Long-run Association Between Bank Credit and Output: A Study on Districts’ Panel of West Bengal,India

Abstract

In a globalized world, the financial sectors and the real sectors are interlinked. Although it is a common phenomenon to a developed economy in its national as well as provincial levels, it has hardly been tested for the low-income countries like India. It is further difficult to have such linkage effects at the provinces and district levels. This article aims to examine whether per capita commercial bank credit and per capita net district domestic product for the districts of West Bengal state in India have long-run associations for the period 1993–2014 in a panel data framework. Using the panel cointegration and Vector error correction mechanism (VECM) technique, the study reveals that both the financial and real sector indicators are cointegrated and the short-run errors are corrected significantly to establish that there is bilateral causality between credit and output in both long run and short run.

Keywords

Per capita credit per capita net district domestic product panel unit roots panel cointegration VECM causality

Introduction and Background

There have been long debates between two groups of economists with regard to whether financial sectors’ development do have any impact upon the real sectors of the economies. The group led by Adam Smith did not believe that the financial institutions have any influence upon the production activities of the real sectors and so growth of a nation. Smith, in his famous book, ‘Wealth of Nation’ has pointed out that farmers, producers and businessmen are important carriers of economic growth. Another group led by Schumpeter (1911) offered the reverse argument. According to him, economies make progress through the trade cycle in a dynamic and discontinuous system. In order to break the circular flow, the workings of the innovative entrepreneurs are to be financed by banking funds.

After a long period of debates, Patrick (1966) is probably the first to clearly define the one-to-one correspondence between bank credit (an important indicator of financial sector) and growth of aggregate output, especially for the underdeveloped countries. To him, there are two ways of explaining the correspondences among the bank credit and growth of domestic product—supply leading approach (SLA) and demand following approach (DFA). Under SLA, it is argued that expansion of credit may boost up growth of domestic product of a country. On the other hand, under DFA, it is assumed that the financial development of a country is an outcome of the development of the real sector. Over time, there have been empirical verifications of the two opposite stands on whether financial and real sectors are interlinked. The studies so far available in the literature cover the so-called developed and developing economies of today. There are hardly studies that deal with the issue at sub-provincial, even at sub-national levels of the economies. This article tries to penetrate into this dry area with the specific aims.

The entire study has been arranged as review of literature, objectives, rationale of the study, methodology, data source, theory and empirical model, analysis of results and conclusion.

Survey of Relevant Literature

Over time, there have been empirical verifications of the two opposite stands on whether financial and real sectors are associated in short- and long-run framework. First, we address the literatures that deal with the positive impacts of financial sector in general, and the banking sector in particular on the economic growth of a country. Diamond (1984) has shown the positive role of bank finance on the allocation of capital by reducing the cost of capital by the process of delegated monitoring of the financial intermediary. In another study, Greenwood and Jovanovic (1990) have explained the role of the finance on economic growth by the help of the endogenous growth theory. According to the study, the financial sector collects and analyses the risk attached to the credit delivery and thereby allocates the financial capital at minimum costs, thus helping the economic growth of a nation. The study of King and Levine (1993) has supported the Schumpeterian version of a positive and significant correlation between the financial sector and economic growth. Based on the study of 80 countries over the period 1960–1989, it has established that various financial indicators like the ratio of credit allocated to private firms to GDP and the size of formal financial intermediation relative to GDP have a strong and robust correlation with economic growth and rate of physical capital accumulation. Demetriades and Hussein (1996) have shown that there is a bilateral relationship between financial development and economic growth for a set of 16 less developed countries. Beck et al. (2000) also highlight a positive correlation between development of financial intermediation and economic growth, rate of total factor productivity, savings rate and rate of physical capital accumulation. In another study, Kiran et al. (2009) have shown that a long-run relationship exists between financial development and economic growth for a panel of 10 emerging countries including India for the period 1968–2007. In their discussion paper on the Indian economy, Bhanumurthy and Singh (2009) have shown that the high growth of Indian GDP has been due to financial inclusion, among others. They observe that there has been cointegrated relation between credit-deposit ratio and State Domestic Product ratio. For a study at an Indian district level, Rajesh Raj et al. (2014) examined whether differences in banking sector penetration explain the differences in firm start-ups in informal sector of the country for the period from 1994–1995 to 2010–2011. The results confirm that local bank availability is associated with significant increase in enterprises in the informal sector, and the effect is more pronounced for larger enterprises in the sector. In a study specific to Indian states, Sehrawat and Giri (2015) examined the impact of financial development on growth of the 28 states for the period 1993–2012. It revealed that there was causality from per capita credit as well as per capita deposits to economic growth. In a district level study for Tamil Nadu in India, Nishanth and Baby (2016) attempted to analyse the relationships between bank credit and economic growth for the panel of 32 districts for the period from 2004–2005 to 2011–2012. The study observed that the elasticity of credit to growth and the reverse are all non-zero, justifying the existence of relationships between the two variables. Best et al. (2017) opined that financial deepening in Jamaica has positive effects upon the growth rate of income of the country. In another study related to India, Sehrawat and Giri (2017) confirm the long-run relationship between financial development index, trade openness and economic growth, and the causality test shows that there is a unidirectional causality running from financial development to economic growth.

