Copyright UNU-WIDER 2008

* Department of Economics and Finance, Baruch College, CUNY.

This study has been prepared within the UNU-WIDER project on the Southern Engines of Global
Growth.
UNU-WIDER gratefully acknowledges the financial contributions to the research programme by the
governments of Denmark (Royal Ministry of Foreign Affairs), Finland (Ministry for Foreign Affairs),
Norway (Royal Ministry of Foreign Affairs), Sweden (Swedish International Development Cooperation
AgencySida) and the United Kingdom (Department for International Development).
ISSN 1810-2611 ISBN 978-92-9230-121-7
Research Paper No. 2008/67

Indian Economic Growth

Lessons for the Emerging Economies

Suparna Chakraborty *

July 2008
Abstract
Can we use neoclassical growth model to single out the important transmission channels
through which external factors or primitives affected the Indian economy and caused
the remarkable growth of the period 19822002? In this paper, we answer the question
by applying the new technique of business cycle accounting to the Indian economy. Our
results show us that the primary conduit of policies that brought about significant
growth in India was productivity that registered an unprecedented increase particularly
in the 1990s. Our results further indicate that changes in labour market frictions and
investment market frictions did not play a significant role, though increased government
consumption aided growth by propping up demand. In addition, we examine the
effective tax rates in India and find that while investment taxes barely fluctuated,
income tax rates were increasing throughout. We suspect other positive developments in
the Indian economy overwhelmed the negative effect of increasing labour income taxes
on growth. Our result suggests that any emerging country that aims to replicate the
Indian experience would do well to formulate policies that target productivity, a lesson
that seems consistent with the Japanese experience since the Second World War.
Keywords: business cycle accounting, India, growth, wedges, neoclassical growth, taxes
JEL classification: E13, E32

The World Institute for Development Economics Research (WIDER) was
established by the United Nations University (UNU) as its first research and
training centre and started work in Helsinki, Finland in 1985. The Institute
undertakes applied research and policy analysis on structural changes
affecting the developing and transitional economies, provides a forum for the
advocacy of policies leading to robust, equitable and environmentally
sustainable growth, and promotes capacity strengthening and training in the
field of economic and social policy making. Work is carried out by staff
researchers and visiting scholars in Helsinki and through networks of
collaborating scholars and institutions around the world.
www.wider.unu.edu [email protected]

UNU World Institute for Development Economics Research (UNU-WIDER)
Katajanokanlaituri 6 B, 00160 Helsinki, Finland

Typescript prepared by the author.

The views expressed in this publication are those of the author(s). Publication does not imply
endorsement by the Institute or the United Nations University, nor by the programme/project sponsors, of
any of the views expressed.
Acknowledgements
I thank Dr. Sugata Marjit and CSSSCal for hosting me during the preliminary analysis
of this paper. All remaining errors are mine.


