ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

Effects of Public Investment on Growth and Poverty

Counterfactual policy simulations of a sustained increase in public investment in infrastructure in India, financed through borrowing from commercial banks, show a substantial increase in private investment and thereby output in this sector. Due to increases in absorption, real private investment and output in all other sectors also seem to increase, resulting in several other macroeconomic changes. With a 20 per cent sustained increase in public investment in infrastructure, the government can accelerate real growth by 1.8 percentage points in the medium to long run, six to 10 years after the policy change. This will be accompanied by a 0.2 percentage point decline in the rate of inflation. The increase in income will lead to a 0.7 percentage point annual reduction in poverty in rural India. This shows the potential for achieving the much-debated 10 per cent aggregate real GDP growth in the Indian economy.

Effects of Public Investment on Growth and Poverty

Counterfactual policy simulations of a sustained increase in public investment in infrastructure in India, financed through borrowing from commercial banks, show a substantial increase in private investment and thereby output in this sector. Due to increases in absorption, real private investment and output in all other sectors also seem to increase, resulting in several other macroeconomic changes. With a 20 per cent sustained increase in public investment in infrastructure, the government can accelerate real growth by 1.8 percentage points in the medium to long run, six to 10 years after the policy change. This will be accompanied by a 0.2 percentage point decline in the rate of inflation. The increase in income will lead to a 0.7 percentage point annual reduction in poverty in rural India. This shows the potential for achieving the much-debated 10 per cent aggregate real GDP growth in the Indian economy.


here has been a lot of public debate in recent months, particularly after the presentation of the union budget for 2006-07 by the union finance minister, P Chidambaram, about (a) the need for achieving 10 per cent GDP growth and its feasibility, (b) the role and potential of the infrastructure sector in achieving the desired GDP growth, and (c) the ways and means of raising resources for public investment in the infrastructure sector and particularly, the use of accumulated foreign capital inflows for this purpose. This paper attempts to address these issues and seeks quantitative answers in a macroeconomic theoretical framework. The tool of counterfactual policy simulation is used for this purpose. The answers to the above questions seem affirmative as detailed below.1

A macroeconometric model is, as a system of simultaneous equations, seeking to explain the behaviour of key economic variables at the aggregate level, based on the received theories of macroeconomics. Macroeconometric modelling, in general, pursues two objectives: forecasting and policy analysis. The latter objective is the focus of this study. Fiscal and monetary policies are the foremost policies that are virtually analysed in macroeconometric models from their inception.

This paper attempts to utilise the tool of an aggregative, structural, macroeconometric model to analyse the macroeconomic effects of changes in selected exogenous variables for India. Before we give the details of the selected model, its estimation, etc, it would be useful to briefly look at the literature on this topic pertaining to India. A detailed review of macroeconometric models built for the Indian economy is beyond the scope of this paper.2 Since this study proposes to analyse the economy from a monetary framework, it would be worthwhile to look into how the monetary sector was modelled in the Indian context.3 This will be useful for identifying the research issues pertinent to this study.

Modelling the monetary sector and its links with fiscal and external sectors became a challenging task in India after the 1970s. Modelling money and monetary policy for the determination of real output and price level have increased considerably in India [Rangarajan and Mohanty 1997; Rangarajan 2000]. In these models, stock of money varies endogenously through feedback from reserve money, which changes to accommodate fiscal deficit and changes in foreign exchange reserves. The output supply is determined by real money balances and net capital stock, both with lags; while the price level depends on the money supply and production. Some models attempt to link the real, monetary and fiscal sectors [Krishnamurty and Pandit 1985; Murty and Soumya 2006a].

Public investment adds to real capital stock, which in turn increases the real output. Analysis of the effect of public investment on private investment indicates crowding-in [Krishnamurty and Pandit 1985]. More recent assessment suggests the weakening of this phenomenon in the last decade possibly due to resource constraint and the negative price effect of public investment financed by fiscal deficit [Krishnamurty 2001; IEG-DSE 1999; Rangarajan and Mohanty 1997].

Modelling the external sector was not a major concern in the earlier models because of restrictions on trade. But in the recent years, several models emerged with detailed emphasis on the external sector and its interlinks with the monetary and fiscal sectors. Krishnamurty and Pandit (1996) modelled the merchandise trade flows in the supply-demand framework and included disaggregated output, prices and investment behaviour.