We now address studies that put forward reverse arguments on the association between these two components. Lucas (1988) did not find any association between economic growth and finance, and he termed the relationship between financial and economic development as ‘over-stressed’. Similar conclusions are observed in Demetriades and Luintel (1996), Sarkar (2009). Driscoll (2004) tested whether changes in bank loan supply affect output for the panel of state-level data and found that shocks to money demand had large and statistically significant effects on the supply of bank loans, but loans had small, often negative and statistically insignificant effects on output. Recent researches have suggested that the level of financial development is good only up to a point, after which it becomes a haul on growth. This means that the relationship between finance and growth is a non-linear one or, more specifically, an inverted U-shape. In their study, Cecchetti and Kharroubi (2013) find that for private sector credit extended by banks, the turning point is close to 90 per cent of GDP. They also find that the faster the financial sector grows, the slower the economy as a whole grows. This finding indicates that big and fast-growing financial sectors may be very costly for the rest of the economy in the sense that more financial sector’s development exploits the existing resources of the economy. The work of Law and Singh (2014) provides new evidence on a sample of 87 developed and developing countries that there is a threshold effect in the finance–growth relationship in the way that up to a threshold point, the level of financial development is beneficial to growth, beyond which further development of financial sector tends to adversely affect growth. In their working paper on finance and economic growth in OECD and G20 countries, Cournède and Denk (2015) establish that finance has been a key determinant of long-term economic growth in countries over the past half-century, but they indicate that at current levels of household and business credit, further expansion slows rather than boosts growth. In a recent study on whether a country’s level of economic development impacts its finance growth relationship for a panel of 90 countries from three different classes of economies, Nguyena et al. (2017) observed that the banking sector has a negative effect for all levels of development but more so for developed economies.

With respect to the study in the head of the link between bank credit and output, at least for India, we do not find such study to follow. However, there are a bit different but related studies. The study by Kumar (2008) explores the relationship between technical efficiency (TE) and profitability in the Indian public sector banking industry using data envelopment analysis. It reveals that the mean level of TE for the industry is found to be 88.5 per cent, which implies that public sector banks can produce 1.13 times as much output from the same inputs, if they operate at the ‘efficiency frontier’. In a recent study, Rakshit (2019) evaluated the performance efficiency of the top 36 commercial banks in India and revealed that in India where large banks perform well on profitability, small banks have done well on marketability efficiency. However, close to 80 per cent of Indian commercial banks are inefficient on both the fronts of profitability and marketability.

The survey of literature briefs the finance–growth linkages at country levels in most cases without having the studies at the district levels. Although one study is available (Nishanth & Baby, 2016) at the district level of Tamil Nadu state in India, it did not focus on examining the long-run associations between bank credit and output. The present study tries to fill the gap in the literature by analysing the short-run and long-run relationships between commercial bank credit and net real domestic products of the panel of districts of West Bengal in India for the period 1993–2014.

Objective and Structure of the Study

The present study aims to examine whether per capita commercial bank and per capita net district domestic product (NDDP) for the districts of the state of West Bengal in India have long-run associations for the period 1993–2014 in a panel data framework.

Rationale for the Study

The increasing socio-economic disparities across the countries, across the regions and across the provincial levels after the phase when almost all the countries have entered into the globalized world have compelled world leaders to think of holistic policy combinations to reduce the magnitude of disparities. It has started with the financial sector reform in the early 1990s to accelerate credit facilities to different sectors of the economy and presently with new agenda, Jan Dhan Yojana, to accelerate the process of financial inclusion. Thus, it is a time to examine whether the financial inclusion mission has been associated with the real economy in a long-run phenomenon. Although there are some studies for India and its states, no study in this regard was for West Bengal state as it is a whole independent economy, and it needs some policy prescriptions in the same issues like that of India and any other country in the world. Hence, the study undertakes the exercise for West Bengal as a notion for micro-level study. The present study, thus, has rationale to attempt for such exercise.

