtaq**, Muhammad Ashfaq***, Abedullah**** and P.J.Dawson*****1

Address for Correspondence

Dr. Khalid Mushtaq,

Assistant Professor,
Department of Agricultural Economics,
University of Agriculture Faisalabad,
Pakistan

Tel: +92 (0)41 9200161-69 Ext: 2802

Email: khalidmushtaq@uaf.edu.pk

1*, **, ***

Lecturer, Assistant Professor and Professor respectively, Department of
****
Agricultural Economics; Assistant Professor, Department of Environmental &
*****
Resource Economics, University of Agriculture Faisalabad; and Reader in Agricultural
Economics, School of Agriculture, Food and Rural Development, University of Newcastle
Upon Tyne, UK.

0
 Macro Determinants of Total Factor Productivity Growth of Agriculture in Pakistan

Abstract

The role of productivity in accelerating the pace of economic growth is well recognized in the
literature. With continual population growth, a diminishing supply of per capita arable land,
limits to further expansion of cultivated land and slowing returns to further input
intensification, there is growing need for food supply increases that could only originate from
productivity growth rather than increase in inputs The present study investigated the impact
of different macro variables on Total Factor Productivity (TFP) of agriculture in Pakistan by
employing cointegration analysis analysis for the period from 1971 to 2006. The results
indicated that human capital, infrastructure development and credit resources were positively
associated with TFP of agriculture. Openness of agricultural economy observed a significant
positive impact on productivity. Macroeconomic stability influenced TFP growth negatively
and significantly. Real per capita income indicated positive but insignificant relationship with
productivity growth. The strong two way Granger-causality was observed between
productivity and human capital development; and infrastructural development. Overall the
results explained that policies which promote human capital, increase credit resources in
agriculture, improve infrastructure development, facilitate openness of agricultural economy,
ensure macroeconomic stability and rise in real per capita income; will improve productivity
and competitiveness of Pakistan agriculture.

Keywords: Tornqvist-Thiel, Total Factor Productivity, Competitiveness, Cointegration.

1. Introduction
The average annual growth of about 3.46 percent in agriculture over the last six
decades has exceeded the population growth rate of about 2.58 percent. This growth rate in
agriculture has been sustained by the technological progress embodied in the high yielding
varieties of grains and cotton, with supporting public investment in irrigation, agricultural
research and extension, and physical infrastructure (Ali, 2005). Agricultural growth, in turn,
has made significant contribution to the overall economic growth of 5.03 percent per year
during the same time period2.
As in many other developing countries, agriculture in Pakistan faces considerable
challenges in the 21st century. The current population of Pakistan is about 177.1 million,
growing at about 2.05 percent per annum, is estimated to be the third populous country in the
world by the year 2050. Such a huge rise in the size of population is indeed termed as an
2
The average annual growth rates of 3.46 percent, 2.58 percent and 5.03 percent in agricultural GDP, population
and overall GDP, respectively have been calculated from the time series data on these variables, obtained from
website of state bank of Pakistan (www.sbp.org.pk).

1
important constraining factor for achieving sustainable economic growth and food self
sufficiency (GOP, 2011). Per capita income in Pakistan is also showing a rising trend. This
increasing population pressure and higher per capita income is expected to increase the
demand for food in future. The elasticity of demand for food is also high among the poor,
indicating that any shortage of food in future will put the poor at high risk of survival. Thus
with continual population growth, a diminishing supply of per capita arable land, limits to
further expansion of cultivated area, slowing returns to further input intensification and
relatively high income elasticity of food in developing countries like Pakistan, there is
growing need for food supply increases that could only originate from productivity growth
rather than from input growth (Ali, 2004).
The present research tries to highlight the effect of public policies and other economic
measures on TFP growth of agriculture in Pakistan. Analyzing total factor productivity of
Pakistans agriculture, using time series data is important for two reasons. First, in the past
few years, Pakistan has been experiencing very high growth in the region and it is important
to know the latest growth accounting. Secondly, the Pakistan government has implemented
many wide ranging economic reforms since 1999- 2000. As these reforms are implemented
with different vigor in different sectors, agriculture being the main pillar of our national
economy needs much more attention. It is important to know how these macro policy reforms
have contributed in improving the productivity and competitiveness of agriculture in Pakistan.
The paper is organized as follows: Section 2 presents the empirical framework; Section 3
discusses the empirical results, while Section 4 concludes.

2. Empirical Framework

2.1: Data and Variable Specification

Annual time series data in logarithmic form for the period 1971-2006 relate to primary school
enrolment (000 numbers), road length (000 kilometers), credit disbursed to agricultural sector
as percent of agricultural GDP, sum of agricultural exports and imports as percent of
agricultural GDP, inflation rate (in percent), real per capita income (in Rs) and total factor
productivity index. Nominal per capita income was transformed into real per capita income by
GDP deflator (2000-01=100). Pakistan Economic Survey, FAO statistical database, Handbook
of statistics on Pakistan economy is the main sources of data.

2
 A set of macro variables, have been used in literature while studying TFP growth of
the economy. The study at hand used macro variables particularly related to agriculture sector
and can be expected to effect directly or indirectly the TFP growth of this sector. The
description of these factors contributing to TFP growth is given in the following sub-section.

