Article Type : Research Article
Authors : Kumar Sinha J
Keywords : Economic growth; Per capita state domestic product; Growth convergence; Inequality
This study undertakes
four decades of data for Indian states from the 1980s and traces the
convergence of state-level per capita income, which has always been one of the
most important economic phenomena for Indian states. A total span of four
decades has been broken down into three sub-periods, viz., 1980-1992,
1993-2003, and 2003-2020 based on time and levels of income using panel unit
root tests. The results show no discernible evidence of convergence across the
states, especially after post-liberalization. The panel unit root tests suggest
no evidence of convergence over the whole-time period of 1980-2020 when all the
regions are taken together. The pre-liberalization period of 1980-1992 showed
more evidence of convergence among the regions as compared to the
post-liberalization period of 1992–2020 which does not exhibit any convergence.
After dividing the post-liberalization period into sub periods of 1992-2003 and
2004-2020 some sign of convergence is found; however, the number of regions
exhibiting any significant convergence is less in number than such regions in
the pre-liberalization period (three states in 1992-2003 and five states in
2004-2020 vs six states in 1980-1992). However, taking into account control
variables significant evidence for convergence of state-level per capita GDP
was traced. This indicates that inequality across states could be reduced
through active policy interventions through control variables for capital
expenditure, development expenditure, and fiscal deficit.
Seventy-five years since independence, eradication of
poverty and inequality have been the prime objective of every government. Yet,
India remains the home of one-third of the world’s poor. As India aims to join
the global superpowers, shedding the tag of a developing country, there is a
pressing need to evaluate how successful the country has been in this
endeavour. India is a diverse country with significant regional variations in
economic development. The per capita GDP of Indian regions has shown significant
disparities over the years, with some regions consistently outperforming
others. According to data from the World Bank, the per capita GDP of India in
2020 was $1,947. However, the per capita GDP of individual states and union
territories varied widely. In 2020, Delhi had the highest per capita GDP of
$7,172, while Bihar had the lowest per capita GDP of $528. Data from the
Reserve Bank of India shows that there has been some convergence in per capita
GDP across Indian states over the past few decades. Between 1980 and 2020, the
gap between the highest and lowest per capita GDP states narrowed. In 1980, the
ratio of the per capita GDP of the highest and lowest states was 9:1, while in
2020, it was 5.5:1. However, despite this convergence, significant disparities
remain. In 2020, the per capita GDP of Delhi was more than 13 times higher than
that of Bihar, the state with the lowest per capita GDP. Several factors
contribute to the regional disparities in per capita GDP in India, including
differences in natural resources, infrastructure, education levels, and
economic policies. Efforts to reduce these disparities have included targeted
government programs and policies to promote investment in underdeveloped
regions and sectors, as well as efforts to improve education and
infrastructure. Since independence, the performance of several regions has
consistently been below the national average. Economic policies of the
government since independence were aimed at pulling these regions and their
people out of poverty. But the question of whether and how far these policies
have been effective has remained a concern. The inter-region disparity in per
capita income has been an ever-present feature in the Indian economy. Although
the regional disparity cannot be eliminated, policies should work toward
narrowing down the inter-regional differentials in per capita state gross
domestic product, leading to a reduction in the disparity. India is considered
one of the fastest-growing economies in the world along with China, in recent
times. Amid complex regional heterogeneity across 29 states and seven union
territories of India, in terms of culture, language, social norms, and economic
outlook, one common question is often asked-“are the regions all growing toward
convergence in the backdrop of economic growth?”1 Several studies have
evaluated the per capita income convergence of regions, but no clear consensus
has been reached so far. More surprisingly the existing economic bilateralism
in the subcontinent makes us extremely curious about the distribution dynamics
of income across the Indian regions. Thus, there is a need to examine the idea
of income convergence of the various income-divergent regions primarily for two
reasons. First, more than seven decades of planned economic development have
already taken place. This exercise may help us figure out how successful the
strategic planning was in terms of a reduction in income inequality among
states. Second, more than three decades since economic liberalization (1991)
that aimed to move away from years of protectionism to make the economy more
market-driven, it would be judicious to evaluate the impact of releasing the
caged tiger on the per capita income levels of Indian States. The existence of
huge shopping complexes, plants of multinational firms, the development of
infrastructure, the achievement of BSE-SENSEX (S&P Bombay Stock Exchange
Sensitive Index2) crossing the 40,000 mark (in May 2019), immense expansion of
telecommunication networks, the presence of multi-national food chains, the
rapid expansion of real estate sector, and many more such examples are at the
core of the growth narrative. However, the seminal work of shows that shining
India corresponds to only the top 10 percent of the population, rather than the
middle class or the middle 40 percent [1]. The 1980–2020 growth story of the
middle class (102%), when put in the context of average growth (187%), seems
significantly low. Thus, a reality check in terms of per capita income, which
proxies the standard of living, is very relevant. Despite the importance of
income growth, what remains even more crucial is equitable growth across all
income groups. This brings to the concept of convergence and divergence of
incomes.