1 Introduction
Since the last two decades, Indias name has gured prominently amongst the list of emerging
economies. After decades of slow growth, India entered into a new phase of economic reconstruction
during the mid eighties and the nineties. In comparison to a measly 1% growth rate in the years
following independence from the British Raj in 1947, Indian GDP per capita growth rate reached
1.5% in the eighties, still not quite up to the benchmark but not dismal either. The picture
drastically changed in the nineties, when the growth rate of GDP per capita reached an average of
5%, much beyond the benchmark and a small miracle on its own.
This economic transition has been attributed to many factors among which the development
of the software sector and a pro-liberalization approach of the government as opposed to a more
centrally planned socialist approach of yesteryears has been hailed as the primary drivers of growth.
There are quite a few papers which agree with the contribution of Indian IT sector as the catalyst of
economic growth. On one hand we have Nirvikar Singh (2004) who argues in favor of the important
role played by the Indian IT sector in promoting growth. This view, perhaps not surprisingly, nds
great support among the IT pioneers of India. NR Narayan Murthy, Chairman of Infosys, one
of the fastest growing IT companies that originated in India hails the changing climate in India
by arguing that "...the economic reforms of 1991 changed the Indian business context from one
of state-centered, control orientation, to a free, open market orientation - especially for hi-tech
companies. It allowed Indian companies to start competing eectively on a global scale"
1
. On the
other end of the spectrum, we have Dani Rodrick and Arvind Subramanian (2004) who investigate
"... a number of hypotheses about the causes of this growth favorable external environment, scal
stimulus, trade liberalization, internal liberalization, the green revolution, public investment and
nd them wanting." They argue that ".....growth was triggered by an attitudinal shift on the part
of the national government towards a pro-business (as opposed to pro-liberalization) approach". In
a previous paper
2
we examined the role of technology or productivity growth in bringing about the
economic miracle in India using a neoclassical framework following the growth accounting procedure
of Edward Prescott and Finn Kydland (1982), where technology is treated as an external shock.
However, in addition to merely identifying the macroeconomic fundamentals
3
that have played a
role in Indian economic development, another issue of interest is "how" they aect the economy.
Thus a study of numerically accounting for Indian economic growth is incomplete unless in addition
to identifying the primary forces that were particularly successful in generating economic growth
in India, we also identify the "transmission channels" through which these factors worked. In this
paper, we concentrate on this second issue and this leads us to examine the growth of India during
the eighties and the nineties through the lens of business cycle accounting. We further extend our
analysis to see if changes in taxes played any role in generating increased growth by comparing the
time series of tax rates in India with the results of our business cycle accounting.
The Business Cycle Accounting (henceforth BCA) procedure, a relatively new procedure
4
is
based on the fundamental observation that there are primarily three channels or "wedges" through
1
Quotation from speech delivered at the Indian Economy conference at Cornell in 2002.
2
Chakraborty (2006)
3
In technical jargon, the external forces or macro fundamentals that aect an economy causing a change in output
is also referred to as a "primitive".
4
Two seminal papers in this area is by Casey Mulligan (2002) and V.V. Chari, Patrick Kehoe and Ellen McGrattan
(2002, 2006).
1
which any external friction or policy can aect an economy: through an impact on productivity,
through an impact on labor market, and through an impact on investment market. For example,
if we believe that economic liberalization and free market policies were the primary forces behind
Indian economic growth, two obvious channels through which these policies aected the economy
seems to be the productivity channel and the investment market channel where the frictions that
made it costly for Indian rms to gain access to funds were considerably lessened if not completely
eradicated. Economists also consider government spending to be an important channel through
which government can directly aect the economy.
There are a couple of crucial points that one needs to keep in mind while conducting a business
cycle accounting analysis for any economy. In an ideal world with no frictions, the economy would
be able to achieve the rst best outcome, or grow at a balanced growth rate as assumed in real
business cycle literature. However, if an economy is subject to market frictions, these frictions
would prevent the economy from achieving the rst best outcome and the economy would move
away from a balanced growth path. In a real business cycle model, on which the BCA procedure is
based, these frictions or "wedges" would show up as distortions in the rst order conditions. What
BCA tells us is that there are primarily three ways (four if you consider government expenditure
channel) through which external frictions aect an economy. If we knew the numerical value of the
frictions exactly, then if we feed in all the frictions jointly in a benchmark real business cycle model,
we should be able to replicate the data exactly. The question we are interested in is: which of these
frictions play a major role in the economy? Here the trick is to allow for the frictions one by one
to assess their importance by evaluating how well can a friction on its own replicate the data. This
exercise is called "decomposition".
From a methodological perspective, to apply the BCA procedure to India, we take a neoclassical
growth model and extend it to include time varying eciency or productivity wedge that is the Solow
residual, labor wedges that are modelled as labor income taxes, investment wedges that resemble
taxes on investment expenditure and government wedges that is actually government spending.
The solution of the model involves two parts: since the wedges represent external market frictions,
there is no data available on them and we need to use the results of our model to calculate the
wedges (this step is somewhat like back-calculating the value of frictions). The primary idea here is
that the wedges summarize all possible frictions that aect the economy. Hence in theory the data
on output, investment or labor that we observe is a function of these wedges. Since we know the
data from National income accounts, we can use them and the policy functions from our model to
calculate the wedges. Next, we feed in the time series of our model generated wedges one by one
and in various frictions and check how well they match up with the observed output.
Our decomposition results show that primitives aected the Indian economy primarily by causing
changes in productivity. Labor market frictions or investment market channel was not particularly
important. The growth in productivity was also supported by increasing government expenditure
that propped up demand. Comparing the Indian experience with that of Japan in the reconstruc-
tion period after the 2nd World War, the common channel of growth that one can identify is a
productivity increase. Note that we are not saying that the primitives in form of increased impetus
towards liberalization and pro-business policies were not important. In fact we are not commenting
on the primitives at all at this point. Our premise is that whatever means (or primitives as we call
them) any developing nation decides to implement, it would best serve to jump-start growth if it is
2
directed to improve productivity. At least that is the lesson that the Indian miracle seems to teach
us.
Note that the wedges or frictions that aect the rst order conditions of our model, at least
on the face value, resemble taxes. In other words, the impact of these frictions are similar to the
impact of changing taxes, though the frictions represent much more than just taxes. They embody
all forms of distortions that can aect an economy. For example, the labor wedge, on its face value,
resembles a tax on labor income. An increase in the labor wedge is associated with an economic
depression just like an increase in labor income tax would do. However, it is entirely possible that
the labor income tax rates of the economy dont change much but other factors aecting the labor
market do. For example, union bargaining for higher wages would result in a slow growth and
would be captured by the "labor wedge" channel. The eect is thus similar to what would have
happened had the labor income tax rates changed.
In this context, it is worthwhile to note that while the tax rates in an economy are not the be
all and end all of wedges, they do comprise one of the important factor that show up as wedges
in our model and can have far reaching consequences for the economy. In addition, they are the
only market friction for which we have reliable data as provided by the government of a country.
Hence, one natural curiosity is to see to what extent did changing tax rates aect an economy? For
example, if we notice from our decomposition results that labor wedge is important in accounting
for uctuations in an economy, the natural question would be to what extent is it due to changes
in labor income tax rates?
In context of India, note that neither labor wedge nor investment wedge turns out to be impor-
tant in accounting for economic uctuations in India. Does this imply that labor and investment
taxes did not change? Or is it that the eect of changing labor and investment taxes were mitigated
by opposing forces that shows up in our model as labor and investment wedge? To answer this
question, we compute the eective labor and investment taxes for the Indian economy and nd that
while investment taxes did not change much, there was a steady increase in labor income taxes that
by itself would have discouraged growth but were obviously overwhelmed by positive developments
in the economy that kept the labor wedge in our model from worsening.
The rest of the paper is organized as follows: in Section 2 we outline our model. Section 3
discusses the methodology and the results. In Section 4 we present the results of comparing the
wedges with the taxes and Section 5 concludes the paper.
2 Business Cycle Accounting
BCA procedure uses a standard growth model with four stochastic variables or wedges: eciency
wedge
|
, which appears like time varying productivity; the labor wedge t
n|
, which acts like a time
varying tax on labor income, and the investment wedge t
r|
, which acts like a tax on investment
expenditure. Further, per capita government expenditure q
|
, is also considered as government
wedge, which can have a signicant impact on the economy. Each of the wedges represent the
overall distortion to the relevant rst order conditions.
3
2.1 Theoretical model
We assume that the economy every period comprises of a measure
|
of identical and innitely lived
agents who are endowed with one unit of time that can be used for work and leisure. The economy
also consists of measure one of identical rms that own the production process. For purposes of
analysis, we assume that population grows at a constant rate j every period and is exogenously
determined. We assume that there is one output that is produced and consumed in the economy.
There is a government that collects income and investment taxes and uses the proceeds to nance
government expenditure and transfers in such a fashion as to balance the government budget every
period. Given the structure of the economy, we can summarize the problems facing the agents of
the economy as:
2.1.1 Representative consumers problem
The representative consumer in the economy chooses per period consumption c
|
and labor |
|
to max-
imize present discounted value of lifetime utility. The consumer receives income from two sources:
labor income and rental income from capital. In addition, the consumer also receives some transfers
from the government. The proceeds of the income and transfers are used to nance consumption
and investment expenses. Further, the consumer has to pay income (t
n|
) and investment taxes
(t
r|
) to the government at an exogenously determined rate. Thus the representative consumers
problem can be written as:
'ar 1
0
1