Theoretical Model

To establish the link between output per capita (y) and bank credit per capita (c) (assuming that all the physical capitals are purchased through loans from banking and financial institutions) by means of a theoretical model, suppose Y is the real NDDP, L is total labour force (or population), C is the total volume of bank credit and A is all other factors of production. So, the production function is given as follows:

Y = f (A, L, C)

Taking the intensive form of the function, we have y as the per capita real NDDP of the districts, c as the per capita bank credit and a as the per capita value of the other factors and the reduced production function is as follows:

y = f (a, c)

with dy/da > 0 and dy/dc > 0. Taking the changes of all the three indicators, y, a and c, over time and dividing both sides by y, we get a relation as follows:

\begin{array}{l} (d y / d t) / y = f ((d c / d t) / c) + f ((d a / d t) / a) \\ or, \dot{y} / y = f (\dot{c} / c) + f (\dot{a} / a) \end{array}

where $\dot{y}$ , $\dot{c}$ and $\dot{a}$ indicate rate of growth of per capita GDP and credit. This shows that the rate of growth of per capita NDDP is the summation of the growth rate of per capita credit and growth of other factors per capita.

Taking a Cobb–Douglas form, the function is considered as follows:

y = a^{(1 - α)} c^{α}

where 0 < α < 1. We get a one-to-one correspondence between y and c. Hence, we have the following modified relation:

\dot{y} = (1 - α) \dot{a} + α \dot{c}

This shows that growth rate of NDDP over time is the weighted average of the growth rates of per capita credit and other factors per capita. If α = 0.5, then the above relation is reduced to

\dot{y} = 1 / 2 (\dot{a} + \dot{c})

which means the growth rate of NDDP over time is the half of the summation of the growth rates of two factors over time.

Let us come to the empirical methodology. As mentioned earlier, existence of long-run association between credit and output of the panel of the districts is a necessary precondition to have a balanced growth and development of the state of West Bengal via the financial inclusion mission.

Data, Theory and Empirical Methodology

We have borrowed the district-level data on credit of the scheduled commercial banks in India from the database of Reserve Bank of India for the period 1980–2014. There are 18 undivided districts in our study—Bankura, Bardhaman, Birbhum, Calcutta, Dakshin Dinajpur, Howrah, Hooghly, Jalpaiguri, Koch Bihar, Malda, Medinipur, Murshidabad, Nadia, North 24 Parganas, Purulia, South 24 Parganas and Uttar Dinajpur. The data on NDDP at constant 1993–1994 prices have been borrowed from Bureau of Applied Economics and Statistics (BAES) Division, Government of West Bengal. We have thus a balanced panel of the number 35 × 18 = 630. To standardize the data on credit and NDDP, we have converted the total credit and NDDP data into per capita credit for the districts by dividing the population data derived from the average growth rates of population for the available periods.

Panel Unit Roots

If there is time series data across the number of cross-sections, then running individual unit root tests suffer from power problem that would lead to spurious regression results. A panel unit root test overcomes this problem and provides results with more power. The present study has, therefore, concentrated on the panel data analysis.

For a data set of a variable y (y_i,t, i = 1, 2, …, N (here N = 8) and t = 1, 2, ..., T (here, T = 26), where t denotes time, let us consider the following linear regression model for panel unit root test in line with the ADF(p) (1979) regression, viz.:

Δ y_{i,}_{t} = (ρ_{i} - 1) y_{i,}_{t - 1} + \sum_{j = 1}^{p} γ_{j} Δ y_{i,}_{t - j} + {Z^{'}}_{i, t} α_{i} + u_{i, t}

(1)

where Z_it represents the exogenous variables in the model, including any fixed effects or individual trends. The null hypothesis for this model is ρ_i = 1 against the alternative hypothesis ρ_i < 1. Equation (1) can be rewritten as

Δ y_{i,}_{t} = β_{i} y_{i,}_{t - 1} + \sum_{j = 1}^{p} γ_{j} Δ y_{i,}_{t - j} + {Z^{'}}_{i, t} α_{i} + u_{i, t}

(2)

The null hypothesis for this model is β_i = 0 against the alternative hypothesis β_i < 0.

There are two approaches of testing panel unit roots, depending on homogeneity or heterogeneity of the regression coefficients. Testing techniques for panel unit roots where the coefficients (β_is) are restricted to be homogeneous across all units of the panel have been offered by Levin and Lin (1993) and Levin et al. (2002), and for the heterogeneous coefficients have been by Im et al. (1997, 2003), ADF—Fisher chi-square and PP—Fisher chi-square of Maddala and Wu (1999) and Choi (2001). The assumption of homogeneity (β_is = β, say) is clearly restrictive and subject to the possible homogeneity bias of the fixed effect estimator.