2.1.1: Human Capital Development

Human capital is often regarded as the accumulation of education. The studies have
put forward that educational change influence markedly productivity and economic growth.
Sharpe (1998) has argued that with stable macroeconomic environment, public support for
training, education, and research and development enhances overall productivity of the
economy. Pasha et al., (2002) emphasized the contribution of primary and secondary
education in productivity growth. Khan (2006) used expenditure on education as a proxy for
human capital development to investigate its impact on TFP of the economy. Akinlo (2005)
and Njikam et al., (2006) used secondary school enrolment to capture the effect of education
on TFP. Similarly Nachega and Fontaine (2006) stated that a well-educated and healthy work
force directly or indirectly increases TFP and thus economic growth. They used average
number of years of schooling of the labor force as a proxy for human capital accumulation.
Investment in education promotes more skilled and specialized labor input. Since more skilled
workers are better able to adjust in a dynamic, knowledge-based economy, and this result in
enhanced productivity performance. As the present study confines itself to TFP growth of
agriculture sector, thus, the indicator of education expenditure, used by Khan (2006), is
somewhat a broader measure of human capital, while investigating the impact of education on
TFP of agriculture sector. The present study uses primary school enrolment as a proxy for
human capital development of the labor force in agriculture.

2.1.2: Infrastructural Development

Infrastructure is frequently pointed out in the literature to be a crucial factor effecting
TFP. Extended infrastructure reduces the direct and indirect cost of production. Hazell and
Fan (2002) stressed the importance of infrastructure in enhancing productivity in developing
economies. Public policies in developing countries, which tend to reduce the public
investment in infrastructure in rural areas, are against the phenomenon of productivity growth.

3
It has been proved in many studies that the public investment on infrastructure in rural areas is
playing the role of engine for agricultural productivity growth. Infrastructure can be measured
either in monetary or in physical form, depending on the availability of the data. While
conducting analysis at national level, both measures can be used as in the form of expenditure
on infrastructure or length of the paved roads etc. Fan et al., (1999) explained that rural roads
appear to be the important determinant while analyzing productivity growth of agriculture in
India. Fan and Zhang (2004) also discovers the high importance of rural roads in productivity
of rural areas in China. The present study uses roads length to specify infrastructural
development variable.

2.1.3: Credit Resources

Easy access to credit not only enhances economic growth but also the productivity of
firm and contributes to TFP of the overall economy. Broadly speaking, it is the development
of financial sector that facilitates the credit, necessary for healthy business and reflects
positive relationship with TFP. Credit finds new areas of investment under the efficient
resource allocation. Akinlo (2005) used credit as percent of GDP as an indicator of financial
development. Nacheja and Fountain (2006) used credit to GDP ratio as a proxy for credit
resources. Njikman et al., (2006) used credit disbursement to private sector as a proxy for
financial depth. The present study used credit disbursement to agriculture sector as percent of
agricultural GDP as a proxy for financial sector development in agriculture.

2.1.4: Openness of Agricultural Economy

Openness is generally believed to have a favorable impact on economic growth
through increasing productivity of the economy. It is believed that more open economies can
grow more rapidly through greater access to cheap imported intermediate goods, larger
markets, and advanced technologies that contribute to TFP growth3. In literature, openness of
trade is proxied as export to GDP ratio, export plus import to GDP ratio, export plus import as
percent of GDP (Miller and Upadhay, 2002; Akinlo, 2005; Khan, 2006; Nachega and
Fontaine, 2006; Njikman et al., 2006). The present study used the sum of agricultural exports

3
Lewis (1980), Grossman and Helpman (1994), Miller and Upadhay (2000), Miller and Upadhay (2002), Akinlo
(2005), Khan (2006).

4
and imports as percent of agricultural GDP as a proxy for the openness of agricultural
economy.

2.1.5: Macroeconomic Stability

The theorists and policy makers have conflicting views on several occasions, while
investigating the impact of inflation on growth and productivity. Akinlo (2005) in a study on
macroeconomic factors and total factor productivity growth in Sub-Saharan countries, used
inflation as an indicator for macroeconomic stability. As the developing economies signal the
impact of money illusion that is why, inflation is necessary to be included as a macro
determinant of TFP. The present study used inflation in the model to capture the stability in
the economy, which is considered a necessary player for TFP growth. Inflation may add to
economic growth by generating employment. A positive relationship between inflation and
TFP can be expected. On the other hand, inverse relationship between inflation and TFP can
also be found. It might be that high and unstable prices create economic uncertainties and
discourage investment. Inflation can also encourage capital flight which adversely effect the
investment and hence TFP growth.

