The Export-induced Investment-led Growth Process in India

The macroeconomic (Keynesian) stability condition of the model economy will be fulfilled if the responsiveness of ‘σ’ to a unit change in capacity utilization rate is greater than that[r]

Working Paper

The Export-induced Investment-led Growth Process in India

Zico Dasgupta

Azim Premji University, Bangalore

I Introduction

It is widely known by now that the Indian economy embarked upon a new growth trajectory during the period of economic liberalization as it witnessed higher growth rate of output and investments as compared to the earlier period What remains to be a matter of contestation, however, is the explanation for such a phenomenon Depending on their theoretical foundations, the explanations that are provided in the recent literature can be broadly distinguished into two categories as follows

The first set of explanations can be located within the supply-side theoretical framework, the defining features of which are the presence of a production function and the existence of an adjustment mechanism where exante savings necessarily generates an equivalent level of investments under the assumption of perfect price flexibility India’s recent growth story has been explained within this framework by referring to the phenomenon of higher growth rate of total factor productivity (Bosworth et al, 2007; Basu and Maertens, 2007 and Robertson, 2012), which in turn, has been perceived as a logical corollary of successful implementation of trade liberalization policies since the decade of 90s (Panagariya, 2008; Bhagwati and Panagariya, 2013 and Panagariya, 2013) Accordingly, the recent trend in India’s output growth rate has been explained in terms of a binary between “triumph of liberalization” and obstacles arising for ‘reforms’, where episodes of high growth have been perceived as a logical corollary of successful implementation of liberalization policies while the explanation for episodes of economic slowdown has been sought in the claim that “reform was halted” (ibid) There are at least two criticisms which can be levelled at these explanations:

investment behavior is introduced by dropping the unrealistic assumption that all savings are necessarily invested In the absence of such an automatic adjustment process, the explanation for the rise in investments itself remain incomplete within this framework

The second limitation involves the proposition of a causal relationship between economic liberalization and output growth rate Not only the relationship between degree of trade openness and output growth rate itself has been found to be statistically insignificant in various cross-country regression analyses (Rodrik, 1999 and Rodriguez and Rodrik, 1999), but in itself, it also falls short of explaining one key aspect of India’s growth process While the formal inauguration of liberalization policies in India came about as early as 1991 with the implementation of New Economic Policies and Structural Adjustment Program, the output growth rate during the entire decade of the 90s remained roughly similar to that of the 80s (Chandrasekhar and Ghosh, 2002) and it was primarily due to the high growth rates achieved during the decade of the 2000s that the period of economic liberalization as a whole registered higher average output growth rate as compared to that of the 80s (Azad et al, 2018) This specificity of the decade of 2000s remains largely unaddressed within these explanations

In sharp contrast to the first set of explanations, the second set of explanations attempts to address both these limitations These explanations can be located within the demand-side framework, the defining features of which are the existence an investment function and the existence of an adjustment mechanism where investments necessarily generates an equivalent level of savings The central mechanism through which the phenomenon of higher output growth rate is explained in this framework is higher investments There are primarily two routes through which the phenomenon of higher investments and output in India during the decade of 2000s has been explained

While both these routes have pointed towards a domestic demand-led growth process this paper proposes a third route within the demand-side framework, which can be described as an export-induced investment led growth process Though the Indian economy witnessed a fall in net exports during the period of high growth, what we shall argue is that it was the exogenous rise in exports, brought about by changes in the level and pattern of global demand, which led to higher investments and output during the relevant period

The bulk of the empirical analysis in this paper is restricted till the year 2012-13 on account of comparability problem between India’s new GDP series (base 2004-05) and the older one1 The rest of the paper is divided in the following manner: section II outlines some stylized facts of India’s growth process during the recent period Section III discusses the theoretical framework through which an export-induced investment led growth process can be conceptualized Section IV empirically tests the investment function for the Indian economy which is laid out in the theoretical model Section V reinterprets India’s growth story during the liberalization period in the light of this analysis Section VI provides the conclusion

II Some Stylized Facts of India’s Growth Process during the Post-Liberalization Period

If the year of implementation of New Economic Policies in India (1991) is regarded as the most distinct dividing line between its pre-liberalization and post-liberalization period, then there are at least four distinct features of India’s growth process which can be noted during the period of economic liberalization as the following

(1) Rise in GDP Growth Rate: The period of economic liberalization as a whole registered a

higher average growth rate of GDP when compared to that of the 80s as shown in figure While the average growth rate during the decade of 90s remained more or less the same (5.8%) as compared to that of the 80s (5.6%) , the average growth rate of the 2000s (7.1%) was much higher than the earlier periods It is this higher economic growth witnessed during the 2000s which ensured that the average growth of GDP during the period of economic liberalization (1992-93 to 2012-13) was higher than that during the 80s

Figure 1: Average Decadal Growth Rate of GDP at Fi

Source: National Account Statistics, CSO, various years

(2) The Two Booms of 2000s:

evident from figure 2, was primarily on account of two booms: first, 2007-08 and second, a brief period between 2009

witnessed a discernible slowdown since 2011 reflected in the figure for the years 2011

Figure 2: GDP Growth Rate at Factor Cost and Constant Prices (as %), 1992 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 1980-81 to 1989-90 1990-91 to 1999-00 5.6 5.8 0.0 2.0 4.0 6.0 8.0 10.0 12.0 1992 -93 1993 -94 1994 -95 1995 -96 1996 -97 1997 -98 1998 -99 1999 -00

Figure 1: Average Decadal Growth Rate of GDP at Fixed Cost and Constant Prices (as %)

Source: National Account Statistics, CSO, various years

(2) The Two Booms of 2000s: The rise in the output growth rate during the decade of 2000s, as

evident from figure 2, was primarily on account of two booms: first, the period from 2003 08 and second, a brief period between 2009-10 and 2010-11 The Indian economy witnessed a discernible slowdown since 2011-12 (Subramanian, 2019), a trend which is also reflected in the figure for the years 2011-12 and 2012-13