The Levin and Lin (1993) and Levin et al. (2002) models are captured by the following Equation (3), where β_is = β:

Δ y_{i,}_{t} = β y_{i,}_{t - 1} + \sum_{j = 1}^{p} γ_{j} Δ y_{i,}_{t - j} + {Z^{'}}_{i, t} α + u_{i, t}

(3)

The test statistics proposed by Maddala and Wu (1999), based on the suggestion of Fischer, is of the form

χ^{2} = - 2 \sum_{i = 1}^{N} (log p_{i})

(4)

(where i = 1, 2, …, N). It follows chi-square distribution under the null hypothesis of p_i = 0 for all the is. The simulation suggests that the Maddala and Wu’s Fisher test is more powerful than the Im, Pesaran and Shin test, which is again more powerful than the Levin, Lin and Chu test in a variety of situations.

Panel Cointegration Test

In a panel data, two tests are usually done for testing whether there are cointegrating relationships between/among the variables. The Pedroni (1999, 2004) and Kao (1999) tests are based on Engle and Granger (1987) two-step residual-based cointegration tests, and Fisher test is a combined Johansen test. We apply all the three tests in our study. The methodologies are as follows.

The Engle and Granger (1987) cointegration test is based on an examination of the residuals of a spurious regression performed using I(1) variables. If the variables are cointegrated, then the residuals capturing the linear combinations of both the variables should be I(0) or first differenced stationary. On the other hand, if the variables are not cointegrated, then the residuals will be I(1). Pedroni proposes several tests for cointegration that allow for heterogeneous intercepts and trend coefficients across cross-sections. Let us consider the following regression with no intercept constant and trends:

y_{i, t} = β_{i} x_{i, t} + u_{i, t}

(5)

For t = 1, 2, …, T and i = 1, 2, …, N; where y and x are assumed to be integrated of order one. Under the null hypothesis of no cointegration, the estimated residuals e_i,t will be I(1). The general approach is to obtain residuals from Equation (5) and then to test whether residuals are I(1) by running the auxiliary regression for each cross-section as

e_{i, t} = ρ_{i} e_{i, t - 1} + \sum_{j = 1}^{p_{i}} γ_{i j} Δ e_{i, t - j} + ε_{i, t}

(6)

Pedroni describes various methods of constructing statistics for testing for null hypothesis of no cointegration (ρ_i = 1). There are two alternative hypotheses: the homogenous alternative (ρ_i = ρ) < 1for all i (which Pedroni terms the within-dimension test or panel statistics test), and the heterogeneous alternative (ρ_i < 1) for all i (also referred to as the between-dimension or group statistics test). The Pedroni panel cointegration statistic $ℵ_{N, T}$ is constructed from the residuals of Equation (6). A total of 11 statistics with varying degrees of properties (size and power for different N and T are generated.

On the other hand, Kao (1999) presents two tests for the null hypothesis of no cointegration in panel data: the DF and ADF type tests. He considers the special case where cointegration vectors are homogeneous between individuals. Kao presents two sets of specifications for the DF test statistics. The first set of test statistics depends directly on consistent estimation of long-run parameters. The second set of test statistics does not. Kao considers the model as depicted in Equations (5) and (6). The ordinary least squares (OLS) estimate of ρ is given by:

ρ = \frac{\sum_{i = 1}^{N} \sum_{t = 2}^{N} {\hat{e}}_{i t} {\hat{e}}_{i t - 1}}{\sum_{i = 1}^{N} \sum_{t = 2}^{N} {\hat{e}}_{i t} {\hat{e}}^{2}_{i t - 1}}

(7)

The ADF test statistic for the null hypothesis of no cointegration is based on the above expression.

In a different testing configuration, Johansen (1988) proposes two different statistics, one of them is the likelihood ratio trace statistics, and the other one is maximum eigenvalue statistics, to determine the presence of cointegration vectors in non-stationary time series. Using Johansen test for cointegration, Maddala and Wu (1999) consider Fisher’s (1932) suggestion to combine individuals tests, to propose an alternative to the two previous tests, for testing for cointegration in the full panel by combining individual cross-section tests for cointegration.