2.1.6: Per Capita Income

Per capita income was used to capture the direct or indirect effects of income level of
the masses on agricultural productivity growth. In Pakistan, non-farm income makes the
larger proportion of the per capita income. The share of non-farm income in the per capita
income in rural areas is about 59 percent, while the share of crop and livestock income is
about 41 percent (Adams, 1993). Kamal (2006) also indicated that non-farm income
represented the largest source of rural household income and had favorable impact on income
distribution. Thus per capita income does not purely reflect the income from agriculture
production in Pakistan and there are many other sources contributing in per capita income. Per
capita income of the country may contribute towards the increase in agricultural productivity
through: 1) Increasing demand for food and other agricultural products. Income elasticity of
demand for food is high in developing countries like Pakistan. That increase in demand may
act as an incentive for farmers through change in price and farmers start making efforts for
efficient utilization of the resources to increase their production; 2) Increase in per capita

5
income improves the health and education level of the masses that, in turn, assume to have
positive impacts on productivity through greater access to sources of information and better
decision making; and 3) Increase in per capita income, especially in the rural areas may assure
greater access to new technology at farm level which will add to agricultural productivity.

2.1.7: Total Factor Productivity Index

The TFP index of agriculture in Pakistan has been estimated by Ali et al., (2009) and
that estimated TFP index of agriculture was used as a dependent variable in the present study.

2.2: Model Specification

To investigate the impact of different macro variables on TFP growth, the model is
specified as:

LTFP f ( LPSE , LRL, LCRD, LINF , LSXM , LPCI ) (1)

Where;
LTFP = log of total factor productivity index;
LPSE = log of primary schools enrolment (proxy for human capital development);
LRL = log of road length (proxy for infrastructural development);
LCRD = log of credit disbursed to agriculture sector as percent of agricultural GDP
(proxy for credit resources in agriculture);
LINF = log of inflation rate (proxy for macroeconomic stability);
LSXM = log of sum of agricultural exports and imports as percent of agricultural GDP
(proxy for openness of agricultural economy);
LPCI = log of real per capita income

2.3: Estimation Procedure

2.3.1: Testing For Unit Root

The present study begins by testing for the presence of unit roots in the individual time

series, using the augmented Dickey-Fuller (ADF) test (Dickey and Fuller, 1981), both with

6
and without a deterministic trend. The number of lags in the ADF-equation is chosen to

ensure that serial correlation is absent using the Breusch-Godfrey statistic (Greene, 2000,

p.541). The ADF equation is required to estimate the following by OLS.

k
Yt 3 3 t (3 1)Yt 1 iYt i ut (2)
i 1

Where Yt is the series under investigation, t is a time trend4 and ut are white noise

residuals. It is not known that how many lagged values of the dependent variable to be

included on the right-hand side of (2). There are several approaches but the present study used

the Lagrange Multiplier (LM) test (Holden and Perman, 1994, p.62).

2.3.2: Testing For Cointegration

If two series are integrated of the same order, Johansen's (1988) procedure can then be
used to test for the long run relationship between them. The procedure is based on maximum
likelihood estimation of the vector error correction model (VECM):
z t 1z t 1 2 z t 2 p 1z t p 1 z t p x t u t (3)

where zt is a vector of I(1) endogenous variables, zt=zt-zt-1, xt is vector of I(0) exogenous

variables, and and iare (nn) matrices of parameters with i=-(I-A1-A2--Ai), (i=1,,k-

1), and =I-1-2--k. This specification provides information about the short-run and long-

and
run adjustments to the changes in zt through the estimates of respectively. The term
i

z t k provides information about the long-run equilibrium relationship between the

variables in zt. Information about the number of cointegrating relationships among the

variables in zt is given by the rank of the -matrix: if is of reduced rank, the model is subject

to a unit root; and if 0<r<n, where r is the rank of , can be decomposed into two (nr)

4
The rationale for having a trend variable in the model is that as most of the series are trended overtime. So it is
important to test the series for unit root having a stochastic trend against the alternative of trend stationary.

7
matrices and , such that ' where 'zt is stationary. Here, is the error correction term

and measures the speed of adjustment in zt and contains r distinct cointegrating vectors

that are the cointegrating relationships between the non-stationary variables. Johansen (1988)

uses the reduced rank regression procedure to estimate the - and -matrices and the trace test

statistic is used to test the null hypothesis of at most r cointegrating vectors against the

alternative that it is greater than r.

2.3.3: Error Correction Mechanism

When the variables are cointegrated, there is general and systematic tendency for the
series to return to their equilibrium value. It means that short-run discrepancies may be
constantly occurring but cannot grow indefinitely. This shows that the adjustment dynamics is
intrinsically embodied in the cointegration theory. The theorem of Granger representation
states that if a set of variables is cointegrated (I, I), it implies that the residual of the
cointegrating regression is of order I(0), thus there exists an ECM describing that relationship.
This theorem explains that cointegration and ECM can be used as a unified theoretical and
empirical framework for the analysis of both short-run and long-run behavior. The ECM
specification is based on the idea that adjustments are made to get closer to the long-run
equilibrium relationship.