|=0
,
|
n(c
|
, 1 |
|
)
subject to:
c
|
+ (1 + t
r|
)r
|
n
|
|
|
(1 t
n|
) + r
|
/
|
+ Tr
|
/
|+1
(1 c)/
|
+ r
|
:o::cqatiitj co::trai:t:
where /
|
denotes per capita capital stock, r
|
denotes per capita investment, after-tax labor
income is n
|
|
|
(1 t
n|
) and rental income is r
|
/
|
where n
|
is the wage rate and r
|
is the rental
rate on capital stock, , is the discount factor, c is the depreciation rate on capital stock, and Tr
|
denotes transfers from the government received at period t.
2.1.2 Representative rms problem
Every period, the representative rm produces a single output using labor and capital to maximize
prots. Output is subject to an exogenously given production technology. Hence the representative
rms problem every period is given by:
4
'ar j
|
n
|
|
|
r
|
/
|
subject to:
j
|
1(/
|
,
|
|
|
)
where j
|
denotes per capita output and
|
denotes productivity. For my analysis I assume that
the production technology is labor augmenting. The long run rate of technical progress is denoted
by (1 + q
:
).
2.1.3 Equilibrium
The equilibrium in this economy is given by a vector of price functions fn
|
, r
|
g
1
|=0
and a vector
of allocation functions fc
|
, |
|
, /
|+1
, j
|
g
1
|=0
such that the price and allocation functions satisfy the
following equations every period:
c
|
+ /
|+1
(1 c)/
|
+ q
|
= j
|
(1)
j
|
= 1(/
|
,
|
|
|
) (2)
n
n|
(c
|
, |
|
)
n
c|
(c
|
, |
|
)
= (1 t
n|
)1
l|
(/
|
,
|
|
|
) (3)
,1
|
n
c|+1
(c
|+1
, |
|+1
)f1
||+1
(/
|+1
,
|+1
|
|+1
) + (1 c)(1 + t
r|+1
)g
= (1 + t
r|
)n
c|
(c
|
, |
|
)
(4)
where notations like n
c|
, n
n|
, 1
l|
, 1
||
etc. denotes the rst derivative of the utility function
and production function with respect to dierent arguments like consumption, labor, and capital.
Equation (1) represents the resource constraint faced by the economy every period and is the output
market clearing condition. Equation (2) shows that output every period is subject to the production
technology. Equation (3) equates the marginal rate of substitution between consumption and leisure
to the after tax marginal return to labor, where in equilibrium, the marginal return to labor or the
wage rate is equal to the marginal product of labor. Equation (4) is the inter-temporal equation
taking into account the fact that in equilibrium, rental rate on capital is equal to the marginal
product of capital. The four equations outlined above summarize the equilibrium conditions of the
economy every period. Note that the time varying productivity and taxes on labor income and
investment expenditure distort the rst order conditions and keeps the economy from achieving the
rst best outcome.
5
2.2 Application to India
We want to apply the BCA technique to India to account for uctuations in output during the
period 1982 to 2002.
The steps involved in BCA accounting are twofold:
(1) Given the parameter values and the rst order conditions, we derive the time series of the
wedges.
(2) Once we have the realized values of the wedges, we feed in the wedges one by one and in
various combinations to see to what extent our model matches the data
5
. This exercise is called
decomposition.
In this paper, we simplify the original method of BCA by assuming that investment wedge t
r|
does not vary over time and let only productivity
|
, the labor wedge t
n|
and the government
consumption wedge q
|
vary. The logic of our exercise is described in the technical appendix, but
the basic idea is that we assume initially that investment wedge is constant at its steady state value
for simplicity. This assumption makes calculating the time series of the productivity and labor
wedges easier as we do not have to rst estimate how agents form expectations over time which is
necessary to get the investment wedge. Assuming investment wedges to be constant circumvents
this problem. Once we get the productivity and labor wedge series, we feed them in our model.
If eciency and labor wedges jointly can well replicate the data, it means that investment wedges
did not play an important role and we were right in holding them constant at their steady state
value. On the other hand, if even after feeding in eciency and labor wedge, the model falls short
of explaining the data, we argue that investment wedges (that our model assumes does not change)
must in reality have varied and have played a major role in the economy. In the technical appendix,
we discuss this method in greater detail and also explain how we derive the time series for the
wedges.
For our exercise we need to specify the utility and the production functions and take into account
the population growth rate while deriving the rst order conditions. We assume a Cobb-Douglas
production function and a standard monotonically increasing and strictly concave utility function
represented by:
n(c
|
, |
|
) =
(c