GDP Growth Rate at Factor Cost and Constant Prices (as %), 1992 91 to 00 2000-01 to 2012-13 1992-93 to 2012-13 5.8 7.1 6.9 1999 -00 2000 -01 2001 -02 2002 -03 2003 -04 2004 -05 2005 -06 2006 -07 2007 -08 2008 -09 2009 -10 2010 -11 2011 -12 2012 -13

xed Cost and Constant Prices (as %)

The rise in the output growth rate during the decade of 2000s, as the period from 2003-04 to 11 The Indian economy 12 (Subramanian, 2019), a trend which is also

Source: National Account Statistics, CSO, various years

(3) Exogenous changes in Global Demand: The two booms in the Indian economy more or less

coincided with two changes in global economy during the same period, one involving the level of global demand and the other involving the pattern of global demand

Firstly, the decade of 2000s was marked by a sharp rise in the growth rate of aggregate import demand and merchandise import demand of rest of the world (world import demand excluding that of India) as compared to the late 90s (see figure 3) Such a rise in global demand occurred in two separate phases during the 2000s The first phase involved a global boom between 2003-04 and 2007-08 which came to an end with the emergence of global financial crisis in 2008 The second phase involved the brief recovery period of the global economy between 2009-10 and 2010-11 following synchronized fiscal stimulus packages implemented all over the world India’s export growth rate followed more or less the same trend as the global import demand during these two booms

Figure 3: Imports of Rest of the World and India’s Exports (at current prices), 1992-93 to 2013-14

Source: Calculated from UNCTAD Stats and WITS, COMTRADE Database

Note: Share of Resource-based commodities in merchandise exports is measured in thte secondary axis The commodity group is defined by following the technological classification of Lall (2000) and are constructed from 3-digit commodity groups of SITC-Rev2

(4) Rise in Private Investments and exports: What remained to be the common features

between the decades of the 90s and that of the 2000s, are the phenomena of (a) declining consumption ratio and (b) rising import ratio and (c) stagnant shares of government final consumption expenditures and public gross capital formation in GDP (see figure 4) What was distinct about the decade of 2000s, were (a) the sharp rise in the shares of exports and private sector’s capital formation in GDP and (b) decline in net exports

The rise in private investments in turn was primarily on account of higher corporate investments, as reflected by a sharp rise in the share of corporate sector in fixed capital formation during this period (Nagaraj, 2013) In terms of sectors, higher investments were driven by the registered manufacturing sector, the share of which increased in gross capital formation during the decade of 2000s as compared to the earlier period (ibid)

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 -30.0 -20.0 -10.0 0.0 10.0 20.0 30.0 199 -93 199 -94 199 -95 1995 -96 199 -97 199 -98 1998 -99 1999 -00 2000 -01 2001 -02 2002 -03 2003 -04 2004 -05 2005 -06 2006 -07 2007 -08 2008 -09 2009 -10 2010 -11 2011 -12 2012 -13 2013 -14

Growth Rate of RoW aggregate imports Growth Rate of India Exports

Growth Rate of RoW Merchandise Import

Figure 4: Share of Expenditure Components in GDP (as %)

Source: National Account Statistics, CSO, various years

Note: Consumption-GDP ratio is measured in the secondary axis

Though many countries witnessed higher growth rate of output and investments during the decade of 2000s as compared to the earlier period, what has been specific to India’s growth story is the very manner in which it achieved higher growth In sharp contrast to the experience of the US during the early 2000s where the stimulus for higher investment was argued to be provided by higher consumption ratio, or that of many East Asian countries where higher investments was accompanied by improvement in net exports, India witnessed a rise in investment rate despite the decline in both the consumption ratio as well as the net exports In the backdrop of the observed relationship between investments and gross exports, this paper attempts to explain the specificity of India’s growth story by locating gross exports as a stimulus for investments The next section attempts to build a conceptual framework to understand the relationship between the two

III Gross Exports as the Exogenous Stimulus: A Theoretical Framework

In a demand-constrained economy, there are at least two possible routes through which exports can be perceived to provide stimulus for investment in the domestic economy The first route is the one where an increase in net exports leads to a rise in effective demand and output of the

domestic economy through the conventional multiplier2, which in turn, can lead to further increase in the investment rate as capitalists expect higher realized profit in the future period on the basis of improved demand conditions of the present period In other words, here the external sector provides stimulus for investment by acting as a location of realization of surplus or exante savings for the domestic economy Our emphasis in this paper, however, would be on the second route

This second route pertains to a situation where instead of net export, it is the gross export which in itself acts as an inducement to invest in the domestic economy, over and above the stimulus provided by the changes in aggregate demand3 The possibility of such a stimulus for investment would arise if there exists any structural difference between the domestic market and the external market, such that for a given level of demand, the capitalists perceive the two markets differently while taking investment decisions The raison d’être of such a structural difference between the two markets from the perspective of the domestic capitalists in a developing country like India, in turn, can be argued to be the following:

In a globalized world, the domestic big capital competes with the foreign capital in both the foreign and the domestic market while catering to a given level of demand However, the difference between the two markets lies in the fact that in the domestic market, on account of decades of protection received from the state, the big corporate houses of the domestic economy had already emerged as “trusted brands”, built extensive distribution network and developed experience to cater to the domestic economy according to local tastes and preferences by the time of economic liberalization Such opportunities for the domestic capital are simply missing in the foreign markets where they have limited footholds and lack adequate technological capabilities to break into these markets In other words, compared to the domestic markets, the scope of domestic corporates carving out their niche markets vis-à-vis the foreign capital is much reduced Thus, from the perspective of the domestic capitalists, catering to one unit of demand in the foreign market would be fundamentally different from catering to one unit of demand in the domestic economy in terms of the competition they face from other capitalists In the backdrop of this fundamental difference between the two markets from the perspective of domestic

2

See Polak (1947) for a review of different forms of trade multiplier in a demand-constrained economy

3 As one of the possible interpretation of Luxemburg(1951), this route was explored by Patnaik (1972) to highlight

capitalists, there are at least two possible ways through which the rise in global demand can be argued to induce higher investment in India