Let assume that X t and Yt variables are cointegrated and the relationship between these
two can be expressed as ECM. Assuming that the X t is the cause of Yt and both variables
are considered in logarithmic form. The ECM can be written as:

DLX t = 0 1 DLYt 2 ECTt 1 t (4)

Where D denotes the first difference operator and t is the random error term. The
ECTt 1 is the one period error correction term from the cointegration regression. The

equation (4) states that DX t depends on DYt and also on the error correction term (ECT).
2.3.4: Granger-Causality Analysis

8
 After establishing cointegration, Engle and Granger (1987) error correction
specification was used for testing of Granger Causality. If the series Yt and X it are I(1) and
are cointegrated, then the ECM model is represented in the following form:

n n
Yt 0 i Yt 1 j X it 1 ECTt 1 t (5)
i 1 i 1

n n
X it 0 i X it 1 j Yt 1 ECTt1 t (6)
i 1 i 1
where is difference operator, t and t are the white noise error terms, ECTt 1 is the
error correction term derived from the long-run cointegrating relationship and n is the optimal
lag length orders of the variables. The null hypothesis was constructed as H o: X it will
granger-cause Yt , if t 0. Similarly, Yt will granger-cause X it , if t 0. For its
implementation, F-statistics are calculated under the null hypothesis that coefficients of t
and t are equal to zero in the above equations. When the computed F-value is greater than
the F-tabulated value, the null hypothesis was rejected, explaining the granger cause of one
variable on the other.
3. Empirical Results
3.1: Unit Root Results

ADF- test was performed for testing the unit roots in the variables. Null hypothesis of
the unit root were tested against the alternative hypothesis of stationarity by the ADF
regressions, including an intercept but not a trend and with an intercept and a linear trend.
Maximized log-likelihood (LL), Akaike Information Criterion (AIC), Schwarz Bayesian
Criterion (SBC) and Hannan-Quinn Criterion (HQC) were used to determine the optimal lag
length for the augmented terms. The computed absolute value of the test statistic was checked
against the maximum values of these criteria with the 95 percent absolute critical value for the
ADF- statistic. When the computed absolute test statistic value was greater than the absolute
critical value, the null hypothesis of unit root was rejected which employed the stationarity in
the time series. On the other hand, when the absolute test statistic value was less than the

9
absolute critical value, the null hypothesis of the unit root was accepted, employing that series
was non-stationary. The results are presented in Table 3.1.

Table 3.1: ADF- Unit Root Results of the Selected Variables in Level Form
Variables Non-Trended Trended 3 Conclusion
LTFP -0.41 -2.20 2.51 I(1)
LPSE -0.37 -2.60 3.35 I(1)
LRL -3.89 -3.32 2.67 I(1)
LCRD -1.64 -1.61 4.53 I(1)
LINF -3.97 -2.59 9.12 I(0)
LSXM -2.82 -3.91 8.44 I(0)
LPCI -0.37 -4.25 9.31 I(0)
C.V -2.96 -3.57 6.73
Source: Authors own calculations
Note: C.V is critical values for 5 percent significance level
Table 3.1 shows that the absolute computed values of the variables [(Total Factor Productivity
index (LTFP), Primary Schools Enrolment (LPSE) and Credit Disbursed to Agriculture Sector
as percent of Agricultural GDP (LCRD)], in the level form were less than absolute critical
values (5 percent significance level), both for trended as well as for the non-trended models.
The 3 -test also supported the results of the first two models, as the computed value was less
than the critical value for the said variables. Thus the null hypothesis of unit root was
accepted and concluded that the above mentioned data series were non-stationary in the level
form. The absolute computed value for the variable of Road Length (LRL) was less than the
critical value for the ADF-statistic in the trended model but greater than the critical value in
the non-trended model. The 3 -test was performed and the results indicated that computed
value was less than the critical value. Thus out of three models, two suggested the non-
stationarity in the data series of road length and thus the null hypothesis of unit root for the
variable of road length was accepted. The variables of Openness (LSXM), Inflation Rate
(LINF) and real Per Capita Income (LPCI) were also analyzed for presence of unit root in
level form. The results indicated that absolute value of test statistic for the variable of
Inflation Rate (LINF) appeared to be stationary in non-trended model. However, the other two
tests (trended and 3 ) suggested that data series was stationary at its level form as the
computed values were greater than the absolute critical values. Thus out of three models, two
suggested the stationarity in the series of inflation rate, so the null hypothesis of unit root for
this variable was rejected. Thus it was concluded that the variable of inflation rate (LINF) was

10
stationary at the level form. The absolute computed value was less than the critical value in
the non-trended model for the variables of Openness (LSXM) and real Per Capita Income
(LPCI). The computed absolute value of the test statistic for the trended model for these
variables in the level form was greater than value for ADF- statistic. These results were also
confirmed by 3 -test, which showed the stationarity in the variables of openness and per
capita income. Thus the null hypothesis of the presence of unit root was rejected for these
variables and it employed that these variables were stationary at the level form. Thus these
variables (LINF, LSXM, LPCI) were said to be integrated of order zero denoted by I(0). The
variables which were non stationary at the level form, analyzed again in the first difference
form to check stationarity. All the series become stationary at their first difference form.