Vector Error Correction Mechanism

Once we establish that the series are cointegrated or there are long-run equilibrium relations between the variables, we need to test whether the errors in the short-run deviations from the equilibrium relations are corrected and the series converge to the long-run relation. Vector error correction mechanism (VECM) captures this phenomenon. VECM is a restricted Vector Autoregressive (VAR) model, which is intended for using with cointegrated non stationary series. The VECM has cointegration relations built into the specification so that it restricts the long-run behaviour of the endogenous variables to converge to their cointegrating equilibrium relationships while allowing for short-run adjustment dynamics. The cointegration term is known as the error correction term since the deviation from long-run equilibrium is corrected gradually through a series of partial short-run adjustments.

To present the VECM, let us consider a two-variable system with one cointegrating equation therein and no lagged difference terms. The cointegrating equation for no intercept and trend is given by the following equation:

y_{t} = β x_{t}

(8)

The estimated error term in first difference is given as

e_{t - 1} = y_{t - 1} - β x_{t - 1}

(9)

Therefore, the corresponding VEC model is:

\begin{array}{l} Δ y_{t} = α_{y} (y_{t - 1} - β x_{t - 1}) + ε_{y} \\ Δ x_{t} = α_{x} (x_{t - 1} - β y_{t - 1}) + ε_{x} \end{array}

(10)

In the simple model, the only right-hand side variable is the error correction (EC) term, which is zero in the long-run equilibrium. However, if y and x deviate from the long-run equilibrium, the error correction term will be non-zero and each variable adjusts to partially restore the equilibrium relation. The coefficient α measures the speed of adjustment of the ith endogenous variable towards the equilibrium. If the error correction term is found to be negative in sign and statistically significant (with probability less than 0.05), then we say that the short-term errors have been corrected, and the series are back to the long-run relation. Additionally, we say that there is long-run causality from y to x or vice versa.

Finally, the short-run causality can be tested in this VECM set-up by applying Wald test for coefficient diagnosis.

Analysis of Results and Discussion

Before going into the proposed empirical exercises, we present the panel data in diagrams to have a look at the possible associations between PCCredit and PCNDDP for the districts. Figure 1 presents the same for all the districts except Calcutta and Figure 2 for Calcutta only to have better idea about the trends of the panels.

Figure 1.

Source: Sketched by the authors on the basis of RBI and BAES data.

Figure 2.

Source: Sketched by the authors on the basis of RBI and BAES data.

It is observed that over time, both PCCredit and PCNDDP are increasing for the districts and depicting similar types of movements of both the series. The series for PCCredit lies below the series for PCNDDP but with exception for Calcutta. For Calcutta, the positions of the series are reverse in the way that PCCredit lies above PCNDDP, since most of the state head offices of all the commercial banks are situated in Calcutta. Again, the trend for the PCNDDP lying below PCCredit is due to the fact that giant share of credits is cleared from state head offices but used in the districts other than Calcutta.

Graphical presentation cannot ensure the association between credit and output; we thus move to the next quantitative exercises.

Panel Unit Roots Test Results

We have run panel unit root tests for common unit root process (LLC) and individual unit root process (IPS, MW) following Equations (1)–(4) for the series on PCCredit and PCNDDP of individual districts for the study period of over 396 panel observations. The results of panel unit roots test for both the series have been presented in Table 1 using Eviews 7.

The results given in Table 1 show that both the series in their first differences are highly significant under ‘common unit processes’ and ‘individual unit root processes’. This means that the null hypothesis of ‘presence of panel unit roots’ is rejected, and hence, the series are stationary at their first differences. Further, it is also observed that the results of stationarity hold under ‘with’ and ‘without’ intercept conditions. Hence, we move to the next step.

Panel Cointegration Test Results

Since the primary criteria to run tests for cointegration, that is both the series are to be integrated of order one, have been met, we now test for the existence of long-run associations between credit and output for the selected districts. As mentioned in the methodology, we run cointegration test under three lines—Pedroni test, Kao test and Johansen–Fisher test.

Table 1.

Panel Unit Root Test Results for PCCredit and PCNDDP at Their First Differences

Method	Null Hypothesis	Test Statistics with Intercept (Prob.)		Test Statistics without Intercept (Prob.)
		PCCredit	PCNDDP	PCCredit	PCNDDP
Levin, Lin and Chu	Unit root (under common unit root process)	−4.81 (0.00)	−11.64 (0.00)	−7.92 (0.00)	−17.54 (0.00)
Im, Pesaran and Shin	Unit root (under individual unit root process)	−4.46 (0.00)	−10.93 (0.00)	—	—
MW–ADF–Fisher chi-square	Unit root (under individual unit root process)	102.86 (0.00)	188.86 (0.00)	147.16 (0.00)	304.15 (0.00)
MW–PP–Fisher chi-square	Unit root (under individual unit root process)	166.76 (0.00)	232.35 (0.00)	216.26 (0.00)	342.06 (0.00)

Source: The authors.