This study aims to find out the convergence pattern in
the per capita state/region GDPs (SGDP) across the Indian states and union
territories, and tries to answer the questions;
1. Is
there any convergence among the states/ territories before and after the
liberalization policy introduced in 1992?
2. Is
there any convergence among the states/territories for the sub-period before
and after the liberalization policy?
The study examines if the poorer Indian states are
indeed converging toward the richer states and whether liberalization has had
any beneficial effect on the said convergence. In the case of convergence, it
may fairly be assumed that the economy has been partially successful in
addressing the poverty issue. If there is no evidence of convergence, then it
points toward a serious situation that calls for an immediate policy
revaluation and deeper introspection. The study contributes to the existing
literature primarily from three different aspects: i) the dataset covers more
than four decades from 1980; ii) relevant variables are controlled as suggested
by literature and then the panel unit root test is applied to find any evidence
of convergence; and iii) finally, the entire data- set both in terms of the
period and levels of income at their initial stage are delved deeper and divided
to trace out any evidence of convergence. However, if initial results indicate
that over various periods and sub-time periods, states have shown little or no
convergence over time the study proceeds to find evidence of convergence by
taking into account capital expenditure, fiscal deficit, and total development
expenditure which strongly implies that there is an active role of state
policies in eliminating the lack of convergence or divergence among Indian
states.
The theory of convergence to steady state stems from
the central assumption of the Solow growth model-diminishing returns to
reproducible capital [2]. It states that “Two countries that are the same in
all their parameters—savings rates, population growth rates, rates of technical
progress, etc. must ultimately exhibit similar levels of per capita income.” As
per the Solow–Swan model in a closed economy, the capital-labor (K/L) ratio
causes variations in per capita incomes across economies. Keeping the savings
rate constant, a lower initial K/L ratio is associated with a faster
proportionate increase in K/L. This occurs due to factor mobility—labor will
migrate from poor to richer states while capital will migrate from rich to poor
[3]. The two types of tests often used in convergence analysis are—?
convergence, which measures the proportionate growth in per capita income on
the initial level of per capita income, and ? convergence, which measures the
cross-sectional dispersion of per capita incomes. ? convergence can be further divided
into conditional and absolute convergence. The former is perceptible only after
other factors which may cause variation in steady states have been accounted
for while the latter is a stronger kind of convergence, where initially poor
states grow faster, notwithstanding the differences in initial conditions [4].
In most studies, researchers test for conditional convergence. While
conditional convergence is crucial for cross-country regressions, in the case
of intra-country regression, suggest that the effective output per capita and
technological progress do not differ much, and hence, can be tested for
absolute convergence instead [5]. Using the same reasoning, we test for
absolute convergence for the various states of India. Tested for convergence among
states of the US and found that the value of responsiveness to the average
growth rate (?), is directly proportionate to the speed of convergence to the
steady state. This methodology has been adopted by several researchers in the
context of India [6]. At the time of independence in 1947, India’s GDP was `2.7
trillion which has grown almost exponentially to stand at `57 trillion in
2013–2014. Despite India’s overall stellar growth, several critics have
expressed concern over the growth of individual states in India. As per the
poverty estimates (2004–2005 and 2011–2012) during this period, more than half
of the total poor lived in six states, such as Bihar, Chhattisgarh, Jharkhand,
Madhya Pradesh, Rajasthan, and Uttar Pradesh [7]. Some states have experienced
growth better than others. As India celebrated its 75th year of independence in
2022, a key question remains how far it has progressed from 1947 to the third
decade of the twenty-first century? Analysed interstate differentials between
1950 and 1960 and showed no noticeable reduction in income differentials [8].