3.2: Cointegration Results

After testing for unit root, the next step is to test for cointegration. As unit root results
indicate that LINF, LSXM, LPCI series are I (0) i.e., stationary. This I (0) series cannot
interpret the long-run relationship between I (1) variables, but can explain the short-run
behaviour and therefore is allowed to enter the unrestricted VAR as exogenous variable.
Johansens procedure was applied to test the cointegration between the respective variables.
The first step in Johansens procedure is the selection of order of Vector Auto Regressive
(VAR). We use the LR-statistic, adjusted for small samples (Sims, 1980), to test the null
hypothesis that the order of the VAR is k against the alternative that it is four where k=0, 1,,
4 and for all cases, k=15. The second step in Johansen procedure is to test for the presence and
the number of the cointegrating vectors among the series in the model. The rank of the
cointegration i.e. the number of the cointegrating vectors was selected by using the trace
values test statistics. The Johansen cointegrating results of the existence and number of
cointegrating vectors among the series in the TFP growth model are presented in Table 3.2.
The results show that first statistic value was greater than the 95 percent critical value. Thus
on the basis of the results, the study rejected the null hypothesis of no cointegration and
accepted the alternative hypothesis of the existence of cointegration. According to Harris
(1995) the number of cointegrating vector is one when the null hypothesis is rejected for the
5
We also tried the Schwarz Bayesian Criterion (SBC) and Akaike information Criterion (AIC). Both SBC and
AIC selects lag length one and four respectively. To avoid over-parameterisation, we choose one as the lag length
(Pesaran and Pesaran, 1987).

11
first time. It can safely be said that there was one cointegrating vector among the series
concerned.

Table 3.2: Co integration Results --- Trace Statistics

List of variables included in the cointegrating vector:
LTFP LPSE LRL LCRD
INTERCEPT
List of I(0) Variables Included in the VAR:
LINF LSXM LPCI
H0: (No Cointegration) H1: (Cointegration) Test Statistic 95 % C. Values
r=0 r=1 53.895 53.4800

r <= 1 r=2 28.577 34.8700

r <= 2 r=3 10.759 20.1800
r <= 3 r=4 2.127 9.1600
Note: r is the number of cointegrating vectors

In the Johansen model parameters in the cointegrating vector can be interpreted as estimates
of long run cointegrating relationship between variables (Hallam and Zanoli, 1993). The
estimated parameter values of equation (7), when normalized on the series of TFP index were
the long-run elasticities.

LTFP= 0.64LPSE + 0.07LRL + 0.03LCRD1 (7)

3.3: Error Correction Model Estimates Results

The error correction model results are reported in Table 3.3 and show that the signs of
the estimated coefficients of all the macro variables are according to a priori expectations. The
human capital development has a positive sign describing a positive relationship between TFP
of agriculture and human capital. The results indicated that a one percent increase in the
primary schools enrolment (improvement in the educational capability of the labor force)
increased TFP of agriculture by 0.64 percent in the long-run and of only 0.03 percent in the
short-run. The results of this variable explained that human capital improvement accounted

12
for a significant contribution and highlighted the importance of raising the human capital
endowment of the agricultural labor force to achieve increases in TFP of agriculture.
The long-run elasticity of infrastructure development proxied by the road length was
0.07 with a positive sign. It implied that a one percent increase in the road length increased
the productivity of agriculture by 0.07 percent in the long-run. Short-run elasticity of this
variable was positive but insignificant with a magnitude of 0.01. The findings of this study are
in accordance with the previous studies. Evenson and Bloom (1991) observed a positive and
significant impact of road length on productivity of Pakistans agriculture in the long-run.
Zhang and Fan (2001) found that infrastructure development affects the agricultural
productivity in India in the long-run but not in the short-run. Fan et al., (2002) explained that
rural roads to be the important determinant for agricultural productivity growth in China.
The results also explained that a one percent increase in credit resources increased
TFP of agriculture by 0.03 percent in the long-run and 0.05 percent in the short-run. The
coefficient of credit resources was non significant both in the long-run as well as in the short-
run. The insignificance of credit variable in the present study might be due to inefficient and
highly inequitable distribution of agricultural credit. Most of the credit distributed was
directed towards the large land holdings families exerting political influence. Another sound
reason of the insignificance of this variable was the miss-utilization of credit and the
administrative hurdles created during the provision of loans to farmers.
The openness of agricultural economy was positively associated with the productivity
change of agriculture. The coefficient of this variable was significant with a magnitude of
0.19 which implied that one percent increase in the sum of agricultural exports and imports
increased TFP of agriculture by 0.19 percent. The sign of the coefficient was according to a
priori expectation because it is generally believed that openness have a favorable impact on
growth through increasing productivity. The fact is that more open economies can grow more
rapidly through greater access to imported intermediate goods and advanced technologies that
contribute to enhance productivity. The results for macroeconomic stability proxied by the
rate of inflation indicated a significant negative effect on TFP of agriculture. The elasticity of
inflation rate was -0.03, implied that one percent increase in the inflation decreased
productivity of agriculture by 0.03 percent. Though the magnitude of the coefficient was
small yet, it gave the direction between inflation and the TFP of agriculture, necessary for