Note: Automatic lag-length selection is based on AIC: 0 to 6. Probabilities for Fisher tests are computed using an asymptotic chi-square distribution. All other tests assume asymptotic normality.

The Pedroni (Engle–Granger-based) test has been done in three alternative criteria: no deterministic trend, deterministic intercept and trend, and no deterministic intercept and trend. For the ‘within-dimension’ criteria, there are eight panel results for simple and weighted statistic and for the ‘between-dimension’ criteria, there are three group statistics. The results are given in Table 2.

It is observed from the Pedroni test results that the existence of cointegration between the series are maximum observed under the criterion of deterministic intercept and trend. Out of total 11 statistics under this criterion, 6 show statistically significant results, which support for the existence of cointegration. But the other two criteria do not prove the existence of cointegration in gross sense.

Another Engle–Granger-based test results, the Kao residual cointegration test, are presented in Table 3, which proves the unambiguous conclusion that the series are cointegrated and have long-run associations. The ADF test statistic is too high to reject the null hypothesis of ‘no cointegration’.

Table 2.

Pedroni Residual Panel Cointegration Test

	Hypotheses
Test Criteria	Null Hypothesis: No Cointegration		Statistic (Prob.)	Weighted Statistic (Prob.)
No deterministic trend	Alternative hypothesis: common AR coefficients (within-dimension)	Panel v-statistic	−2.58 (0.99)	−2.45 (0.99)
		Panel rho-statistic	4.13 (1.00)	3.88 (0.99)
		Panel PP-statistic	6.49 (1.00)	5.87 (1.00)
		Panel ADF-statistic	−0.37 (0.35)	1.30 (0.90)
	Alternative hypothesis: individual AR coefficients (between-dimension)	Group rho-statistic	5.50 (1.00)	—
		Group PP-statistic	8.45 (1.00)	—
		Group ADF-statistic	1.49 (0.93)	—
Deterministic intercept and trend	Alternative hypothesis: common AR coefficients (within-dimension)	Panel v-statistic	1.60 (0.05)	1.63 (0.05)
		Panel rho-statistic	−1.12 (0.13)	1.06 (0.86)
		Panel PP-statistic	−2.61(0.00)	−0.12 (0.45)
		Panel ADF-statistic	−2.30 (0.01)	−1.71 (0.04)
	Alternative hypothesis: individual AR coefficients (between-dimension)	Group rho-statistic	2.25 (0.98)	—
		Group PP-statistic	0.81 (0.79)	—
		Group ADF-statistic	−1.79 (0.03)	—
No deterministic intercept and trend	Alternative hypothesis: common AR coefficients (within-dimension)	Panel v-statistic	−0.31(0.62)	−0.42 (0.66)
		Panel rho-statistic	2.22 (0.98)	2.24 (0.98)
		Panel PP-statistic	3.23 (0.99)	3.27 (0.99)
		Panel ADF-statistic	−3.40 (0.00)	−1.04 (0.14)
	Alternative hypothesis: individual AR coefficients (between-dimension)	Group rho-statistic	5.47 (1.00)	—
		Group PP-statistic	6.38 (1.00)	—
		Group ADF-statistic	−1.81 (0.03)	—

Source: The authors.

Note: The bold figures indicate significant cointegration results.

Table 3.

Kao Residual Cointegration Test

Null Hypothesis: No Cointegration
	t-Statistic	Prob.
ADF	−5.625	0.001

Source: The authors.

Table 4.

Fisher Johansen Cointegration Test

Fisher Stat.		Fisher Stat.
(from trace test)	Prob.	(from max-eigen test)	Prob.
123.5	0.000	101.2	0.000
87.91	0.000	87.91	0.000

Source: The authors.

Finally, the Fisher–Johansen cointegration test result is presented in Table 4. The results show, on the basis of the estimated values of trace statistics and maximum eigenvalue, that the panels of PCGDP and PCCredit are cointegrated. The probability values are too low to reject the null hypothesis of no cointegration.

Combining the panel cointegration results of the three methods as depicted in Tables 2 –4, we can now conclude that the panels of the series on credits and outputs of the districts of West Bengal are cointegrated, and they have equilibrium relations in the long run in gross sense. Gross sense means two out of three test criteria unequivocally prove the existence of cointegration. There is one cointegrating equation.