t
(1lt)
1
)
1
1c
, when o 6= 1
= clog c
|
+ (1 c) log (1 |
|
) when o = 1
(5)
j
|
= /
0
|
(
|
|
|
)
10
(6)
The functional forms that we use are same as those used by Chari, Kehoe and McGrattan (2002)
as well as by Prescott and Hayashi (2002). Note that on a balanced growth path, the variables
c
|
, /
|+1
, j
|
, and q
|
grow at a rate (1 + q
:
). The model does not allow us to calibrate for the
parameter values as we do not know the steady state value of the wedges. So our st step for
solving the model is to pick parameter values from literature. We assume , or the rate of time
5
For details, please refer to the Technical Appendix.
6
preference to be .95 as is commonly used in business cycle literature. We take c to be .8251 and
0 to be .36 from Chakraborty (2006) and we assume that depreciation rate c = .25 that we derive
from the Indian tax code that allows non-residential corporations to claim tax relief for depreciated
capital stock at a maximum depreciation rate of 25%. Once we have our parameter values, we can
calculate the wedges from the rst order conditions of the model and the data. Given the time
series of the realized wedges, we are interested to get an intuitive idea of if they look promising in
generating economic growth in India.
Let us begin by graphically demonstrating the evolution of GDP per capita over the period
1982 to 2002 with respect to a long term time trend of 1.5%. Figure1 graphically demonstrates the
index of detrended GDP per capita. The way we arrive at this gure is by rst detrending GDP
per capita during 1982 to 2002 at the rate 1.5% which is the long term trend growth rate in India
that we derived by taking the average growth rate of GDP per capita during 1960 to 2002. Then,
we take the value of detrended GDP per capita in 1982 as 100 and recalculate the detrended GDP
per capita in the following years with respect to 100. This gives us the index of detrended GDP
per capita and is quite useful in charting how GDP per capita has performed over the years. From
Figure1, we can summarize that GDP per capita has consistently grown above the trend growth
rate in the eighties and the nineties and the rate of growth accelerated since 1991.
Next, we are interested to see whether eciency or labor wedge could have played a role. We can
also look at government consumption wedge that we are not plotting here for the sake of brevity.
Figure2 plots the index of eciency wedge, where we take the value of
|
in 1982 to be a 100. We
nd that productivity also grew during 1982 to 2002 with respect to the long term trend which at
least intuitively is conducive to economic growth.
Figure3 plots the labor wedge, t
n|
. From Equation (3) note that an increase in labor wedge,
or an increase in t
n|
is associated with labor becoming costlier and would therefore cause a drop
in output. Given Figure (.3), note that except for between 1982 to 1983 when we nd labor wedge
declined, labor wedge has not changed much over the last two decades. Now keeping in mind that
labor wedges represent labor market frictions that keep the economy from achieving the rst best
outcome, we should expect economic growth to be associated with a decline in value of labor wedge
or a decline in labor market frictions. So, if labor wedge does not show any such decline, intuitively
it could not have played a role in bringing about economic growth in India. As for the dramatic
decline in 1982 to 1983, we attribute it to a data mis-specication as we nd that labor data shows
a dramatic shift in 1982 which cannot be attributed to any drastic shifts in labor market policy.
Our belief is that labor market data reporting which has been consistent only after the mid eighties
could have somehow contributed to this big jump, and so we do not take this seriously.
3 Decomposition
In this section we show the model outcomes generated by feeding in the realized values of the
wedges one by one and in various combinations in our decision rules and evaluate how well they
can approximate the macro data.
We solve for the decision rules of our model using the log-linearization techniques of Robert
King, Charles Plosser and Sergio Rebelo (1988). The decision rules are derived as:
7
_
_

/
|+1
j
|

|
|
_
_
= 11

/
|
+ QQ
_
_

|
t
n|
q
|
_
_
where 11 is a 3x1 matrix and QQ is a 3x3 matrix of coecients where the log deviation of a
variable .
|
from its steady state value . is denoted by .
|
. The only exception in this specication is
t
n|
which is equal to t
n|
t
n
where we follow Chari, Kehoe and McGrattans (2002) specication.
Note that given decision rule for capital