Suggested that the degree of inequality had remained unchanged whereas showed a
steady increase in interstate inequalities [9,10]. Before the 1980s, GDP growth
had stagnated at a dismal 3 percent per annum for almost 20 years, which shot
up to 5 percent in 1980–1989, that further increased to 6 percent in the
1990–1999 period [11]. The proportion of poor below the poverty line has
significantly declined from 44.5 percent in 1983-1984 to 27.5 percent in
2004-2005. This brings us to one of the most important questions-have all the
states been able to reap the benefits of liberalization uniformly? The study
states that the probability of convergence is the highest when national or
regional economies are linked by open trade and factor mobility. This makes the
1991 reforms act as a perfect ground for a natural experiment aimed to test the
aforementioned hypothesis. Believe that
growth was kick started post-economic liberalization of 1991 since it provided
domestic firm’s access to capital equipment embodied with new technology, and
better intermediate inputs, and expanded their choice set to act. With free
markets came creative destruction increasing overall productivity, especially
in the service sector. The introduction of cell phones and the diffusion of the
internet paved the way for this revolution which aided the fastest-growing
sector in India—business services. However, it is perplexing how an economy
that was majorly employed in the unorganized sector was able to gain from these
reforms. Point out certain channels, such as direct absorption of technology
(e.g., cell phones), cheaper products from the organized sector leading to an
increase in real wages in the unorganized sector, and demand spill overs from
increased incomes of the organized sector. This further suggests how
liberalization has the potential to affect economies such as India. However,
the question remains whether these sources are significant enough to enhance
the per capita income, particularly in the poor regions of any country. The
broader literature points out that in the presence of significant human capital
and skills, the effect of liberalization is higher, probably three or fourfold
more than that of a skill-adverse region. Our results do support such a claim
to some extent. States which experienced higher growth post-liberalization are
the ones that already had more human capital, better infrastructure, and labour
laws. Specific studies, such as the one conducted by, use both ? and ? tests of
unconditional convergence. The sigma convergence test shows that throughout
1961–1971, there was a reduction in regional disparities due to significant
improvement in the agrarian economy and technological inputs. However,
throughout 1972–1982, with the slowdown of industrial growth in India, there
was a divergence in the disparities. The beta convergence test shows a
statistically significant convergence for the first period whereas for both the
other periods, the beta value comes out statistically insignificant, which
suggests divergence. The results from the study by showed the presence of a
V-factor in Indian states’ growth turnaround that is consistent with policy
reforms [12]. In an extension of this study, found largely ambiguous evidence
of convergence across Indian states [13]. However, they found that states with
higher literacy, urbanization, and access to ports participated more in the
growth process. More concrete pieces of evidence for conditional convergence
came in recent years from Das, where the authors further extended the data till
2007–2008 and found conditional convergence at the district level once
controlled for district characteristics [14]. It is now a fact that incomes
have seen substantial improvement since the 2000s, growing at 4.4 percent;
historically which had never been higher than 2.5 percent. Use tax data to
study how incomes have grown since 1922, the year when the income tax was
implemented. It turns out that current income inequality is much higher than in
the pre-independence period—the top 0.1 percent income share remained between 5
and 7 percent before 1922 as opposed to greater than 8 percent in 2015. The top
1 percent income share is at its highest level (22%) ever since the creation of
income tax. On the other hand, the bottom 50 percent grew at a rate much lower
than average, accompanied by the middle 40 percent that grew at a marginally
better albeit lower than average rate. This data indicates a divergence in
income. Table 1 provides a snapshot of the level of income inequality across
income groups at two different time points 30 years apart. The first column
exhibits the share of incomes in 1982–1983 and the second shows the share in
2013–2014, highlighting the widening gap among the top, middle, and bottom
percentiles. Show that although most Indian states have had an upward trend
from 1961 to 1990, there has not been much evidence of convergence [15]. On the
contrary, the stronger states grew even stronger. In the background of the
contemporary literature in this particular area of research answers to the
research questions mentioned in Section 2 were searched. The following sections
explain the data and discuss the model and estimation, analysis and results,
and lastly conclusions and recommendations.
1. This study covers a period of forty years in this
article from 1980 to 2020. The entire period was divided into three
sub-periods: 1980–1992, 1993–2003, and 2003–2020. The first sub-period from
1970 to 1992 is the
pre-liberalization period. The next 28 years have been also split into two
sub-periods. The sub-periods are created to check if there is any evidence of
convergence pre/post-liberalization. Also, 2002 was a major setback for most of
the countries both in terms of the socio-political and economic scenario. In
2002, mentioned that India was losing economic momentum, inflation had
decreased from high levels, and private balance sheets experiencing the costs
of an investment boom funded by the credit bubble [16]. Hence, this particular year
formed a basis for investigation before and after that year.
2. India is currently divided into 29 states and seven
union territories. However, the limitation of relevant comparable data
restricts our sample to 26 states and union territories, which represent almost
more than 90 percent of the country.
The PCSGDP convergence pattern was examined in two
stages: i) without taking into consideration any control variables, and ii)
with consideration of control variables.
Approach was used to detect any discernible
convergence across the states’ per capita GDP. This approach defines
“convergence as equality of long-term forecasts at a fixed time. States i and j
converge if both states' long-term forecasts of (log) per capita output are
equal at a fixed time t” [17].
lim E (yi,t+k ? y j,t+k |It ) = 0
If the differential yi–yj contains a unit root, the
conditions of convergence are violated. Two-panel unit root tests; namely and
the other one the panel unit root test were conducted [18,19]. Levin, Lin, and
Chu test consider the specification from the augmented Dickey-Fuller test
yt = ????yt – 1 + xt’? + €t
?¥yt = ????yt
–1 + xt’? + €t
Where the difference operator is
????