Vector Error Cointegration Mechanism Test Results

Since we have proved that there is long-run association between PCCredit and PCNDDP, in the next step, we need to test whether the errors in the short-run dynamics are corrected, and the series converge to the long-run relation. As mentioned earlier, VECM captures this phenomenon. By using the set of Equations (8)–(10), we have derived the estimated values of the error correction terms. Needless to mention that a negative and statistically significant error correction term justifies that the series converge to the equilibrium relation, and there is long-run causality from either of the exogenous variables to either of the endogenous variables. We have considered lags up to 2 as the system allows accommodating under optimal condition. The VECM results have been given in Table 5.

Table 5.

VECM Results

Dependent Variables	EC Terms	Probability	Whether Errors Corrected	Remarks
D(PCCREDIT)	C(1) = −0.059	0.002	Yes	Long run causality from PCNDDP to PCCREDIT
D(PCNDDP)	C(1) = −0.032	0.08	Yes	Long run causality from PCCREDIT to PCNDDP

Source: The authors.

Note: Optimum lag has been derived as 2.

Table 6.

Short-run Causality Through Wald Test

Dependent Variable	Chi-square	Probability	Remarks
D(PCCREDIT)	49.30	0.00	PCNDDP→PCCREDIT
D(PCNDDP)	14.25	0.00	PCCREDIT→PCNDDP

Source: The authors.

It is observed from Table 5 that the error correction terms (C(1)) for either of the two variables playing as the endogenous dependent variable are of desired negative sign and statistically significant. This means, short-run deviations from the equilibrium relations are brought back and the errors are corrected. Further, there is 5.9 per cent rate of correction annually for PCCredit as the dependent variable and 3.2 per cent rate of correction for PCNDDP as the dependent variable. In addition to that, we can infer that there is bidirectional causality in the long run.

Again, there may be short-run causality between the variables. We test it through the Wald test. Wald test results are confirmed by the values and corresponding probability in chi-square statistic. The results are given in Table 6.

We observe from Table 6 that the values of chi-square statistics are too high in both cases to reject the null hypothesis of ‘no panel causality’. This means there is bilateral causality between credit and output in the short run like that of the long run. Hence, we conclude that per capita credit and output of the panel of districts of West Bengal state in India have long-run associations with bilateral causal relations in both long-run and short-run periods. It further signifies that the financial sectors and the real sectors of the panel of districts of West Bengal are interlinked and both the supply leading and DFAs of inter linkage have worked well.

Thus, the study supports the proposition that financial institutions and real sectors of the state are interlinked like that of the Indian-specific studies of Bhanumurthy and Singh (2009), Rajesh Raj et al. (2014), Sehrawat and Giri (2015) and for the district-level study of Tamil Nadu by Nishanth and Baby (2016). Hence, the counter-proposition that financial development and real sector’s development have no linkage does not work for West Bengal.

Concluding Observations

Starting with the objective of examining whether credit and output of the districts of West Bengal in India have long-run associations in a panel data format, we conclude that the series for per capita credit and per capita net domestic product for the period 1993–2014 are with long-run associations. The short-term deviations are corrected significantly at the rate of 3.2 per cent to 5.9 per cent. Further, the VECM and Wald test results establish that credit and output of the districts have bidirectional interplays between them. We further infer that the financial sector and the real sectors of the state of West Bengal in India are interlinked and both the supply leading and DFAs have been working well for the districts as a whole.

Managerial Implications

As having long-run relationships between bank credits and outputs of the districts of West Bengal state in India, it is thus prescribed that the Government of West Bengal, India, should concentrate on developing the real and financial sectors by the help of the central government and the central bank of the country so that their interlinking effects will take the state to newer levels of growth and development.

Footnotes

Acknowledgement

The authors are grateful to the anonymous referees of the journal for their extremely useful suggestions to improve the quality of this article. Usual disclaimers apply.

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: On behalf of all authors, the corresponding author states that there is no conflict of interest.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Ramesh Chandra Das

References

Beck

, Levine

, & Loayza

(2000). Finance and the sources of growth. Journal of Financial Economics , 58, 261–300.

Best

, Francis

B. M.

, & Robinson

(2017). Financial deepening and economic growth in Jamaica. Global Business Review , 18(1), 1–18.

Bhanumurthy

N. R.

, & Singh

(2009). Understanding economic growth in Indian states: A discussion paper . Institute of Economic Growth.

Cecchetti

, & Kharroubi

(2013, September 27). Why does financial sector growth crowd out real economic growth ? Finance and the Wealth of Nations Workshop, Federal Reserve Bank of San Francisco & The Institute of New Economic Thinking.

Choi

(2001). Unit Root Tests for Panel Data. Journal of International Money and Finance , 20, 249–272.