Firstly, for a given level of desired and actual capacity utilization rate, an exogenous rise in

exports would be associated with higher investments if the capitalists earmark a part of their investment expenditures to open up or get access to the foreign market and such earmarked investments are assumed to be proportional to the level of foreign demand for domestic goods The global demand or foreign demand for domestic goods in this analytical framework would play the same role as does innovations in Kalecki (1954), since catering to external market itself requires incurring additional investment expenditures over and above the stimulus provided by the existing state of demand

Secondly, for a given level of demand, the desired capacity utilization rate itself can be a

semi-exogenous variable and respond to changes in global demand on account of the following reasons If the capitalists happen to maintain planned excess capacity as an instrument for deterring new entry in the market and for catering to unforeseen expansion in demand without letting it get “snatched away by new competitors” , in a manner described by Steindl (1973), then from the perspective of the domestic capitalists in a developing country like India, the fear of an additional unit of demand “getting snatched away” by competitors would be far greater in the foreign market than in the domestic market for reason discussed earlier Now, if the possibility of an additional unit of output getting “snatched away” by competitors is perceived to be greater in the foreign market than that in the domestic market by domestic capitalists, then ceteris paribus, it can be argued that a rise in the foreign demand would be associated with domestic capitalists holding additional planned excess capacity, over and above the one which the capitalists had initially desired at unchanged global demand But, maintaining excess capacity also implies higher cost from the perspective of the capitalists4 Thus, a decline in global demand from the existing level would also lead the capitalists to cut back their desired excess capacity, in order to reduce their costs from the existing level In short, changes in global demand would lead to similar changes in planned excess capacity of the capitalists at a given level of aggregate demand And since, a rise in desired excess capacity or reduction in desired capacity utilization

4

rate entails addition in capital stock in order to increase capacity output, rise in global demand would also lead to similar changes in investment rate for a given level of aggregate demand In any of the above two cases, gross exports would act as a separate stimulus for investment for a given level of aggregate demand And if it does, as we shall now see, any exogenous rise in gross exports would not only push up the growth rate, but in itself would also open up the possibility of

reducing the net exports, a phenomenon which was witnessed in the Indian economy during the

liberalization period To lay bare the dynamics of such a process, we construct a theoretical model within a standard underconsumptionist Kaleckian framework

The Basic Model

In order to highlight the relationship between gross exports, investment and net exports, we ignore the implication of fiscal policies and consider an open economy without government intervention We assume a one-commodity world with an oligopolistic set-up where investment is undertaken solely by oligopolistic firms and prices are fixed according to a given mark-up In our model economy, neither are firms finance constrained in the credit market nor does there exist any BOP constraints for the external sector The only constraint before this model economy is the level of effective demand, as the actual output is assumed to remain below the potential output throughout the analysis For the sake of simplicity, depreciation level is assumed to be zero

= (1)

Where =

∗

= ∗

0< <1; 0<π<1; β>0

Similar to various open economy Kaleckian models, the level of exports is assumed to be determined by exogenously given foreign demand But unlike many other post-Kaleckian models, as that in Onaran and Galanis (2012) and Hartwig (2013), exports are assumed to be unaffected by changes in real exchange rate This assumption is consistent with India’s export trend, where the real exchange rate did not show any statistically significant relationship with the growth rate of exports during the post-liberalization period5 The level of imports is simply assumed to be the product of an exogenously given import propensity and the level of output Accordingly, the export-capital ratio and the import-capital ratio of this model economy are described as equation (2) and equation (3), where is the exogenously given export-capital ratio, ‘ ’ is the import-capital ratio and ‘m’ is the exogenously given import propensity

= (2)

= (3)

Where, >0; m>0

Though the nature of investment function has been one of the most widely debated and contested issues among the authors adhering to the demand-side framework, nonetheless, what is common within all such models and theories is the emphasis of causal relationship from aggregate demand to desired investment rate of capitalists in the long run Accordingly, taking cue from

5

Amadeo (1986 and 1987) and Hein (2014), this causal relationship is described through an investment function which is positively affected by the capacity utilization rate of firms The specificity of the investment function of corporate sector in a developing country India, for reasons discussed earlier, would be the exports acting as a separate stimulus for investment over and above the one provided by the capacity utilization rate or the aggregate demand Such an investment function is described by equation (4), where ‘g’ is the growth rate of capital stock or the investment rate and ‘a’ is any exogenous parameter For the sake of convenience, we shall henceforth call the parameters ‘b’ and ‘c’ the output coefficient and the export coefficient respectively, both of which are assumed to be exogenous and positive Condition (C.1) is assumed to hold to ensure that the investment function has a positive intercept in the investment-capacity utilization rate space

= + + (4)

+ ≥ (C.1)

Where = b>0; c>0

In the backdrop of the above discussion, we now define the net exports and a σ-function of the economy The net export-capital ratio is given by

= −

, = − (5)

The σ-function is defined as the sum of domestic savings and net imports as described in equation (6)

= + − (6)

= ( + ) − (7)

The macroeconomic (Keynesian) stability condition of the model economy will be fulfilled if the responsiveness of ‘σ’ to a unit change in capacity utilization rate is greater than that in the case of investment rate This condition can be written as the following:

>

, ( + ) − > (C.2)

The above stability condition ensures that any gap between exante investment and the sum of domestic savings and net import at a given capacity utilization rate calls forth output adjustment in a manner that the capacity utilization rate adjusts to equalize ‘g’ and ‘σ’ Throughout this exercise we shall assume this macroeconomic stability condition to hold

In order to bring out the specificity of this model and highlight the implication of a positive export coefficient, an additional condition is imposed by assuming that the responsiveness of domestic savings rate to a unit change in capacity utilization rate is itself greater than that in the case of investment rate In other words, the Keynesian stability condition is assumed to hold even if the model economy is perceived to be closed without incurring any imports whatsoever Such a condition can be described as

>

, − > ( 3) With this we now examine the long-run equilibrium and dynamics of this model

Long-run Equilibrium

The steady-state equilibrium condition in the long run is given by the equality between investment and the sum of domestic savings and net import as shown in equation (8)

The long run equilibrium value of capacity utilization rate (u*) is derived by solving for ‘u’ from equation (4), (7) and (8) and described in equation (9) The numerator of the RHS of the above equation is positive since by hypothesis + is positive, whereas, the denominator of the RHS is positive on account of Keynesian stability condition (C.2) Thus the long run equilibrium level of capacity utilization rate u* is positive