The macroeconomic impact of fiscal deficit on balance of payments in India is an emerging issue since the inception of stabilisation programme. Rangarajan and Mohanty (1997) postulated that fiscal deficit increases the absorption in the economy relative to output and the output effect of deficit follows with a lag.

In a recent paper, Sastry et al (2003) have analysed the sectoral linkages between agriculture, industry and services in the Indian economy. The study emphasised the role of agriculture through its demand linkages with other sectors in determining the overall growth of the economy. The next section outlines the methodology and the proposed model of this study.

I Methodology and Proposed Model

This paper tries to extend the work by the authors [Murty and Soumya 2006a], wherein they attempted to build a small macroeconometric model for India using the absorption approach of Polak. Both these efforts extensively utilise the work of Rangarajan and Mohanty (1997). Some important changes have been made to expand that model and to address the theme of this paper. The basic model is monetarist in focus. It emphasises the inter-relationships between internal and external balances and also the relation between money, output, prices and balance of payments.

The model strives for a balance between the two polarised approaches of the classicals and the Keynesians. While classicals contend that changes in money supply ultimately result in changes in the price level, the Keynesians postulate that the changes in money supply eventually lead to changes in output under conditions of less than full employment. Viewing reality lying somewhere in between these two extremes, one can postulate that changes in money supply affect both the output and the price level. Thus, the model tries to capture the effects of changes in money supply on both output and price level.

The model mainly focuses on the determination of money supply and its links with fiscal operations and the impact of money stock on output. It is postulated that real money balances or credit affects output besides the real capital stock. An increase in real credit results in monetary expansion, which in turn has an effect on the aggregate output and price level. A rise in output through increase in credit neutralises the rise in price level caused by monetary expansion. Further, RBI credit to finance the fiscal deficit, the latter defined as government total expenditure less government total receipts, causes money supply to increase endogenously with the rise in reserve money. This monetary expansion again affects the price level and output to a lesser extent, and the cycle continues.

In the proposed model, private investment is assumed to be explained by (a) public investment in that specific sector, (b) real interest rate, (c) public sector resource gap, and (d) sectoral output price. The public sector resource gap variable defined as the difference between gross public sector savings and investment is common to all the four sub-sectors and is expected to have a negative correlation with private investment. Based on the net effect of the above four explanatory variables of private investment, we classify whether there exists “crowding-in” or “crowding-out” between public and private investments. If the net effect is positive (negative), we say that there exists crowding-in (crowding-out) respectively.

The proposed model also incorporates the savings-investment identity through current account balance. It also has an interest rate equation in a reduced form. The interest rate determinants are changes in bank credit to the commercial sector, current account balance, rate of inflation and equilibrium level of gross domestic savings.

The external sector is modelled through demand (and supply) for exports, demand for imports and balance of payments (BOP) identity. Assuming equilibrium in the exports market, the export supply function is specified as a price equation for unit value of exports. It incorporates world real income, relative price and the export price of the rest of the world. The export demand depends on relative export price and the real domestic income. The import demand function depends on the domestic absorption and the relative import price. The nominal exchange rate is a function of the domestic price level, current account balance and change in foreign assets of the Reserve Bank of India (RBI).

In order to link economic growth with poverty reduction, the model postulates a simple relationship between headcount ratio and the per capita real income, separately in rural and urban areas.

Proposed Model

Based on the methodology outlined above, we propose the following model,4 which consists of four blocks – real, fiscal, monetary and external sectors. These four blocks are regrouped into three separate modules for econometric estimation. Module I consists of all macroeconomic equations covering fiscal, monetary and external sectors. Module II covers all real sector equations, which include production,5 investment, and prices. Module III has only two equations representing rural and urban poverty ratios. The description of variables is given in the Appendix II. Module I: Fiscal, monetary and external sectors Fiscal sector:

  • (1) DT = f (YNAR, PGDP)
  • (2) DIT = f (Y)
  • (3) NTX = f (YM)
  • (4) CONS = f (YM/P)
  • (5) PC = f (PYDR)
  • (6) FD = f (GXP, TR, (P-P-1)/P-1) Monetary sector:
  • (7) P = f (YR, M3, IB)
  • (8) M3 = f (RM)
  • (9) IB = f ((Δ BCP + CAPB), (P-P-1)/P-1, SAV) External sector:
  • (10) EXPT = f ( UVIX/EXR/WPEXP, WYR)
  • (11) UVIX = f (P/EXR, WYR, WPEXP, EXPT-1)
  • (12) IMPT = f (UVII*EXR/P, AD)
  • (13) EXR = f (P, CAB, ΔRBFA) Link equation:
  • (14) PGDP = f (P) Module II: Real sector Production functions:
  • (15) YAR = f (RAIN, AREA, KAGR-1, YINFR-1)
  • (16) YMNR = f (ADD, KMNR)
  • (17) YINFR = f (KINFR-1, M3 -1/P-1)
  • (18) YSRR = f ( KSRR-1, M3/P) Investment functions:
  • (19) PIAG = f (YAR-1,PCFAG-1, PIINF-1, Real IB, PSRG-1, PRAG) where PSRG: public sector resource gap = PCFSAV/PGKE– PCFTOT
  • (20) PIMN = f (PCFMN, PIINF, Real IB, PRMN)
  • (21) PIINF = f (PCFINF-1, PRINF)
  • (22) PISR = f (PCFSR, PIINF-1, Real IB, PRSR)
  • (23) DEPAG = f (KAGR-1)
  • (24) DEPMN = f (KMNR-1)
  • (25) DEPINF = f (KINFR-1)
  • (26) DEPSR = f (KSRR-1) Output prices:
  • (27) PRAG = f (YAR, PYDR, P)
  • (28) PRMN = f (P)
  • (29) PRINF = f (YINFR, PYDR, P) Module III: Poverty ratios
  • (30) HCRRUR = f (PYDR/NTOT)
  • (31) HCRURB = f (PYDR/NTOT) Identities: 1 PYD =YM – TR + TRP + PYDIFF 2 PYDR = PYD/P

    II Trends and Patterns in Indian Macroeconomy

    It is important to understand the trends and patterns in the observed data before estimating the proposed model and using it for counterfactual simulations. This provides a backdrop for interpreting the empirical results to be obtained. The data were taken from the National Accounts Statistics (NAS), published by the CSO, and the Handbook of Statistics on Indian Economy, published by the RBI. The poverty estimates are based on the National Sample Survey (NSS) data.

    The study period is 1978-79 to 2002-03. Although data are now available for two more recent years for GDP and few other variables, there are gaps for many other variables and therefore we confined our analysis to the above period. For any macroeconometric model, the choice of sectoral break-up is very important and it determines the overall size of the model. Here, we chose a four sector disaggregation for the investment and outputs of the real sector from the NAS. These four sub-sectors are (a) agriculture including forestry and fishing (industry group 1), (b) manufacturing including mining (industry groups 2 and 3), (c) infrastructure, which includes electricity, gas, water supply; construction; and transport, storage and communication (industry groups 4, 5 and 7), and (d) services sector, covering all other activities (industry groups 6, 8 and 9). For simplicity these four sub-sectors are called (i) agriculture, (ii) manufacturing, (iii) infrastructure, and (iv) services respectively, in the rest of the paper.

    Most of the variables for the real and external sectors used in the econometric analysis are in real form (constant 1993-94 prices) to avoid inflationary effects. The monetary and fiscal variables are in current prices. All price variables are indices with 1993-94 as unity. To study the macroeconomic trends, decade-wise annual average compound growth rates for all the variables are computed using semi-logarithmic regressions6 and are given in Appendix I, Table 1. To analyse the structural changes/patterns, average percentage shares of important variables are also given in the same table. A few variables are also plotted to understand visually the trends and fluctuations in them (Chart 1).

    Output and Prices

    Real gross domestic product at factor cost, an indicator of total economic activity or proxy for real income, grew by a moderate

    5.7 per cent per annum (pa) during the entire study period 1980-81 to 2002-03. The relatively good performance of the Indian economy during the post-1980s, compared to the earlier period, is attributable to the better utilisation of industrial capacity and favourable demand conditions. The real output growth has accelerated from 5.4 per cent during the 1980s to 6.2 per cent during the 1990s. Between 1993 and 2003, the post-liberalisation decade, which is also our data period for policy simulation analysis, the real output has grown at 6 per cent pa, which implies a significant slowing down in the economy during 2000-03. Real per capita output (income) also shows similar trends.