13
policy implications. The inverse relationship between inflation and TFP might be that high
and unstable prices which tend towards inflation create a lot of economic uncertainties that
discourage investment in agricultural related projects. This negative association might be also
due to the fact that inflation encourage capital flight which adversely effected the investment
and hence TFP growth. The insignificance of the coefficient of real per capita income might
be due to the fact that increases in per capita income not equally distributed among the
individuals in the country. Thus due to high unequal distribution of income, per capita income
could not significantly influence the TFP growth of agriculture sector.
The coefficient of the error correction term has a negative sign which is according to
the theory and it tells about adjustment measures towards long-run equilibrium. The error
correction term has the coefficient of -0.23 which was highly significant, implies that the
deviation of productivity growth from the long-run equilibrium level was corrected by about
23 percent in a year.
All other diagnostic tests provided satisfactory results. The LM-test indicated that
there is no problem of serial correlation among the residuals. The RESET-test also verified the
correct functional form of the model. The Jarque-Bera test gave conclusion about the normal
distribution of the residuals. The R2 value of 0.52 indicated that about 52 percent variation in
the total factor productivity in agriculture was explained by the factors included in the model.
Similarly Durbon-Watson statistics also verified the fact of no serial correlation among the
residuals.

Table 3.3: Error Correction Model Estimates

Regressor Short-run Long-run
Constant 2.52 (1.95)
DLPSE 0.03 (0.20)NS 0.64 (4.44)*
DLRL 0.01 (0.04)NS 0.07 (1.91)**
DLCRD 0.05 (0.59)NS 0.03 (0.38)NS
LINF -0.03 (-1.87)**
LSXM 0.19 (1.82)**
LPCI 0.06 (0.30)NS
ECM 1 (-1) -0.23 (-2.39)**
R2 0.52 LM- 2 0.062

14
 D.W 2.14 RESET-2 0.937
Jarque-Bera 0.612
Normality- 2

Note: a) t-ratios are given in parenthesis

b) * and ** indicates significances level at 5 and 10 percent level
respectively; and
c) NS denotes the non-significances of the coefficients.

3.4: Granger Causality Analysis

The results of causality analysis between Total Factor Productivity (TFP) and macro
variables included in the model are presented in the Table 3.4, using the equations 5 and 6
described in the previous section. The regression was run separately for each of the
explanatory variable which are of I(1) with the dependent variable (LTFP) index including
error correction term and obtained the Granger-Causality results. The first row shows the F-
statistics value of 2.29 which was significant at about 10 percent level. While the second part
of the first row was also significant at 4 percent. Thus, strong causality could be concluded
between productivity and development in human capital. The direction of causality is both
ways i.e. Bi-Directional. The Granger-Causality results between human capital development
and total factor productivity index supported the results discussed in the previous section.
Human capital development had a strong impact on agricultural productivity and it emerged
as a major determinant of Total Factor Productivity (TFP) growth of agriculture. The results
also indicated that increase in agricultural productivity promotes primary education. This
showed that development in human capital granger cause productivity of agriculture and vise
versa.
F-statistics for causality from credit to Total Factor Productivity (TFP) is 1.94 which
was non-significant and the reverse of this also showed the same result. Thus the results
explained no causality between credit disbursement to agriculture and productivity growth
and similarly between productivity and credit disbursement. The causal relationship between
infrastructure development and Total Factor Productivity (TFP), showed the significance of F-
statistics value at 5 percent level in case of causality from infrastructure development to
productivity growth. This explained the sound influence of infrastructure development
towards change in the productivity. Healthy infrastructure also attracts new foreign and local

15
investments which help to increase productivity. The causality from Total Factor Productivity
(TFP) towards infrastructure was also highly significant which showed that a change in
agricultural productivity caused a change in infrastructural development. More agricultural
production increases GDP of the country and it enhances the resources at national level. Thus
the governments invest more in the infrastructural development from the increased capital
accumulation caused by the productivity of agriculture. Thus, the direction of the relationship
in this case was of two ways.

Table 3.4: Granger-Causality Results

Causality F-statistics P-value Direction

LPSE LTFP 2.29 0.10 Bi-Directional

LTFP LPSE 3.14 0.04

LCRD1 LTFP 1.94 0.14 No-Direction

LTFP LCRD 1.33 0.28

LRL LTFP 2.86 0.05 Bi-Directional

LTFP LRL 3.47 0.02
4. Conclusions and Policy Recommendations

The present study was designed to investigate the impact of different macro variables
(human capital development, infrastructure development, credit resources, and openness of
agricultural economy, macroeconomic stability and real per capita income) on productivity
growth of agriculture in Pakistan. In order to analyze this impact, TFP index of Pakistans
agriculture, estimated by Ali et.al (2009) is used as a dependent variable in the present study
and determined the productivity response to selected macro variables described above, using
cointegration and error correction model approach.
The results of cointegration and error correction model indicated that the magnitudes
of the elasticity estimates are smaller in size in the short-run as compared to long-run for all
the variables except credit disbursement. The study concluded that improvement in human
capital and infrastructure development has a significant positive effect on TFP of agriculture
in the long-run. The credit resources showed a positive but insignificant effect both in the