= ????
– 1,
xt
are exogenous regressors, ????,
and ? are parameters and €t is white noise
Taking
p lagged differences,
¥yit = ????yit-1 +??ij ¥yit-1 + xt’it? + €t (1)
The
LLC test uses two additional regressions, regressing ¥y’it and y’it-1 and lagged value of ¥yit and the exogenous
regressors.
¥y’it =¥yit
- ??ij ¥yit-1 + xt’it?
&
y’it-1 = yit-1 - ??ij ¥yit-1 + xt’it?
(2)
Both ¥y’it
¥y’it & y’it-1 are adjusted using the standard error from
the regression, i.e.
¥y’it = ¥y’it /si (3)
Y’it-1 = y’it-1 / si (4)
And
¥y’it =????
y’it-1 + ?it is the pooled proxy equation from (3) and
(4) (5)
The
t-statistic for the estimated ????,
for LLC, under the null hypothesis, is
T*?
= [T? – (NT) SN ????*2se
(????*)
µmT*]/ ????mT*
is N (0,1) (6)
?i = 0 Where T* = T – (?p/N) – 1, and p is the number
of lags in each cross-section ADF regression. Acceptance of the null hypothesis
in LLC would imply the presence of a unit root.
The IPS panel unit root test begins with a similar
premise by specifying a separate ADF regression from each cross-section.
¥yit = ????yit-1 +??ij ¥yit-1 + xt’it? + €t , where the null hypothesis is H0 : ?i
=0, for all I ; and the alternate hypothesis H1 = {?i
=0 , for i; ?i <0
for i= N+1, N+2, ….}
Average
Per Capital SGDP of the States
This study covers 26 Indian States. Table 2 provides
the average per capita state GDP (PCSGDP) for all states for the whole period,
that is, from 1980 to 2020. While the nationwide average PCSGDP has been about
`4,501.1, Delhi, Goa, and Sikkim boast of average PCSGDP of `11001.6,
`11,141.6, and `11,409.6 respectively. On the other hand, Bihar (`1,581.1),
Assam (`1,969.2), Uttar Pradesh (`2069.8), Orissa (`2,232.0), Madhya Pradesh
(`2,349.5), Jammu and Kashmir (`2,638.1), and Rajasthan (`2,672.7) have been
the poorest performing states. Only Delhi, Goa, Gujarat, Haryana, Maharashtra,
Pondicherry, Punjab, and Sikkim have a higher PCSGDP than the national average.
The average performance over the four decades indicates the plight of the
poorly performing BiMARU (Bihar, Madhya Pradesh, Rajasthan, and Uttar Pradesh) states
with Jammu and Kashmir as an inclusion.
Growth
Rate of PCSGDP
The growth of State Gross Domestic Product (SGDP) is
calculated using the available values of SGDP provided in the Indian Public
Finance Statistics (IPFS) published by the Government of India. The ranking of
these 26 states in the terminal period of the study, viz., 1980-81 and 2019-20,
changed in their respective rank during the span of four decades and the growth
rates are mentioned in Table 3. The performance of Sikkim has been outstanding
during this overall period 1980-81 to 2019-20 (Table 3). It went from one of
the middle-performing states (rank 12) in 1980 to the best-performing state in
2019–2020; it’s PCSGDP increased more than 28 times during this phase. Tripura
has shown the largest jump in ranking, from the 22nd ranked state in 1980–1981,
it has gone to become 9th ranked state in 2019–2020, its PCSGDP increasing more
than eight times. Jammu and Kashmir have been the poorest performing state in
terms of drop in the ranking. The state has declined from the 8th
best-performing state in 1980–1981 to the 21st position in 2019–2020, with its
PCSGDP rising only 2.71 times.
Table 1: Income Inequality in India.
Percentile |
1982-83 |
2013-14 |
Top 1 % |
6.2 |
21.7 |
Top 0.1% |
1.7 |
8.6 |
Middle 40% |
46.0 |
29.6 |
Bottom50 % |
23.6 |
14.9 |
Table 2: Average Per Capita State GDP between 1980 and 2020 (values in Indian Rupees).