∗= + (1 + )

( + ) − (9)

Plugging in the value of u* in the investment function of equation (4), we derive the long run equilibrium level of investment rate g* as described in equation (10) Since u*, b and a+c are positive, the long run steady-state investment rate is also positive

∗= + + (1 + )

( + ) − + (10)

Similarly, plugging in the value of equilibrium level of capacity utilization rate u* in the net export function of equation (5), we derive the steady-state equilibrium level of net exports as equation (11) and 6.13a In other words, n* is that level of net exports at which the steady-state equilibrium condition holds with g being equal to σ

∗ = − + (1 + )

( + ) − (11)

Figure graphically reflects the manner in which the steady state capacity utilization rate (u*), investment rate (g*) and net exports (n*) are determined in our model The upper panel measures ‘g’ and ‘σ’ in the vertical axis and capacity utilization rate (u) in the horizontal axis The investment function described by equation (4) is shown as g0 and the σ function described by

equation (7) is shown as σ0 The intercept term of g0 is a+cx, while its slope is ‘b’ Similarly, the

intercept term of σ0 is - , while its slope is ( + ) The slope of g0 is flatter than that of σ0

due to stability condition (C.2) The investment rate g equals σ at the intersection point E0 to

determine the equilibrium level of investment rate g0* and capacity utilization rate u0*

intercept term of n0 is , while the slope is − The negatively sloped n0 cuts the horizontal

axis at ub While any u>ub would be associated with trade deficit, any u<ub would lead to trade

surplus In the present exercise, the steady state capacity utilization rate (u0*) happens to be

lower than ub and determines the net exports at ∗ For the sake of convenience, in all the

subsequent diagrams we shall stick with this assumption that the steady state capacity utilization rate is such that the level of net exports is positive

Figure 5: Determination of Steady State Investment Rate, Capacity Utilization Rate and Net Exports

Having described the steady state solutions, let us now proceed to examine the impact of exogenous change in exports on steady state investment rate, capacity utilization rate and the net exports

∗

+

O

,

∗

∗

(+)

(−)

Impact of Change in Exports

The impact of an exogenous change in gross exports on steady-state capacity utilization rate is shown in equation (12) by taking partial derivative of equation (9) w.r.t ‘ ’ As evident from the equation, that

∗

is unambiguously positive Here the exogenous rise (fall) in exports increases (reduces) the capacity utilization rate through two routes The first route, in turn, involves two mechanisms –(i) the conventional multiplier effect where capacity utilization rate rises (falls) at a given investment rate on account of rise in gross exports and (ii)the consequent rise (fall) in the investment rate through the conventional output coefficient which further increases (reduces) the capacity utilization rate over and above the rise (fall) brought about by the multiplier effect Since this process does not involve the export coefficient whatsoever, the extent of change in capacity utilization rate through the first route can be calculated simply by putting the value of ‘c’ at in equation (12) and would be equal to

( ) The second route involves

specifically the export coefficient, where capacity utilization rate rises (falls) on account of such rise (fall) in investment rate which is brought about by the rise (fall) in gross exports at a given capacity utilization rate The extent of rise in capacity utilization rate through this second route can be calculated simply by deducting

( ) from equation (12) and would be equal to

( )

∗

= +

( + ) − > (12)

The impact of a unit change in gross exports on steady state investment rate is shown in equation (5.21) by taking partial derivative of equation (5.12) w.r.t ‘ ’ As evident from the equation, that

∗

case of capacity utilization rate, the extent of increase in investment rate through the first route would be

( ) , whereas its rise through the second route would be equal to ( ) +

∗

= (1 + )

( + ) − + > (13)

The impact of change in gross exports on net exports is calculated by taking a partial derivative of equation (11) w.r.t ‘ ’ and shown in equation 5.22

∗

=( + ) − − (1 + )

( + ) − (14)

Thus, the sign of the above partial derivatives will be determined by the following conditions:

∗

> ( + ) − − (1 + ) ( + ) − > ,

∗

≤ ( + ) − − (1 + ) ( + ) − ≤

Since ( + ) − >0, the above conditions can be re-written as condition C.4 and C.5:

∗

> < ( 4)

,

∗

≤ ≥ ( 5) Where

=( − ) (15)

The critical value of import propensity is denoted by ‘mc’, which is defined as that level of

critical value of import propensity (mc) as described in equation (15) is positive even if the

stringent stability condition (C.3) is assumed to hold since c>0 Now, as evident from conditions (C.4) and (C.5), that the sign of the partial derivative of net exports w.r.t gross exports depends on the precise value of import propensity (m) compared to the critical value of import propensity (mc) Thus, the sign of the partial derivative would be ambiguous as there is no guarantee that the

import propensity would be lower than this critical value

This result is in sharp contrast to the conventional Keynesian-Kaleckian proposition without export coefficient, where the stringent stability condition (C.3) would act as a sufficient condition to ensure a positive relationship between gross exports and net exports This can be easily checked by calculating the impact of change in gross exports on net exports in the absence of a export coefficient, i.e by putting the value of ‘c’ at in equation (14) as shown in equation in (16) Since − >0 (by condition C.3), ∗| >0 But why does the inclusion of a positive export coefficient at all make the direction of change in net exports ambiguous in the case of change in gross exports?