16
long-run as well as in the short-run. The results also indicated that openness of agricultural
economy and real per capita income was positively associated with TFP of agriculture. The
openness of agriculture economy was found to be significant while the variable of real per
capita income was non-significant. The analysis also concluded that inflation rate has a
significant negative effect on productivity of agriculture in Pakistan. Over all the results
showed that policies which promote human capital, increase credit
resources, improve infrastructural development, facilitate openness of
agricultural economy and ensure macroeconomic stability; would lead to
higher productivity growth in Pakistans agriculture.
Based on the findings and research implications, following policy measures/recommendations
are suggested to improve total factor productivity of agriculture in Pakistan.
1) The importance of education is beyond any doubt in uplifting the productive capacity
of the farming community. Education of the labor force is important in increasing the
efficiency of resource use and strengthening research for technological progress. The
coefficient of primary schools enrolment (proxy for human capital development) was
positive and highly significant in the long-run which showed that education improves
the human capital of the country and leads to enhance the productivity of the labor
force. The magnitude of the coefficient in the long-run was largest (0.64) amongst all
the variables. Thus the results strongly suggest that the primary education should
remain a priority agenda for the government and in this regard, specific steps should
be taken to promote the basic education. Investment in this sector is expected to
enhance productivity of agriculture significantly in the long-run. Budget allocation
should be increased to enhance primary schools enrolment.

2) Infrastructure is an important determinant in promoting the traditional agricultural

economies to transitional agricultural economies by increasing marketable surplus and
reducing post harvest losses of agricultural commodities. The results of the study
showed a positive and significant relationship between infrastructure and total factor
productivity growth of agriculture in the long-run, advocating and justifying further
investments in this sector on sustainable basis. Road network should be expanded to
ensure the timely availability of inputs and easy access to the markets for agricultural

17
 products which will help in increasing the resource use efficiency and thus
productivity. This will in turn ensure better returns to the farming community making
them more productive. Better infrastructure attracts more domestic and foreign
investments that will further increase productivity of agriculture. In this perspective
government should focus on improving the access through roads particularly in remote
and rural areas. Public private joint ventures may be a fruitful option in this regard.
This will ensure sustainability of infrastructural developments in Pakistan.

3) Credit resources in agricultural economy, measured by the credit disbursement to

agriculture sector as percent of agricultural GDP showed positive association with
small magnitude of coefficients which were non-significant both in the long-run and in
the short-run. Though the sign of the coefficient was according to a priori expectation,
but its non-significance may be due to the discrimination in its distribution and
improper utilization. Thus in order to have a significant effect of credit on
productivity, it is suggested that the small farmers should be provided an easy access
to credit. Administrative hurdles should be eliminated and strict vigilance in the use of
the credit be ensured. It is also recommended that credit should be given to farmers in
the shape of kind rather than cash to reduce the chances of its misutilization. Credit for
mechanization in agriculture should be increased in order to capture its long-run
impact on productivity. The field officers responsible for monitoring the activities of
farmers should also be trained and motivated to ensure proper utilization of resources.
This will help in fetching fair returns from the utilization of credit thus leading to
improvement in total factor productivity.

4) It is an established fact that openness stimulates growth of the economy including

agriculture sector. This factor becomes more important in the present scenario of trade
liberalization. The coefficient of sum of agricultural exports and imports as percent of
agricultural GDP (proxy for openness of agricultural economy) was 0.19 which was
significant and largest next to primary schools enrolment (human capital
development). Thus, the expansion in the volume of agricultural trade should be the
priority agenda of trade policy in Pakistan. It is recommended that agricultural trade

18
 volume should be expanded through increasing the exports. This needs heavy
investments in agriculture sector to increase marketable surplus. Government should
also adhere to the notion of trade liberalization by promoting further trade of
agricultural commodities. In this regard, to comply with above mentioned objectives, a
more open and liberal trade policy should be the focus of the government. Awareness
should be created among the farming community in adopting and imitating technology
that trickles through trade. It is also strongly recommended that the optimal share from
agricultural exports should be transferred to farmers in order to create incentive to
enhance productivity. Trade barriers should be removed and new markets for the
exports of agricultural commodities should be searched out. At the same time private
sector should be motivated to comply with emerging requirements of trade
liberalization. In addition more diplomatic efforts are also needed to develop a good
image for Pakistani products in the international market.

5) High rate of inflation is an important factor which adversely affects the purchasing
power of the farming community and leads to misallocation and underutilization of the
resources. This is also evident from the results through negative and significant impact
of inflation rate on total factor productivity growth of agriculture. On the basis of the
results, it is recommended that government should adopt contractionary monetary
policy on the one side and productive utilization of resources on the other side to
control inflation. This policy option will stabilize agricultural prices of inputs and
technological intervention through continuous monetary and regulatory measures. This
policy initiative will strengthen the economy and confidence of stakeholders in
government policies and through multiplier effect, Pakistan may get numerous
benefits through increasing productivity of agriculture.

19
References
Adams, Jr. R. (1993). Non-Farm Income and Inequality in Rural Pakistan. Pakistan
Development Review, Vol. 32, No. 4, pp. 1187-1198.

Akinlo, A. E. (2005). Impact of Macroeconomic Factors on Total Factor Productivity in Sub-

Saharan African Countries. Research Paper # 2005/39, World Institute for Development
Economics Research, UNU.

Ali, A., K. Mushtaq, M. Ashfaq, and M. A. Raza (2009). An Analysis of Productivity Growth
of Agriculture in Pakistan 1971-2006. J. of Agric. Res, Vol. 47(4), 439-450.