State |
PCSGDP |
State |
PCSGDP |
1. Andhra Pradesh |
3541.5 |
2. Maharashtra |
6381.7 |
3. Arunachal Pradesh |
3501.3 |
4. Manipur |
2629.2 |
5. Assam |
1969.2 |
6. Bihar |
1581.1 |
7. Nagaland |
3431.9 |
8. Meghalaya |
2668.5 |
9. Delhi |
11001.6 |
10. Orissa |
2232.0 |
11. Goa |
11141.6 |
12. Pondicherry |
7115.8 |
13. Gujarat |
5328.5 |
14. Punjab |
5114.5 |
15. Haryana |
5203.3 |
16. Rajasthan |
2672.7 |
17. Himachal Pradesh |
3608.7 |
18. Sikkim |
11409.6 |
19. Jammu & Kashmir |
2638.1 |
20. Tamil Nadu |
4297.5 |
21. Karnataka |
4102.2 |
22. Tripura |
3822.2 |
23. Kerala |
3295.5 |
24. Uttar Pradesh |
2069.8 |
25. Madhya Pradesh |
2349.5 |
26. West Bengal |
3756.1 |
Average of the 26 States |
4501.1 |
||
Source: Department of Economic
Affairs, Indian Public Finance Statistics (GoI). |
|
Table 3: Ranking of Indian States according to their PCSGDP Between 1980–1981 and 2019–2020 and Annual Growth Rate.
State |
Rank
in 1980-81 |
Rank
in 2019-20 |
Change
in Rank during 1980-81 to 2019-20 |
Average
annual Growth Rate during 1980-81 to 2019-20 |
Andhra Pradesh |
17 |
16 |
1 |
5.64 |
Arunachal Pradesh |
11 |
17 |
- 6 |
3.95 |
Assam |
23 |
26 |
-3 |
2.46 |
Bihar |
26 |
25 |
1 |
3.67 |
Delhi |
1 |
3 |
-2 |
7.47 |
Goa |
2 |
2 |
0 |
10.21 |
Gujarat |
7 |
6 |
1 |
7.13 |
Haryana |
6 |
7 |
-1 |
4.84 |
Himachal Pradesh |
10 |
14 |
-4 |
4.65 |
Jammu & Kashmir |
8 |
21 |
-13 |
2.73 |
Karnataka |
13 |
10 |
3 |
6.24 |
Kerala |
14 |
15 |
-1 |
5.25 |
Maharashtra |
5 |
5 |
0 |
6.53 |
Manipur |
16 |
20 |
-4 |
3.45 |
Meghalaya |
18 |
18 |
0 |
4.18 |
Madhya Pradesh |
20 |
22 |
-2 |
3.47 |
Nagaland |
19 |
12 |
7 |
6.47 |
Orissa |
21 |
23 |
-2 |
3.06 |
Pondicherry |
3 |
4 |
-1 |
6.63 |
Punjab |
4 |
11 |
-7 |
3.46 |
Rajasthan |
25 |
19 |
6 |
4.53 |
Sikkam |
12 |
1 |
11 |
28.20 |
Tamil Nadu |
15 |
8 |
7 |
7.41 |
Tripura |
22 |
9 |
13 |
8.15 |
Uttar Pradesh |
24 |
24 |
0 |
2,74 |
West Bengal |
9 |
13 |
-
4 |
4.49 |
Source: Department of Economic Affairs, Indian Public
Finance Statistics (GoI.). |
Table 4: Panel Unit Root Test (Levin, Lin, and Chu) Including Subperiods.
State |
1980-2020 Whole
Study period |
1980-1992
Pre
liberalization period |
1992-2020 Post- liberalization period (PLP) |
||
1992-2003 PLP
part-1 |
2004-2020 PLP
part-2 |
1992-2020 Total
PLP |
|||
Andhra Pradesh |
-0.68 |
-1.30 |
-2.61 |
-2.03 |
-1.76 |
Arunachal Pradesh |
-1.68 |
-1.76 |
-1.83 |
-1 84 |
-1.84 |
Assam |
0.08 |
-1 68 |
0.55 |
- 2.59 |
- 1 58 |
Bihar |
0.09 |
- 0.91 |
- 4.18* |
- 2.14 |
1 75 |
Delhi |
0.70 |
- 071 |
- 2.71 |
- 2.81 |
1.73 |
Goa |
-0.70 |
-1.86 |
-2 54 |
-2.06 |
-1.81 |
Gujarat |
0.05 |
-2.79 |
-2 85 |
-2 55 |
1 35 |
Haryana |
-0.30 |
-3.28 |
0.42 |
-2.78 |
-1.62 |
Himachal Pradesh |
-0.30 |
-1.97 |
-2.76 |
-1.65 |
-1 61 |
Jammu & Kashmir |
-0.84 |
-2.57 |
-2.57 |
-3.75* |
-1.80 |
Karnataka |
-0.98 |
-1.87 |
-3.98* |
-2.00 |
-2.01 |
Kerala |
0.08 |
-1.97 |
-0.32 |
-4.76** |
-1.45 |
Madhya Pradesh |
2.42 |
-2.87 |
-2.34 |
-1.67 |
-0.55 |
Maharashtra |
2.63 |
-2.07 |
-267 |
-4.56** |
-1.23 |
Manipur |
-0.65 |
-5.96** |
-3.43 |
-1.65 |
-1.68 |
Meghalaya |
0.23 |
1.87 |
-2.24 |
-1.35 |
-1.98 |