Cournède

, & Denk

(2015, June). Finance and economic growth in OECD and G20 countries . OECD Economics Department Working Paper, No. 1223, ECO/WKP (2015)41.

Demetriades

P. O.

, & Luintel

K. B.

(1996). Financial development, economic growth and banking sector control, evidence from India. The Economic Journal , 106(435), 359–374.

Demetriades

P. O.

, & Hussein

(1996). Does financial development causes economic growth? Time series evidence from sixteen countries. Journal of Development Economics , 51(2), 387–411.

Diamond

(1984). Financial intermediation and delegated monitoring. Review of Economic Studies , 51(3), 393–414.

10.

Driscoll

(2004). Does bank lending affect output? Evidence from the U.S. states, Journal of Monetary Economics , 51(3), 451–471.

11.

Engle

R. F.

, & Granger

C. W.

(1987). Co-integration and error correction: Representation, estimation and testing. Econometrica , 55, 251–276.

12.

Fisher

R. A.

(1932). Statistical Methods for Research Workers . Edinburgh: Oliver and Boyd.

13.

Greenwood

, & Jovanovich

(1990). Financial development, growth and the distribution of income. Journal of Political Economy , 98(5, Part 1), 1076–1107.

14.

K. S.

, Pesaran

M. H.

, & Shin

(2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics , 115, 53–74.

15.

K.S.

, Pesaran

M. H.

, & Shin

(1997). Testing for Unit Roots in Heterogeneous Panels. Mimeo.

16.

Johansen

(1988). Statistical analysis of cointegration vectors. Journal of Economic Dynamics and Control , 12(2–3), 231–254.

17.

Kao

(1999). Spurious regression and residual-based tests for cointegration in panel data. Journal of Econometrics , 90, 1–44.

18.

King

R. G.

, & Levine

(1993). Finance and growth: Schumpeter might be right. Quarterly Journal of Economics , 108(3), 717–738.

19.

Kiran

, Yavus

N.C.

, & Guris

(2009). Financial development and economic growth: A panel data analysis of emerging countries. International Research journal of Finance and Economics , 30, 1450–2887.

20.

Kumar

(2008). An analysis of efficiency–profitability relationship in Indian public sector banks. Global Business Review , 9(1), 115–129.

21.

Law

, & Singh

(2014). Does too much finance harm economic growth. Journal of Banking & Finance , 41(C), 36–44.

22.

Levin

, Lin

C. F.

, & Chu

C. S. J.

(2002). Unit root tests in panel data: Asymptotic and finite-sample properties. Journal of Econometrics , 108, 1–24.

23.

Levin

, & Lin

C.F.

(1993). Unit Root Tests in Panel Data: New Results. UC San Diego Working Paper 93–56.

24.

Maddala

G. S.

, & Wu

(1999). A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics , 61, 631–652.

25.

Lucas

R. E.

(1988). On the mechanics of economic development. Journal of Monetary Economics , 22(1), 3–42.

26.

Nguyena

Y. N.

, Brown

, & Skully

(2017, January). Economic development levels and the finance and growth nexus. Paper presented at the 2014 Paris Financial Management Conference, Monash University, Australia.

27.

Nishanth

, & Baby

(2016, December). Determining optimal credit allocation at a district level. IFF Working Paper Series No. WP-2016–01.

28.

Patrick

H. T.

(1966). Financial development and economic growth in UDCs. Economic Development and Cultural Change , 14, 174–189.

29.

Pedroni

(1999). Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and Statistics , 61, 653–670.

30.

Pedroni

(2004). Panel cointegration: Asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econometric Theory , 20, 597–625.

31.

Sarkar

(2009). Does finance matter for growth? What the data show ? https://mpra.ub.uni-muenchen.de/32937/1/MPRA_paper_32937.pdf.

32.

Rajesh Raj

S. N.

, Sen

, & Kathuria

(2014). Does banking development matter for new firm creation in the informal sector? Evidence from India. Review of Development Finance , 4(1), 38–49.

33.

Rakshit

(2019). Evaluating profitability and marketability efficiency: A case of indian commercial banks. Global Business Review . https://doi.org/10.1177/0972150918822569.

34.

Schumpeter

(1911). The theory of economic development . Harvard University Press.

35.

Sehrawat

, & Giri

A. K.

(2015). The role of financial development in economic growth: Empirical evidence from Indian states. International Journal of Emerging Markets , 10(4), 765–780.

36.

Sehrawat

, & Giri

A. K.

(2017). Financial structure, interest rate, trade openness and growth: Time series analysis of Indian economy. Global Business Review , 18(5), 1278–1290.