∗

| = ( − )

( + ) − > (16)

Since ceteris paribus any rise in output reduces net exports by increasing imports, the adverse impact of higher output on net exports, brought about by rise in gross exports, would be greater when there exists a positive export coefficient ‘c’ than it would be otherwise But to what extent would such imports rise or net exports fall on account of a given change in output level, ceteris

paribus, would be dependent on the precise value of ‘m’ Accordingly, an economy with import

Figure 5a shows the conventional case when m<mc Since the intercept term of σ0 contain the

parameter ‘ ’, the exogenous rise in gross exports makes the σ function shift rightward from σ0

to σ1 by increasing its intercept term The steady state investment rate increases from g0* to gA*,

whereas the steady state capacity utilization rate increases from u0* to uA* as the intersection

point between σ and investment function moves from E0 to EA This process is associated with

the first route through which investment rate and capacity utilization rate rises involving the multiplier and output coefficient

Since intercept term of net export function is positively affected by rise in ‘ ’, the net export function in the lower panel shifts upward from n0 to n1 on account of rise in gross exports As the

capacity utilization rate rises from u0* to uA, the level of net exports rises from n0* to nA with the

intersection point between net exports function and steady state capacity utilization rate moving from F0 to FA This rise from n0* to nA describes the process where the rise in gross exports leads

to higher output and net exports involving only the multiplier effect and output coefficient at a given level of profit share and import propensity

But the existence of a positive export coefficient implies that the rise in capacity utilization rate will not be restricted to uA as the investment function g0 shifts upward from g0 to g1 on account

of rise in gross exports This leads to a further rise in steady state capacity utilization rate and investment rate to u1* and g1* respectively, as the intersection point between σ and investment

function moves from EA to E1 This process describes the specificity of a positive export

coefficient where investment rate positively responds to gross exports even at a given capacity utilization rate and hence, constitutes the second route through which investment rate and capacity utilization rate rises Such a rise in capacity utilization rate, ceteris paribus, would lead to a fall in net exports from nA to n1* as the steady state capacity utilization rate moves along the

net export function n1 from FA to F1 Since in this case m<mc, the positive impact of higher gross

exports dominates over the negative impact of higher capacity utilization rate on the net exports with n1*>n0* In other words, despite the fact that the extent of improvement itself remains

Figure 5a: Impact of Rise in Gross Exports: The Conventional Case (m<mc)

Let us now consider the puzzling case where m>mc and the net exports decline on account of rise

in gross exports Here, the slopes of σ and net export functions are steeper than they were in the previous case as the import propensity is higher As shown in Figure 5b, the impact of change in gross exports on capacity utilization rate and investment rate in the upper panel is essentially similar in this case compared to the previous one as the steady state capacity utilization and investment rate ultimately settle at u1* and g1* respectively, while the intersection point between

σ and investment function moves from E0 to E1 involving the outward shift of both the σ and

investment function6 And similar to the previous case the net export function shifts upward from n0 to n1 While the movement of intersection point between net exports function and steady state

6

While the direction of change in steady state investment rate and capacity utilization rate remains similar between the conventional and the puzzling case, nonetheless, it can be noted that the extent of their increase on account of unit increase in gross exports would be lower in the puzzling case due to higher import propensity compared to the conventional case

O

(+)

(−)

∗ ∗

,

∗

∗ ∗

∗

capacity utilization rate from F0 to FA describes the impact of multiplier effect and output

coefficient, as in the previous case, the movement of the intersection point along the net export function n1 from FA to F1 reflects the impact of the export coefficient But the specificity of the

puzzling case lies in the fact that the decline in net exports on account of rise in capacity utilization rate is sharper compared to the conventional case, such that it dominates over the positive impact of higher gross exports on net exports and pushes the level of net exports to n1*

with n1*<n0* This is because of the steeper slope of net export function here, brought about by

higher import propensity Thus, the existence of an export coefficient in the puzzling case gives result which stands in sharp contrast with the conventional proposition, as the rise in gross exports in this case leads to a deterioration of trade balance

Figure 5b: Impact of Rise in Gross Exports: The Puzzling case (m>mc)

While central to this conclusion is the existence of a positive export coefficient, the relevant question in the Indian context is whether such an investment function can be empirically observed during the liberalization period This is precisely what we test in the next section

∗

O

∗

(+)

(−)

∗

∗ ∗

,

∗

IV Gross Export as the Exogenous Stimulus: The Empirical Illustration

In this section we examine the hypothesis of a rise in the rate of investment on account of rise in exports for a given level of capacity utilization rate Taking the investment rate as the dependent variable and the capacity utilization rate and the exports-capital ratio as the independent variable, what we want to test specifically is whether the parameter associated with exports is positive We test this hypothesis in two different ways On the basis of ASI data, first we a panel data estimation in the case of the registered manufacturing sector to examine whether higher export was also associated with higher investment after accounting for changes in the capacity utilization rate Secondly we a time-series econometric exercise to examine the same relationship between corporate investment rate and exports, but now on the basis of National Account Statistics data Let us start with the panel data estimation

Panel Data

For the purpose of panel data regression, the disaggregated data at 2-digit level of registered manufacturing sector is taken as the unit of observation One of the major problems which one confronted till recently in doing such an analysis from the data provided by the Annual Survey Industries (ASI), was extending the analysis beyond 2007-08 on account of lack of concordance between the base NIC-2004 and NIC-2008 at the digit level7 However, the EPWRF has recently provided concorded series of most of the relevant indicators of Annual Survey of Industries at the 2-digit level, which has made the disaggregated analysis of registered manufacturing sector possible even beyond 2007-08 Further, unlike the National Account Statistics data, this has the added advantage that one can construct a comparable time-series data even beyond 2012-13 Thus, using the concorded series of ASI as provided by the EPWRF, the time period selected for panel data regression is 1992-93 to 2014-15 Since, the concorded series is in terms of NIC-2004, the relevant 2-digit sectors which need to be included within the manufacturing category are the ones from 15-36 Thus, the number of sectors at the 2digit level is selected to be 22 While the rate of investment and the indicator for capacity utilization rate would be calculated from the above source, for the export data of India and import data of rest of the world we shall rely on the COMTRADE database provided by the World Bank And since, the NIC-2004 base is comparable with ISIC revision 3.1, the trade data at the digit level shall include sectors from 15-36 for ISIC revision 3.1 From these data sources, we construct four variables namely, the rate of investment (g), the capacity utilization ration rate (u), the export-capital ratio (x) and the ratio between import of rest of the world and the export-capital stock (row) These four variables, in turn, are calculated in the following manner:

We calculate the investment rate (g) for each sector in a given period by deflating its gross capital formation with the stock of fixed capital measured at the beginning of the period It can be noted that for a given period, the ASI gives the data for fixed capital stock as the closing value of assets during the end of the period Since, conceptually the calculation of rate of investment requires the data for capital stock at the beginning of the period, the rate of investment for a given financial year ‘t’ is calculated by deflating gross capital formation of year ‘t’ with fixed capital stock estimated in year ‘t-1’