Nagaland |
-0.03 |
-4.90** |
-1.87 |
-2.14 |
-1.55 |
Orissa |
-1.29 |
-4.98** |
-3.56* |
-0.65 |
-1.75 |
Pondicherry |
-1.32 |
-2.62 |
-2.91 |
-3.13 |
-1.78 |
Punjab |
-0.23 |
-4.68** |
-2.03 |
-1.41 |
-1.57 |
Rajasthan |
-0.62 |
-2.97 |
-3.57 |
-1.97 |
-1.55 |
Sikkim |
2.49 |
-2.06 |
-1.76 |
-2.08 |
4.06 |
Tamil Nadu |
0.24 |
-0.56 |
-1.76 |
-4.78** |
-1.24 |
Tripura |
1.49 |
-4.58** |
-2.18 |
-0.34 |
0.76 |
Uttar Pradesh |
-0.23 |
2.11 |
-2.78 |
-2.34 |
1.45 |
West Bengal |
-0.25 |
-4.78** |
2.73 |
-4.47** |
-1.45 |
All 26 States |
5.66 |
-6.10*** |
-3.91*** |
-11.28*** |
0.19 |
Source: The computations are the author’s own. Note: ***, **, and * represent significance
at 1%, 5%, and 10% levels, respectively. |
Table 5: Panel Unit test result.
Time
Phase |
All
States |
High-Income
States |
Middle-income
States |
Low-Income
States |
||||
LLC |
IPS |
LLC |
IPS |
LLC |
IPS |
LLC |
IPS |
|
Whole Period (1980-81 to 2019-20) |
4.39 (1.00) |
12.89 (0.99) |
2.42 (0.98) |
7.10 (1.00) |
1.99 (0.98) |
6.80 (1.00) |
3.44 (0.99) |
7.88 (1.00) |
Pre-Liberisation Period (1980-81 to 1991-92) |
-8.87* (0.00) |
-2.56* (0.00) |
-4.70* (0.00) |
-1.43** ()0.05 |
-5.60* (0.00) |
-1.00 (0.15) |
-5.03* (0.00) |
-2.03** (0.02) |
Post-Liberalisation Period (1992-93 to 2019-20) |
2.60
(1.00) |
4.87 (1.00) |
-1.18*** (0.06) |
2.76 (0.99) |
-1.94** (0.02) |
2.58 (0.99) |
-1.13* (0.10) |
3.03 (0.99) |
Post-Liberalisation Period (1992-93 to 2002-03) |
7.56 (1.00) |
0.90 (0.18) |
-4.70* (0.00) |
-0.66 (0.02) |
-4.61* (0.00) |
-0.62 (0.27) |
-4.08 (0.00) |
-0.27 (0.39) |
Post-Liberalisation Period (2003-04 to 2019-20) |
19.73 (1.00) |
2.98 (0.98) |
-8.79* (0.00) |
-1.65 (0.05) |
-14.98* (0.00) |
-0.22 0.41() |
-8.00 (0.00) |
-0.47 (0.31) |
Source: The computations are the
author’s own. Note: ***, **, and
* represent significance at 1%, 5%, and 10% levels, respectively. |
Table 6: Capex, Fiscal Deficit & Devextot.
Table 7: Capex, Education, Health, Welfare, and Fiscal Deficit.
|
All States |
High Income |
Middle Income |
Low Income |
||||
LLC |
IPS |
LLC |
IPS |
LLC |
IPS |
LLC |
IPS |
|
Fixed Effect |
-6.64*** (0.000) |
-2.70*** (0.002) |
-2.18*** (0.010) |
-0.83** (0.132) |
-4.44*** (0.000) |
-1.81** (0.037) |
-5.66*** (0.000) |
-2.12** (0.007) |
Random Effect |
-6.65*** (0.000) |
-2.70*** (0.002) |
-2.18*** (0.010) |
-0.84** (0.132) |
-4.43*** (0.000) |
-1.81** (0.037) |
-5.05*** (0.000) |
-2.11** (0.007) |
Source:
The computations are the author’s own. Note: *** and ** represent significance at 1%
and 5% level, respectively. |
The performance of the BiMARU states during this
period has been less than satisfactory. While Bihar’s PCSGDP rose 3.61 times in
2019–2020, it finished only one rank above its 1980–1981 rank of 25. Madhya
Pradesh slipped two ranks from its 1980–1981 position of 20 and ended at the rank
of 22 in 2019–2020, with its PCSGDP rising 3.49 times. Rajasthan performed
relatively better and jumped six ranks from the 1980s 25th to 19th position in
2019–2020, with 4.51 times rise in PCSGDP. Uttar Pradesh showed no change in
its ranking in 1980–1981 and 2019–2020 at the 24th position.