/
|+1
, output j
|
and labor

|
|
we can implicitly derive the
decision rule for consumption c
|
using the market clearing condition for nal output.
We begin by rst stating the correlations between output, eciency wedge and labor wedge
during the period 1982 to 2002. The correlation between GDP per capita and eciency wedge is
.98 and that between GDP per capita and labor wedge is .1. Given our model, we expect a positive
correlation between productivity or eciency wedge and output and that is supported by correlation
gures. However the correlation between output and labor wedge is positive though small which
indicates that frictions in labor market and output per capita moved in the same direction. This
suggests that output increased despite of labor market frictions, not because of a decline in labor
frictions in which case the observed correlation would have been negative. This begets the question:
was the increase in productivity enough to account for the magnitude of increase in output?
To answer this question, we graphically depict the model outcomes feeding in various realizations
of wedges in our model and comparing them with data. As Figure4 depicts the model outcome
with eciency wedge alone can very well replicate the output per capita observed in India, however
the model outcome feeding in labor wedges cannot explain any of the observed data on output per
capita.
If we feed in eciency, labor and government wedges jointly in our model we can almost wholly
account for observed output per capita in the data which leads us to conclude that investment
wedges played a limited role if at all in the Indian economy during the eighties and the nineties.
This result is consistent with the view that economic growth in India was a result of a sharp increase
in productivity. Government expenditure, that continued to increase over the nineties, also helped
by propping up demand. Labor market rigidities, on the other hand, were still prevalent and
tempered growth to some extent.
To verify our analysis we also look at an alternative macro variable, the capital-output ratio
(Figure5). The results are pretty similar to what we saw for output per capita. With eciency
wedges alone the model well replicates the data on capital-output ratio but feeding in labor wedges
alone we cannot account for the observed capital-output ratio. However, feeding in eciency, labor
and government wedges jointly in the model, the model outcome closely replicates the data which
supports our previous conclusion that economic development in India was a handiwork of increased
productivity and increased government spending.
4 Tax rates and wedges
In the introduction, we mentioned that the labor and investment wedges in business cycle ac-
counting at least on the face value resemble time varying labor income taxes and investment taxes
respectively. We also noted that though market frictions other than policy induced changes in tax
rates also aect the wedges, time varying tax rates by themselves can also aect the wedges and
8
move the economy away from a balanced growth path. While it is dicult to get data for most
frictions and we need to resort to using our model and available national income accounts data to
back out the value of the wedges, we can independently get the data for tax rates and compare that
with the model generated wedges. This analysis helps us with two issues: (1) it helps us answer to
what extent are policy induced changes in tax rates responsible for the time varying wedges and the
resulting eect on the economy (2) if the pattern of time varying taxes and wedges do not match
up, it indicates a need for us to turn our attention to other possible sources of the frictions that
have the potential to aect the wedges and hence the economy.
The results of our decomposition show that productivity wedges by themselves can almost wholly
account for output increases in India during the period 1982 to 2002. This result itself tells us that
there is a limited role of labor wedges or investment wedges. In fact the labor wedge (Figure3)
shows little uctuation except during 1982-1983 that we chalk up to data mis-specication.
Next, we plot the labor income tax rates from the Indian economy. The data is provided by the
Reserve Bank of India
6
. In our model, we do not have heterogenous agents, where one group owns
labor while the others are entrepreneurs, so we do not have a distinction between labor income taxes
and corporate taxes. In our model, the representative agent owns all the income and is responsible
for all the income taxes. While we do not have the eective tax rates per se, we need to calculate
the time varying tax rates from the data on tax revenues and output. The data is split between
central government revenue and the state government revenue. To calculate the labor income tax
rates, we add the revenue from personal income taxes and the corporate taxes of both the central
and state government and divide it by the output. The result is plotted in Figure6.
The eective labor income tax rates according to our measure has gone up steadily throughout
the last two decades. This, by itself, would have put a damper on economic growth but when
we compare this with the labor wedge in Figure3, we nd that labor wedge does not show much
uctuations. This result indicates that while income tax rates in India might have been steadily
increasing throughout the eighties and nineties, there were other changes in the economy that acted
to counter the eect of rising labor income taxes and kept the labor wedge from deteriorating. Hence,
even though the tax rates were increasing, the Indian economic growth continued unabated.
Next, we turn our attention to investment taxes. Now, our decomposition exercise also points to
a limited role of investment wedges. Does this indicate that investment taxes did not change much,
or was the impact of investment taxes overwhelmed by other frictions? While India does not have
an explicit taxes on investment expenditure, from real business cycle literature, we know that we
can use taxes on consumption expenditure as a proxy for investment taxes. To this end, we again
turn to national income accounts data to calculate the eective tax rates. We take the indirect
taxes net of subsidies as a proxy for our investment taxes and divide it by the private consumption
expenditure to get the eective tax rates. We plot the result in Figure7. Note that the eective
tax rates on investment do not show any signicant trends except for a mild decline since 1994 that
would have been conducive to economic growth.
Thus given the time series of taxes, we can conclude that while labor income taxes were rising,
they did not cause a decline in the growth rate possibly due to overwhelming eects of other positive
developments in the economy. The investment tax rates did not change much, thus it is consistent
6
Data is available at the following link: http://www.rbi.org.in/scripts/Statistics.aspx
9
with the fact that we see a limited role of investment wedges, except since 1994, when a decline in
investment tax would have been conducive to growth.
5 Conclusion
After decades of unremarkable growth following independence, Indian economy took o in the
eighties and continued to grow well into the nineties. While a number of economists have suggested
theories to explain this growth phenomenon, empirical studies that attempt a numerical growth
accounting of India is at best limited. In an earlier paper (Chakraborty 2006), we had attempted