7

In the absence of any information on the technological output-capital ratio, we simply assume the ratio between gross valued added and fixed capital stock to indicate the trend in the capacity utilization rate during the given period In other words, we assume the technological output-capital ratio to remain unchanged during this period Since technological changes in India during the post-liberalization period have been typically characterized by the adoption of labour saving technologies, instead of augmenting ones which increases the technological output-capital ratio (Patnaik, 2003), the assumption of constant technological output-output-capital ratio seems to be perfectly plausible Thus, capacity utilization rate (u) is calculated for a given period by deflating the gross value added with fixed capital stock measured at the beginning of the period The data source for the gross value added and fixed capital stock is the concorded series of Annual Survey of Industries provided by the EPWRF

From the COMTRADE database, we take the data for India’s export at the digit level of ISIC revision 3.1 and convert it in terms of the domestic currency by multiplying the export value with India’s nominal exchange rate per unit of US dollar during a specific period The export-capital ratio for each sector is then calculated by deflating exports with stock of fixed export-capital measured at the beginning of the period Similarly, the import of rest of the world at the digit level of ISIC revision 3.1 is calculated by subtracting India’s imports from the world imports and subsequently, converting it to domestic currency by multiplying the nominal exchange rate vis-à-vis US dollar in any given period The data source for fixed capital, as earlier, is the concorded series of Annual Survey of Industries as provided by the EPWRF

multicollinearity For different 2-digit sectors during the period 1993-94 to 2014-15, figures and show the scatter plots between the independent variables in the two models respectively, namely, the lagged value of the export-capital ratio and the lagged value of capacity utilization rate in the first model and that between the lagged value of rest of the world import-capital ratio and lagged value of capacity utilization rate in the second model As evident from the two charts, the correlation between the independent variables is weak in most of the sectors In fact, except for sector 36 (in both the models) and sector 33 (in model 2), scatter plots for all other 20 sub-sectors show this weak statistical relationship between the independent variables It can be noted, that this weak statistical relationship between the ex post values of the independent variables is particularly due to the decade of 90s, when the indicator of capacity utilization rate declined despite the export-capital ratio and the ratio between world imports and capital stock remaining more or less unchanged Thus, even though the independent variables moved in similar direction during the decade of 2000s, the R-squared value between the variables remains low for the entire period 1993-94 to 2013-14 Since the problem of multicollinearity is not significant here, we now proceed to estimate our two models through the following steps

Figure 7: Scatter Plot between the Independent Variables of Model

Table 2: Panel Unit Root Test Results (p-values) Fisher Test

Levin-Lin-Chu Test Inverse

chi-squared

Inverse normal

Inverse logit

Modified inv chi-squared

g 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.034 0.000 0.000 0.001 0.006 0.015 0.008 0.002 0.000

In the second step, we specify the equations which we shall estimate in the regression analysis For the sake of simplicity, we shall consider one lag for each of the independent variables We use the F-test to choose between the pooled OLS method and the fixed effect method, where the null hypothesis states that all the constants are homogenous across panels and accordingly, the pooled OLS method is applicable The F-test rejects this null hypothesis at 5% significance level for both of our models In order to choose between random effect and fixed effect regression, we apply a standard Hausman test in the case of our two models The null hypothesis of the Hausman test states that the random effect model is valid In both the cases, the Hausman tests reject the null hypothesis at 1% significance level and suggest the use of fixed effect model Accordingly, our two models can be written in the form of following equations which we respectively term as Model and Model For reasons discussed in the earlier section, we would expect statistically significant and positive values for all the coefficients associated with the independent variables ( , , )

= + + + (Model 1) = + + + (Model 2) Where

i=1…N and t=1…T

is the capacity utilization rate in sector ‘i’ at period ‘t-1’

is the export deflated by capital stock in sector ‘i’ at period ‘t-1’

is the import of rest of the world deflated by capital stock in sector ‘i’ at period ‘t-1’ and are the constant terms for sector ‘i’ in model and model respectively

and are the error terms for model and model respectively

In the third step, we check for the presence of autocorrelation, heteroskedasticity and cross-sectional dependence in our models after performing fixed effect regression8 We a Wooldridge test in order to check for autocorrelation where, the null hypothesis states that there exists no first order autocorrelation In order to check for the presence of heteroskedasticity, we use a modified Wald test where the null hypothesis states that there exists no heteroskedasticity And lastly, to check for the presence of cross-sectional dependence, we Pesaran’s Test where the null hypothesis is of no cross-sectional dependence The relevant p-values associated with each of these tests are reported in table The null hypothesis of each of these tests are rejected at 1% significance level Thus, not only we have the problem of autocorrelation and heteroskedasticity, we also have cross-sectional dependence in the regression models But, as argued by Hoechle (2007), that in the presence of cross-sectional dependence, “standard error

estimates of commonly applied covariance matrix estimation techniques-e.g., OLS, White, and Rogers or clustered standard errors-are biased, and hence statistical inference based on such standard errors is invalid” Thus, as suggested by Hoechle, first we re-examine whether fixed

effect model is preferred to random effect models by performing a robust Hausman test which is consistent with spatial dependence and uses the Driscoll and Kraay standard errors Similar to the earlier result, however, the robust Hausman test also indicates the use of fixed effect models at 1% significance level Thus, we continue with our fixed effect models

8

Table 3: Test for Autocorrelation, Heteroskedasticity and Cross-Sectional Dependence

Ho Model

(p-value)

Model 2(p-value) Wooldridge Test for Autocorrelation No first order

auto-correlation

0.03 0.03

Modified Walt Test for group-wise Heteroskedasticity

Homoskedasticity 0.00 0.00

Pesaran’s Test for Cross-Sectional Dependence

No Cross-sectional Dependence

0.00 0.00

Lastly, in the presence of cross-sectional dependence, as suggested by Hoechle (2007), we re-estimate the fixed effect models using the Driscoll and Kraay standard errors, which are consistent with heteroskedasticity and autocorrelation and are robust to general forms of spatial and temporal dependence The result of this robust fixed effect regression analysis in the case of our two models is reported in table The coefficients for both the capacity utilization rate and the export-capital ratio in model are positive and statistically significant at 1% level Similarly, in model 2, the coefficients for both the capacity utilization rate and the ratio of rest of the world imports are positive and statistically significant at 1% level The within R-squared value in the two models are 0.23 and 0.24 respectively