Stationarity
Check
The study considers five periods to check for
stationarity: i) the period 1980–2020 (the whole period); ii) the
pre-liberalization period of 1980–1992; iii) the post-liberalization period of 1992–2020
– which was further divided into two sub-periods in between, iv) from 1992 to
2003 and v) from 2004 to 2020. Was applied to test for stationarity for each of
the periods. Table 4 shows the results of the Levin–Lin–Chu panel unit root
test for all the sub periods taken together. The result shows no evidence of
convergence over the whole time period of 1980–2020 with all the states taken
together (Table 4). When the post-liberalization period as a whole is taken,
from 1992 to 2020, it again showed no evidence of convergence. Table 4 shows
the regional-level convergence for the four-time periods. The full-time period
exhibits no stationarity and hence no discernible evidence of convergence among
the regions. Between 1980 and 1992, only six states, Manipur, Nagaland, Orissa,
Punjab, Tripura, and West Bengal have shown stationarity and hence signs of
convergence. During 1992–2020, again no region exhibited stationarity. Between
1992 and 2003 only three states (Bihar, Karnataka, and Orissa) exhibited stationarity
while during 2004– 2020 five states (Jammu and Kashmir, Kerala, Maharashtra,
Tamil Nadu, and West Bengal) showed significant stationarity. We further
investigate the analysis as mentioned below in (Table 5).
a) The results of the panel unit root test on the
dataset were analysed by dividing the states into high-income, middle-income,
and low-income using the standards followed by World Bank. Results indicate
that for all the regions, and for the entire period the null hypothesis that
not be rejected as that series contains a unit root. In other words, there is
no such evidence of the mean reversion in the entire pool of states and the
time period of four decades. Both tests confirm the same fact. This leads us to
conclude that there is no resilient feature of convergence happening among the
regions in this period and takes us to investigate if the same is true for the
different groups of regions as designed by various income levels. We fail to
reject the null hypothesis of the presence of unit roots in all the groups and
indicate to state that even among the groups there is clearly no sign of any
convergence taking place.
b) The time period 1980 to 1992
(Pre-liberalization): The exercise for
all regions as well as for the groups was conducted. Interestingly, we can
reject the null hypothesis of the presence of unit roots across all regions.
Except for one case (IPS—Middle Income Group), we find stationarity. And if we
take stationarity as a proxy for convergence, then, in this case, we must admit
that for the mentioned period we see some presence of convergence. This is true
for all three groups indicating that not only overall convergence was taking
place but also among the group the process was active. So, one can argue that
in the last decade or so just before the liberalization there is some evidence
of convergence.
c) The Post- liberalization period: There is no such
evidence of convergence anymore in the data across all states. We fail to
reject the null hypothesis—the existence of a unit root in the series. However,
when we look into the groups, we have mixed results in terms of test results.
LLC test rejects the null hypothesis while the IPS test fails to reject the
null hypothesis. Surprisingly the pattern is the same for all three groups. Unless
both tests give the result in favor of rejecting the null hypothesis, we did
not conclude in favor of convergence. Thus, for the entire period of 1992-1993
to 2019-20, we do not find any confirmation of convergence, both across regions
as well as for each group. In order to further explore convergence, we divide
the post-liberalization era into two halves; primarily 1992-1993 to 2002-2003
and 2003-2004 to 2019-20. There is a lack of convergence for all regions for
both subperiod. Results of the designated tests indicate the presence of unit
root in the panel and thus fail to reject the null hypothesis in both cases.
Only for high-income regions for the time period 2002-2003 to 2019-20, we can
reject the null hypothesis of the presence of unit root for both tests. Using
our benchmark, we reinstate that there is some evidence of convergence among
the rich regions. For all the other cases, we find support only from the LLC
test in terms of rejecting the null hypothesis. The IPS panel in all cases
fails to reject the null hypothesis. Thus, our conclusion from these results
indicates a lack of convergence among these regions over the mentioned period.
d) To summarize, we find that post-liberalization the
high-income regions are in some alignment with convergence for the time period
2002-2003 and 2019-20. The middle-income and low-income regions do not reflect
any significant evidence of mean reversion post-liberalization. Thus, the
bigger picture of the table shows some convergence for all regions and among
each group for the period before liberalization (1980–1992). However, the
post-liberalization period is primarily dominated by a lack of convergence
among all regions and groups. Fair evidence of some convergence is only
available for high-income states for the subperiod 2002/2003–2013/2014.
4. Analysis with Control Variables: Control variables were included in the
analysis as mentioned in the literature to have a better understanding of the
nature of convergence or the lack of it [20,21]. The control variables included
for consideration in this study are;
1.