to understand the role of technological progress, held by many as the primary reason for Indian
growth using a neoclassical growth analysis.
In this paper, our attempt is to apply a comparatively new procedure of growth accounting called
"Business Cycle Accounting (BCA)" procedure to India. Our objective is to gure out not the macro
fundamentals or "primitives" that brought about economic growth in India but the "transmission
channels" through which the primitives played a role. For example, if we establish that liberalization
policies in the eighties and nineties generated the economic revolution, did liberalization policy act
by increasing productivity, did they reduce frictions in the labor market thus encouraging growth
or did liberalization policies break down investment market barriers? This paper helps us answer
such questions. The BCA procedure is particularly suited for this job as it is based on the key
observation that most primitives aect the economy through productivity, or the labor market or
the investment market and neoclassical model can be used to study these frictions as the frictions
resemble taxes at least on the face value. Thus by solving the neoclassical growth model and
by inserting the frictions one by one and in various combinations we can decipher which frictions
aected the economy the most and thereby understand the most important transmission channel
of the primitive forces that aected the economy. Our results show that primitives aected the
Indian economy primarily by causing changes in productivity. Labor market frictions or investment
market channel was not particularly important. The growth in productivity was also supported by
increasing government expenditure that propped up demand.
The lesson from the Indian growth experience therefore seems to be that targeting productivity
would help jump-start growth for any emerging country that is in the same boat today as India was
a decade ago. This lesson is not a radical one. It was also corroborated by the Japanese experience
during the reconstruction period following the 2nd World War, and the Japanese fall from grace
during the nineties. Prescott and Hayashi (2002) shows us that it was the drop in total factor
productivity that knocked Japan o from its steady growth path that was a consistent feature it its
almost three decades of growth. Note that in this paper, we are not identifying primitives that help
jump-start growth. Instead, we focus on the often neglected channel of growth. Our argument is
that government policies and regime changes that will be most successful in initially jump starting
growth would be the ones that would work by increasing productivity. At least that is what the
Indian and Japanese experience seems to teach us.
Looking back at the eective tax rates, we conclude that labor income taxes are increasing over
time and by themselves would have slowed Indian growth rates though investment taxes do not
show much changes. It is worthwhile, therefore, to explore other changes in the economic climate
that aected the Indian economy through the labor wedge channel and kept the negative eect
10
of increasing labor income taxes from overwhelming the economy. A decline in union bargaining
activities would be a possible direction to look at, particularly in the private sector.
In our view future research should concentrate on the primitives that might have caused increases
in productivity. It would also be interesting to look at micro data and provide evidence of such
technical progress that in turn would help us target the improvement areas at a micro level.
References
[1] Amaral, Pedro & Macgee, James (2002), The Great Depression in Canada and the United
States: A Neoclassical Perspective, Review of Economic Dynamics 5(1), pp. 45-72
[2] Ahearne, Alan; Wynne, Mark and Kydland, Finn (2006), "Irelands Great Depression", Eco-
nomic and Social Review (forthcoming)
[3] Bergoeing, Raphael; Kehoe, Patrick J; Kehoe, Timothy J and Soto, Raimundo (2002), A
Decade Lost and Found: Mexico and Chile in the 1980s, Review of Economic Dynamics 5(1),
pp. 45-72
[4] Chari, V V; Kehoe, Patrick J and McGrattan, Ellen R (2002), Accounting for the Great
Depression, Federal Reserve Bank of Minneapolis Quarterly Review, Volume 27, Number 2
[5] . . . . . . . . . . . . . . . . . . . . . . . . . . . Business Cycle Accounting Revision (2006), Federal Reserve
Bank of Minneapolis Sta Report
[6] . . . . . . . . . . . . . . . . . . . . . . . . . . . Accounting for the Great Depression, American Economic Review
Papers and Proceedings, Vol. 92, No. 2, May 2002
[7] Cole, Harold L. & Ohanian, Lee E (1999), The Great Depression in the United States from
a Neoclassical Perspective, Federal Reserve Bank of Minneapolis Quarterly Review, Volume
23, pp. 2-24
[8] Fischer, Jonas (2004), "Technology Shocks Matter," Working Paper Series WP-02-14, Federal
Reserve Bank of Chicago
[9] Gali, Jordi (1999), Technology, Employment, and the Business Cycle: Do Technology Shocks
Explain Aggregate Fluctuations?, American Economic Review, March 1999, pp. 249-271
[10] Gali, Jordi & Rabanal, Pau (2004), Technology Shocks and Aggregate Fluctuations: How Well
Does the RBC Model Fit Postwar U.S. Data?, NBER Macroeconomics Annual, Forthcoming
[11] Kehoe, Timothy J & Ruhl, Kim J (2003), Recent Great Depressions: Aggregate growth in
New Zealand and Switzerland 1973-2000 New Zealand Economic Papers, 37, pp. 5-40
[12] King, Robert; Plosser, Charles and Rebelo Sergio (1988), Production, growth, and business
cycles: The basic neoclassical model, Journal of Monetary Economics 21(2), pp. 195-232
[13] Kiyotaki, Nobuhiro and Moore, John (1997), "Credit Cycles", Journal of Political Economy,
pp 211-248
[14] Kobayashi, Keiichiro and Inaba, Masaru (2005), "Business Cycle Accounting of the Japanese
Economy", REITI Working Papers
11
[15] Kydland, Finn E. and Prescott, Edward C (1982), Time to Build and Aggregate Fluctua-
tions. Econometrica, November 1982, 50(6), pp. 134570
[16] Mulligan, C. B. (2002). A dual method of empirically evaluating dynamic competitive equi-
librium models with market distortions, applied to the Great Depression and World War
II.NBER Working Paper 8775.
[17] Narayan Murthy, N.R (2002) "The Impact of Economic Reforms on the Hi-Tech Industry in
India: A Case Study of Infosys", The Indian Economy Conference, Cornell University, April
19-20, 2002
[18] Prescott, Edward C (1999), Theory Ahead of Business Cycle Measurement, Federal Reserve
Bank of Minneapolis Quarterly Review, 10(Fall) , 9-22
[19] Prescott, Edward C & Hayashi, Fumio (2002), The 1990s in Japan: A Lost Decade, Review
of Economic Dynamics 5(1), pp. 206
[20] Sachs, Jerey (2002) "Growth Prospects of the Indian Economy" The Indian Economy Con-
ference, Cornell University, April 19-20, 2002
[21] Singh, Nirvikar and Srinivasan, T.N. (2004), "Indian Federalism, Economic Reform and Glob-
alization," Public Economics 0412007, Economics Working Paper Archive EconWPA
[22] Singh, Nirvikar (2004), "Information Technology as an Engine of Broad-Based Growth in In-
dia," Development and Comp Systems 0412012, Economics Working Paper Archive EconWPA
[23] Subramanian, Arvind and Rodrick, Dani "From Hindu Growth to Productivity Surge: The
Mystery of the Indian Growth Transition," CEPR Discussion Papers 4371, C.E.P.R. Discussion
Papers
12
GDP per capita with respect to long term trend
0
20
40
60
80
100
120
140
160
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Years
I
n
d
e
x