Table 4: Fixed Effect Regression with Driscoll-Kraay Standard Errors

Model Model

Coefficient for 0.19*

(0.0239)

0.18* (0.0262)

Coefficient for 0.01*

(0.0036)

Coefficient for 0.0002*

constant 0.12* (0.0324)

0.11* (.0298)

within R-squared 0.23 0.24

Prob>F 0.00 0.00

Number of Observations 484 484

Note: ‘*’ indicates statistical significance at 1% level The Driscoll and Kraay standard errors are given in the parenthesis

Thus, the panel data exercise strongly supports the claim of the existence of an export accelerator in the registered manufacturing sector With this, we now move on to the time-series analysis

Time Series Analysis

The objective of this exercise would be to estimate the investment function of the corporate sector from the National Account Statistics (NAS) data Similar, to earlier exercise, the rate of investment shall be taken as the dependent variable and the indicators for capacity utilization rate and that of the global demand shall be taken as independent variables Since, comparable annual time series from the beginning of the liberalization period can only be constructed till 2012-13 from NAS, the period of our analysis will be from 1992-93 to 2012-13 Again, for reasons discussed earlier, we shall use the lag values of the indicators as the independent variables for the purpose of estimation

The data for gross capital formation and the capital stock of the corporate sector is provided by the NAS at the current prices Since, the capital stock at a given period is measured at the beginning of the period in NAS, we calculate the rate of investment of the corporate sector (g) of a given period by deflating its gross capital formation with capital stock of that period

aggregate value added economy (GDP at fixed cost) at current prices This value added of the private sector at current prices is then deflated by the capital stock of the corporate sector at current prices to construct the indicator for capacity utilization rate of the corporate sector during a given period (u)

Since, changes in the level and the pattern of global demand has been associated with similar changes in level of India’s exports, we take the aggregate exports of goods and services as an indicator of global demand Accordingly, the aggregate export of goods and services at current prices is deflated by the capital stock of the corporate sector at current prices to construct the export-capital ratio of a given period The change in this export capital ratio (x), in turn, is assumed to reflect similar changes in global demand for a given stock of capital of the domestic corporate sector

Accordingly, on the basis of these three variables we try to estimate the relevant coefficients of corporate investment function from the NAS data For the purpose of estimation, we shall consider the following investment function We would expect a statistically significant and positive value for the coefficients associated with the independent variables (α and β) Before estimating the model, again, we check for the existence of mulicollinearity Similar to the case of registered manufacturing sector, the problem of multicollinearity is not significant in this model as indicated by the fact R-squared value between the independent variables remained below 0.5 (at 0.4) for the sample period as a whole9 The reason why the problem of multicollinearity is not significant in this model, again, is the trend witnessed during the decade of 90s when the indicator for capacity utilization rate declined despite the export-capital ratio remaining more or less unchanged Thus, even though both the export-capital ratio and the indicator for capacity utilization rate moved in similar direction during the decade of 2000s, the R-squared value between the variables remains low for the sample period as a whole With this, we now move on to estimate the regression model as described in equation (17)

= + + (17)

9

To start with, we check for the presence of unit roots in any of these variables Table reports the p-values of the Augmented Dickey Fuller test conducted for each of the variables both at their levels and their first difference values The null hypothesis of the test states that the variables contain unit roots As evident from the table, that the dependent variable ‘g’ is a I(1) process, for which we can reject the null hypothesis at 5% significance level only if we take its first difference value In the case of level-values of independent variables, the null hypothesis can be rejected at 10% significance level, whereas for its first difference values, the null hypothesis can be rejected at 5% significance level Thus, the independent variables in our model are I(0) processes at 10% significance level and I(1) processes at 5% significance level The conclusion that one can draw from the ADF test is that at least one of the variables in the model contains unit roots, while the order of integration for the other two variables is either or depending on the confidence interval one selects

Table 5: Augmented Dickey Fuller Tests for Unit Roots

Level First Difference

g 0.102 0.017

0.097 0.035

0.083 0.048

Note: Model includes a drift and a time trend

Accordingly, we consider an ARDL model where the optimal lag lengths of the variables were selected through the Akaike Information Criteria (AIC) However, since the Breusch-Godfrey LM test rejected the null hypothesis of no serial correlation at 5% significance level, we re-estimated the ARDL model using Newey-West’s heteroskedasticity and autocorrelation consistent (HAC) standard errors The bound test for this re-estimated ARDL model rejects the null hypothesis of no long run relationship at 1% significance level and suggests that the variables are cointegrated Table reports the long run coefficients of the error corrected model along with their associated p-values and (HAC) standard errors As evident from the table, that the coefficients for the both the capacity utilization rate and the export capital ratio are positive and statistically significant at 1% significance level The Adjusted R-squared of this error corrected model is 0.82

Table 6: Result of Cointegration Analysis

Variable Coefficient

HAC Std

Error t-Statistic Prob 0.33 0.1176 9.08 0.00 1.07 0.0573 5.77 0.00 C -0.49 0.0950 -5.21 0.00 Adjusted R-squared 0.82 Prob (F-statistic) 0.00 AIC -4.62 Durbin-Watson stat 2.74

Figure 8: Actual, Fitted and Residual Values

Thus, be it in the earlier panel-data exercise or in the present time-series analysis, the empirical exercises in this section found evidence for the existence of an export accelerator in the Indian economy Both in the case of the registered manufacturing sector and the corporate sector, investment rate is found to have statistically significant and positive relationship with exports and capacity utilization rate

V Reinterpreting India’s Growth Story

If the relationship between exports and investment rate is perceived in the above described manner, then the explanation for the rise in the investment and output during the 2000s and the post-liberalization period can be primarily located in the exogenous changes in global demand during this period Such exogenous changes can be argued to trigger a sequence of expansion in aggregate demand through at least four other routes over and above the one described in this paper They are as follows:

Firstly, higher demand following two episodes of global boom was associated with a rise in

were financed through higher personal loans from commercial banks in a manner described by Ghosh and Chandrasekhar (2009) and Nagaraj (2013)

Secondly, the consequent rise in elites’ income can be argued to be associated with a change in

the domestic demand pattern in favour of luxury goods and consumer durables (Patnaik, 2007) Since these goods are relatively more capital intensive, such a change in the domestic demand pattern led to further increase in investments (Chandrasekhar, 2011)

Thirdly, implementation of synchronized fiscal stimulus packages all across the world in the

immediate aftermath of the global economic crisis not only boosted global demand and India’s exports, but also created room to increase India’s own fiscal deficit during this brief period (Sen and Dasgupta, 2014) Thus it is hardly surprising that the withdrawal of synchronized fiscal stimuli during 2011 would be followed by a period of economic slowdown in the Indian economy

Accordingly, the growth story which emerges out of India’s accumulation process during the post-liberalization period is one of external dependence, where the domestic economy is primarily dependent on buoyant global demand and favourable external economic situation for gaining higher growth rate in domestic investment and output The export-induced investment led growth process outlined in this paper attempted to highlight this external dependence

VI Conclusion

Bibliography

Amadeo, E (1986) The role of capacity utilization in long period analysis Political Economy , 147-160 Amadeo, E (1987) Expectations in a steady-state model of capacity utlization Political Economy , pp 75-89

Azad, R., Bose, P., & Dasgupta, Z (2017) 'Riskless' Capitalism in India Economic and Political Weekly , Vol 52, no 31

Azad, R., Bose, P., & Dasgupta, Z (2017) 'Riskless' Capitalism in India Economic and Political Weekly , Vol 52, no 31

Basu, K., & Maertens, A (2007) The pattern and causes of economic growth in India Oxford Review of

Economic Policy , 143-167

Bhagwati, J., & Panagariya, A (2013) Why Growth Matters: How Economic Growth in India Reduced

Poverty and the Lessons for Other Developing Countries New York: Public Affairs

Bosworth, B., M Collins, S., & Virmani, A (2007, February) Sources of Growth in the Indian Economy

Working Paper 12901 Cambridge, U.K: NATIONAL BUREAU OF ECONOMIC RESEARCH

Chandrasekhar, C (2011) India's New, High-Growth Trajectory: Implications for Demand,Technology and Employment The Indian Journal of Labour Economics , Vol 54, No

Chandrasekhar, C., & Ghosh, J (2002) The Market that Failed New Delhi: Leftword

Chaudhuri, B (2013) Issues in Activity and Product Classification for ASI The Journal of Industrial

Statistics , 208-216

Dasgupta,Z (2018) The Growth in Investment and Output in the Indian Economy during the Period of Economic Liberalization

Ghosh, J., & Chandrasekhar, C (2009) The costs of coupling: The global crisis and the Indian economy

Cambridge Journal of Economics , 725–739

Harrod, R (1939) An Essay in Dynamic Theory Economic Journal , vol 49, no 193, pp 14-33

Hartwig, J (2013) Distribution and growth in demand and productivity in Switzerland (1950–2010)

Applied Economics Letters , 938-944

Hein, E (2014) Distribution and Growth After Keynes: A Post-Keynesian Guide Cheltenham, UK and Northamption, MA, USA: Edward Elgar

Hein, E., & Vogel, L (2009) Distribution and Growth in France and Germany: Single Equation

Estimations and Model Simulations Based on the Bhaduri/Marglin Model Review of Political Economy , 245-272

The Stata Journal , 1-31

Kalecki, M (1954) Theory of Economic Dynamics London and New York: Routledge

Kalecki, M (1971) Selected essays on the dynamics of the capitalist economy 1933-1970 CUP Archive Luxemburg, R (1951) The Accumulation of Capital London: Routledge and Kegan Paul Ltd

Nagaraj, R (2013) India’s Dream Run, 2003-08: Understanding the Boom and Its Aftermath Economic

and Political Weekly , 39-51

Nagaraj, R (2015a) Seeds of Doubt on New GDP Numbers: Private Corporate Sector Overestimated?

Economic and Political Weekly , 14-17

Nagaraj, R (2015b) Seeds of Doubt Remain: A Reply to CSO’s Rejoinder Economic and Political

Weekly , 64-66

Nagaraj, R (2015c) Size and Structure of India's Private Corporate Sector:Implications for the New GDP Series Economic and Political Weekly , 41-47

Onaran, O., & Galanis, G (2012) Is aggregate demand wage-led or proft-led? National and global

effects Geneva: ILO

Panagariya, A (2008) India:The Emerging Giant OUP USA

Panagariya, A (2013) Indian Economy: Retrospect and Prospect Richard Snape Lecture

Patnaik, P (1972) A Note on External Markets and Capitalist Development The Economic Journal , 1316-1323

Patnaik, P (2003) On the Need for Regulating Technological Change Social Scientist , 7-17

Patnaik, P (2007) A Model of Growth of the Contemporary Indian Economy Economic and Political

Weekly , 2077-2081

Pesaran, M., Shin, Y., & Smith, R (2001) Bounds testing approaches to the analysis of level approaches Journal of Applied Econometrics , 289-326

Polak, J (1947) The Foreign Trade Multiplier The American Economic Review , 889-897

Robertson, P E (2012):Deciphering the Hindu growth epic, Indian Growth and Development Review, (1), pp 5169

Rodriguez, F., & Rodrik, D (1999) Trade Policy and Economic Growth: a Skeptic’s Guide to

Cross-National Evidence NBER Working Paper no W7081

Rodrik, D (1999) Making Openness Work: The New Global Economy and the Developing Countries Washington, DC: The Overseas Development Counci

University Press

Sen, A (1970) Interest, Investment and Growth In A ( Sen, Growth Economics (pp Chapter 10, pp 220-231) Harmondsworth: Penguin Books

Sen, S., & Dasgupta, Z (2014) Economic Policies in India For Stimulation or Austerity and Volatility?

PSL Quarterly Review , 423-450