CAPEX—Capital Expenditure
of States and Union Territories as a percent of state GDP
2.
FISCAL DEFICIT—State
Fiscal Deficit as a percent of state GDP
3.
DEVEXTOT—Total
Development Expenditure, comprising of expenditure on revenue and capital
accounts and loans and advances for social and economic development as a
percent of state GDP.
We can reject the null hypothesis of the presence of a
unit root except for two occasions. Thus, in a nutshell, it can be said that
after controlling for CAPEX, DEVEXTOT, and FISCAL DEFICIT (Table 6) and Capex,
Education, Health, Welfare, and Fiscal Deficit (Table 7) there exists
discernible evidence for the existence of convergence. This result, when
combined with the results from the model without control variables, gives us a
clearer picture of the root cause. Results indicate that when relevant control
variables are added, there is evidence of convergence among regions of India.
This apparently looks like a contradicting result to the earlier tests for
absolute convergence, but the underlying reason can be traced to the very
source of this problem. It points toward the fact that the nature of inequality
across regions is not structural in nature and can be reduced through efficient
policy decisions. Increased and efficient spending toward education, health,
and welfare along with capital expenditure and addressing the state-level
fiscal deficit can successfully reduce the gap between regions. However, given
the sheer magnitude of the problem in a country the size of India, achieving
the said efficiency is difficult, especially in a democracy. In most cases,
Hausman tests indicate a fixed effect. However, for our data, the results are
the same in both cases. It is a general practice to report results for both
types of estimation. Also, there is significant literature that criticizes the
Hausman test. Thus, both results are reported. Tables 6 and 7 above summarize
the estimation results for both fixed effect and random effect LLC tests after controlling
for the exogenous variables.
Eradication of poverty and inequality has been the
prime objective of each government ever since independence during the last
seventy-five years. There is a pressing need to evaluate how successful the
country has been in this endeavor. Yet, India remains the home of one-third of
the World’s poor. This study examined the performance of Indian regions in
terms of their per capita GDP and whether there is any evidence of growth
convergence over the years among regions from 1980 to 2020. Since independence,
the performance of several regions has consistently been below the national
average. The economic policies of the government were aimed at pulling these
regions and their people out of poverty. But the question of whether and how
far these policies have been effective has remained a concern. The inter-region
disparity in per capita income has been an ever-present feature in the Indian
economy. Although the regional disparity cannot be completely eliminated,
policies should work toward narrowing down the inter-regional differentials in
PCSGDP, leading to a reduction in the disparity. In the absence of relevant
control variables, findings from panel unit root tests suggest no evidence of
convergence over the whole-time period of 1980–2020 when all the regions are
taken together. Interestingly, the pre-liberalization period of 1980–1992
showed more evidence of convergence among the regions as compared to the
post-liberalization period of 1992–2020 which does not exhibit any convergence.
After dividing the post-liberalization period into subperiods of 1992–2003 and
2004–2020 some sign of convergence is found; however, the number of regions
exhibiting any significant convergence is less in number than such regions in
the pre-liberalization period (three states in 1992–2003 and five states in
2004–2020 vs six states in 1980–1992). Furthermore, when control variables for
capital expenditure, development expenditure, and fiscal deficit are taken into
account, we find significant evidence for convergence of state-level per capita
GDP. This finding corroborates the findings and points toward the fact that the nature of
inequality across regions is not structural and can be reduced through
state-specific policy reforms and effective policy executions, such as
increased and efficient expenditure on capital and development while addressing
the state-level fiscal deficit. Recent evidence corroborates the rising
inequality in India. Provided empirical evidence that between 1980 and 2014 top
1 percent earners captured 22 percent of the total wealth and the top 0.1
percent received a 12 percent share of total growth compared to the 11 percent
share accrued to the bottom 50 percent. The results are indeed, worrying. It
does not provide evidence of the fact that there has been any convergence in
the states’ PCSGDP. In fact, after liberalization, the signature of convergence
among regions is weaker than before liberalization. The clustering of some
high-income regions raises a question of whether the benefits of liberalization
have been reaped only by a handful of regions while the majority of the regions
saw little or no change in their relative position. Also, when we trace
convergence after controlling for variables, such as capital expenditure,
fiscal deficit, and development expenditure—it opens the door for more active
policy interventions both at the Centre and State levels. It is the need of the
hour to have a closer and more critical look at the policies aimed at poverty
eradication, and removal of regional disparity and treat these issues as a
priority. With rapid urbanization and population increase, the economic
disparity will only grow, and if left unchecked, may lead to catastrophic
possibilities. A series of reforms undertaken by the current government in the
recent past, however, if implemented right, can nudge the economy in the right
direction. The long-term impact of those policies on growth and convergence
dynamics remains an important phenomenon to examine in the near future.