Ali, S. (2004). Total Factor Productivity Growth in Pakistans Agriculture: 1960-96. The
Pakistan Development Review, Vol. 43, No. 4, pp. 493-513.

20
Ali, S. (2005). Total Factor Productivity Growth and Agricultural Research and Extension:
An analysis of Pakistan's Agriculture, 1960-1996. The Pakistan Development Review, Vol.
44 No. 4, pp. 729-746.

Dickey, D. A. and W. A. Fuller (1981). Likelihood Ratio Statistics for Autoregressive Time
Series with Unit Roots. Journal of Econometrica, Vol. 49, pp. 1057-72.

Engle, R. F. and C. W. J. Granger (1987). Cointegration and Error Correction:

Representation, Estimation and Testing. Journal of Econometrica Vol. 55, pp. 251-276.

Evenson, R. E., and E. A. Bloom (1991). Research and Productivity in Pakistans

Agriculture. In Haider, A. S., Z. Hussain, R. McConnon, and S. J. Malik (Eds). Agricultural
Strategies in the 1990s: Issues and Policies. Pakistan Association of Agricultural Social
Scientists, Islamabad, Pakistan.

Fan S., P. Hazel, and T. Sukhadeo (1999). Linkages between Governments Spending,
Growth and Poverty in Rural India. Research Report No 110. International Food Policy
Research Institute: Washington D.C., USA.

Fan, S., and X. Zhang (2004). Infrastructure and Regional Economic Development in
Rural China. China Economic Review, Vol.15, No.2, pp. 203-214.

Government of Pakistan (2011). Economic Survey of Pakistan, 2010-11. Ministry of

Finance, Economic Advisors Wing, Islamabad.

Greene, W. H. 2000. Econometric Analysis. New Jersey: Prentice-Hall, Inc.

Grossman, G., and E. Helpman (1994). Endogenous Innovation in the Theory of Growth.
Journal of Economic Perspectives, Vol. 8, pp. 23-44.

Hallam, D., and R. Zanoli. 1993. Error Correction Models and Agricultural Supply Response.
European Review of Agricultural Economics 20: 151-66.
Harris, R. (1995). Using Cointegration Analysis in Econometric Modeling. Prentice Hall,
Harvester Wheatsheaf.

Hazell, P. B. R., and S. Fan (2002). The Importance of Public Investments for the
Functioning of Food Markets in Developing Countries. In Lutz, Clemens, ed. Food Markets
and Food Security in West Africa, Centre for Development Studies (CDS) Research Report
No. 15.

Holden, D. and Perman, R. 1994. Unit Roots and Cointegration for the Economist, in Rao, B.
B., Cointegration for the Applied Economists, St. Martin's Press, New York.
Johanson, S. (1988). Statistical Analysis of Cointegrating Vectors. Journal of Economic
Dynamics: Vol. 2, No. 3, pp. 231-254.

21
Kamal, A. R. (2006). Income Inequalities in Pakistan and a Strategy to
Reduce Income Inequality. Policy Paper PRSP-II Study, Ministry of Finance,
Government of Pakistan.

Khan, S. (2006). Macro Determinants of Total factor Productivity in Pakistan. Working

Paper Series #10, State Bank of Pakistan, Karachi.

Lewis, A. W. (1980). The Slowing Down of the Engine of Economic Growth. American
Economic Review, Vol. 70, pp. 555-564.

Miller, S. M., and M. P. Upadhay (2000). The Effect of Openness, Trade Orientation, and
Human Capital on Total Factor Productivity. Journal of Development Economics, Vol. 63,
pp. 399-423.

Miller, S. M., and M. .P. Upadhyay (2002). Total Factor Productivity, Human Capital and
Outward Orientation: Differences by Stage of Development and Geographic Regions.
University of Connecticut, Department of Economics Working Paper Series, Working Paper
No. 2002-33.

Nachega, J. and T. Fontaine (2006). Economic Growth and Total factor Productivity in
Niger. International Monetary Fund (IMF), working paper, No. WP/06/208.

Njikam.O, J. N. Binam, and S. Tachi (2006). Understanding Total Factor Productivity

Growth in Sub Saharan Africa Countries. WP No.2006/3, Secretariat for Institutional
Support for Economic Research in Africa.

Pasha, H. A., A. G. Pasha, and K. Hyder (2002). The Slow Down of the Growth of Total
Factor Productivity in Pakistan. Social Policy and Development Centre (SPDC), Karachi,
Pakistan, Research Report No. 44.

Pesaran, H.M. and Peraran, B. (1987). Working with Microfit 4.0: Interactive
Econometric Analysis, Oxford University Press.
Sharpe, A. (1998). Productivity: Key to Success. The Report, Ottawa: Centre for the Study
of Living Standards.

Sims, C. A. (1980). Macroeconomics and Reality. Econometrica.48: 1-49.

Zhang, X., and S. Fan (2001). How Productive is Infrastructure? New Approach and
Evidence From Rural India. EPTD Discussion Paper No. 84. Washington, D. C: International
Food Policy Research Institute (IFPRI).

22