(
1
9
8
2
=
1
0
0
)
y(t)
Figure 1: GDP per capita detrended at 1.5%
Note: We plot the GDP per capita after detrending it by the long term growth rate of 1.5%. The GDP
per capita of 1982 is assumed to be 100 and we plot the index.
13
Efficiency wedge
0
20
40
60
80
100
120
140
160
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Year
I
n
d
e
x

(
1
9
8
2
=
1
0
0
)
Efficiency wedge
Figure 2: Realized eciency or productivity wedge
Note: We plot the eciency wedge as a Solow residual. Once again, the eciency wedge in 1982 is
assumed to be 100.
14
Labor wedge
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Years
T
a
x

r
a
t
e
Labor wedge
Figure 3: Labor wedge
Note: We plot the labor wedge as calculated from the data and our model.
15
Output per capita: Data and model outcomes
0
20
40
60
80
100
120
140
160
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Year
I
n
d
e
x

(
1
9
8
2
=
1
0
0
)
y(t)
y(t)-model
y(t)-eff
y(t)-taxn
Figure 4: Output per capita: Data and Model outcomes
Note: We plot the GDP per capita from data and as generated by our model by feeding in eciency
wedge and labor wedge respectively (the legend reads y(t)-e and y(t)-taxn)). Next, we feed in eciency
wedge, labor wedge and government consumption wedge jointly in our model and plot it (legend reads
y(t)-model).
16
Capital-output ratio: Data and Model outcomes
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Years
R
a
t
i
o
k(t)/y(t)
k(t)/y(t)-model
k(t)/y(t)-eff
k(t)/y(t)-taxn
Figure 5: Capital-Output ratio Data and Model outcomes
Note: We plot the capital output ratio from data and as generated by our model by feeding in eciency
wedge and labor wedge respectively (the legend reads k(t)/y(t)-e and k(t)/y(t)-taxn)). Next, we feed in
eciency wedge, labor wedge and government consumption wedge jointly in our model and plot it (legend
reads k(t)/y(t)-model).
17
Direct tax rate
0
1
2
3
4
5
6
7
8
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Years
I
n

p
e
r
c
e
n
t
a
g
e
Direct tax rate
Figure 6: Direct tax rates
Note: We plot the labor income taxes from the data. It is measured as the ratio of the direct tax
revenue of the central as well as state government to output.
18
Indirect tax rate
0
2
4
6
8
10
12
14
16
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Years
I
n

p
e
r
c
e
n
t
a
g
e
Indirect tax rate
Figure 7: Indirect tax rates
Note: We plot the investment taxes from the data. It is measured as the ratio of the indirect tax revenue
net of subsidies of the economy to the personal consumption expenditure.
19
TECHNICAL APPENDIX
The rst order conditions of the model outlined in Equations (1) to (4) can be simplied to
the following three equations by substituting the value of consumption c
|
from Equation (1) and
replacing it in Equations (3) and (4). Taking into account the population growth rate j, and
discounting the model variables with respect to their long term trend (1 + q
:
), the fundamental
equations of our model reduces to:
j
|
= 1(

/
|
,

|
|
|
) (7)
n
n|
(c
|
( j
|
, r
|
, q
|
), |
|
)
n
c|
(c
|
( j
|
, r
|
, q
|
), |
|
)
= (1 t
n|
)1
l|
(

/
|
,

|
|
|
) (8)
,1
|
n
c|+1
(c
|+1
( j
|+1
, r
|+1
, q
|+1
), |
|
)f1
||+1
(

/
|+1
,

|+1
|
|+1
) + (1 c)(1 + t
r|+1
)g
= (1 + t
r|
)n
c|
(c
|
( j
|
, r
|
, q
|
), |
|
)(1 + q
:
)
(9)
where I denote a variable .
|
detrended by the long-term growth rate of technological development
(1 + q
:
)
|
as .
|
where .
|
=
:t
(1+z)
t
.
Given the wedges

|
, t
n|
, t
r|
, and q
|
, the equations (7) to (9) solve for output, investment
and labor in terms of the wedges. The BCA procedure involves feeding in the wedges one by one
and in dierent combinations to see which wedges or combinations of wedges can best replicate the
data. The accounting procedure has two parts: rst we need to estimate the wedges from the data
and then we feed in the wedges in our model to generate output, labor and investment. This later
procedure is called decomposition. Note that by construction of the BCA procedure, if we feed
in eciency, labor, investment and government wedges in the model all together, then we will get
back the data.
Taking into account the population growth rate, and the functional forms outlined in equations
(5) and (6), equations (7) to (9) reduces to:
j
|
=

/
0
|
(

|
|
|
)
10
(10)
_
_
1o
o
_
_
b tq(1+z)
b
|t+1+(1o)
b
|tb t
1lt
_
= (1 0)(1 t
n|
)
b t
lt
_
(11)
_
o
(1+z)
1
|
_
b tq(1+z)
b
|t+1+(1o)
b
|tb t
b t+1q(1+z)
b
|t+2+(1o)
b
|t+1b t+1
__
0
b t+1
b
|t+1
+ (1 c)(1 + t
r|+1
)
_
= (1 + t
r|
)
_
(12)
where the value of o is taken as 1.
Note that given parameter values we can solve equations (10) to (12) and get decision rules for
output j(t), labor |(t), and capital stock next period /(t + 1) in terms of productivity or eciency
wedge (t), labor wedge t
n|
, investment wedge t
r|
and government consumption wedge q(t). Once
we get the decision rules, we can plug in the time series of the wedges one by one in our decision
rules while holding other wedges constant at their steady state values and thereby account for the
contribution of each wedge in generating the macro variables.
20
The problem here is that we do not have time series data available on productivity
|
, labor
wedge t
n|
, and investment wedge t
r|
as they represent market frictions and are therefore intangible.
So we need to use data from national income accounts and our equations to back out the values of
these wedges. The job is relatively simple for eciency wedge (t), and labor wedge t
n|
which we
can derive given equations (10) and (11) and the time series data on output j(t), labor |(t), and
capital stock next period /(t + 1).
The job is not so easy for calculating investment wedges t
r|
as it involves knowing not only the
time series of aggregate macro data but also the expectations that people form about the future as
equation (12) highlights. Researchers have used many variations to get around this problem. Chari,
Kehoe and McGrattan (2003) hold eciency and labor wedges constant at their steady state values
and let investment wedges be whatever they should be so that they can replicate the investment
data exactly to get an approximate idea for the series and then they iterate such that their model
outcomes (feeding in all the wedges) can replicate the data exactly. Others, including Keiichiro
Kobayashi and Masaru Inaba (2005) work with a deterministic form of the model to get around
the problem of forming expectations.
What we do in our analysis is to hold the investment wedges constant at their steady state
value. We then ascertain how much of the observed data can we generate with eciency, labor and
government wedges jointly. If after feeding in eciency, labor and government wedges, we still have
a large part of observed data still unexplained, then we can say that investment wedges must have
played a signicant role as by construction, the model is supposed to replicate the data exactly
when all four wedges are fed in.
Thus with our simplication, the equations reduce to:
j
|
=

/
0
|
(

|
|
|
)
10
(13)
_
_
1o
o
_
_
b tq(1+z)
b
|t+1+(1o)
b
|tb t
1lt
_
= (1 0)(1 t
n|
)
b t
lt
_
(14)
_
o
(1+z)
1
|
_
b tq(1+z)
b
|t+1+(1o)
b
|tb t
b t+1q(1+z)
b
|t+2+(1o)
b
|t+1b t+1
__
0
b t+1
b
|t+1
+ (1 c)(1 + t
r
)
_
= (1 + t
r
)
_
(15)
where t
r
is the steady state value of investment wedge t
r|
.
Rearranging the terms, the equations can be written as:

|
=
_
j
|

/
0
|
|
10
|
_ 1
1
(16)
t
n|
= 1
_ _
1o
o
_

1
(10)

lt
b t
_
b tq(1+z)
b
|t+1+(1o)
b
|tb t
1lt
_
_
(17)
t
r
= 1
0

|
(1+z)
o
1 + c
(18)
21
where equation (18) is the steady state variation of equation (15) and helps us to get the steady
state value of the investment wedge, t
r
. O course, one can easily get the steady state value of
the eciency wedge and labor wedge that we denote by and t
n
respectively from the steady
state version of equations (16) and (17). Note that given the data from National Income Accounts,
equations (16) and (17) give us the time series of productivity

|
and that of labor wedge t
n|
.
Government consumption wedge q
|
is taken from National Income Accounts.
22