    The above aggregate growth was made possible through differential sectoral growth: agricultural output grew by 3 per cent, manufacturing by 6.6 per cent, infrastructure by 6.5 per cent and services sector by 7.2 per cent during 1980-2003. From the decadewise trends, it is clear that the manufacturing sector has slowed down secularly, while infrastructure and services have accelerated. Agriculture has shown acceleration during the 1990s, but decelerated later. Some analysts attribute this slowing down of the Indian economy during 2000-03 to supply related “infrastructural bottlenecks”, which perhaps is due to deceleration of investment in this crucial sector [see also Shetty 2001 for similar findings].7

    The growth rate in the wholesale price index fluctuated between

    6.6 and 7.8 per cent, which declined to 5.5 per cent during 1993-2003. The rate of inflation declined at differential rates, the most rapid decline (12.7 per cent) being during the 1990s. The decline became slower during 1993-2003. The national income deflator shows similar trends but at 0.5-1 per cent higher levels. Sector specific GDP deflators (proxies for sectoral output prices) also show varying rates of changes, the slowest growth (7.2 per cent) being for manufacturing output price and the most rapid (9.4 per cent) for infrastructure output price. The agricultural output price grew at 9 per cent pa during the entire study period, 1980-2003. The recent decade shows deceleration in these prices as well.

    The real GDP share in agriculture fell from 36.4 per cent in the 1980s to 29.1 per cent in the 1990s and it stood at 26.5 per cent during the recent decade (1993-2003), a sizeable decline of 10 percentage points. The non-agriculture sector exhibits the opposite pattern. Within the non-agriculture sector, share of the services sector is the largest, accounting for more than one-third of the GDP. The share has gone up from 32.3 per cent in the 1980s to 37 per cent in the 1990s and more recently to 38.8 per cent of the GDP. The GDP share of infrastructure remained stagnant around 14-15 per cent, although the GDP level has roughly doubled. TheGDP share of manufacturing sector improved marginally from 17.6 per cent in the 1980s to 19.4 per cent in the 1990s and even subsequently. Thus, there is a structural shift in production from agriculture to infrastructure and services in the Indian economy.

    Investment and Savings

    During 1980-2003, real public investment in agriculture and manufacturing sectors has declined by 2.1 per cent and 0.1 per cent respectively, whereas real public investment in infrastructure and services sectors grew by 3.9 per cent and 3.7 per cent respectively. These investment trends are consistent with the production trends discussed above. The public investment in all sectors put together grew by 2.5 per cent in the study period. In fact, the public investment growth has decelerated from 4.5 per cent during the 1980s to

    2.2 per cent during the 1990s. In the post-liberalisation period, growth is only 1.1 per cent. This is the result of massive disinvestment of public sector units in the country during the post 1990s.

    To a certain extent, private investment has substituted for public investment. Private investment in agriculture, manufacturing, infrastructure and services sectors grew by 4.2 per cent, 6.9 per cent, 5.9 per cent and 6.3 per cent respectively in the entire study period. Private total investment in all sectors grew by 6.3 per cent in the study period. Between the 1980s and 1990s, private investment accelerated in agriculture and manufacturing (substantially) but was nearly stagnant or decelerated in the other two sectors. In the post-1993 period, except in agriculture, private investment slowed down in all the three other sectors. The graphs depicting investment shares also confirm this.

    Nominal gross domestic savings in the economy have been growing at an average rate of 16.2 per cent during 1980-2003, which is 0.6 per cent faster than the growth in nominal gross investment (15.6 per cent). However, both gross domestic savings and investment seem to have decelerated by about 4 per cent

    Year Year
    Saving Rate Invst Rate infrastructure

    Source: Based on the authors’ own calculations from data published in NAS and the Handbook of Statistics on Indian Economy, RBI.

    Year Year

    Infrastructure Services

    50 40 30 20 10 0



    Gross Domestic Saving and Investment Rates GDP Share

    (Per cent) (Per cent)











    18 16












    Public Pub Investment Pvt Investment

    (Per cent) (Per cent)




























































    20 15 10 5 0

    Chart 1: Real Private and Public Sector Investments

    (in Rs ‘000 crore)

    Agriculture Manufacturing


    100 80 60 40 20


    Chart 2: Impact of 20 Per Cent Sustained Increase in Public Investment in Infrastructure on Select Macro Variables

    (in Rs ‘000 crore)

    Real Private Investment in Agriculture Real Private Investment in Manufacturing

    10 15 20


    Real Private Investment in Infrastructure Real Private Investment in Services

    60 55 50 45 40 35 30


    Real Output in Agriculture Real GDP

    1300 1100 900

    Real Output in Infrastructure Gross Domestic Investment



    700 200 150 100

    480 380 280


    Ye ar

    Base simulation Policy simulation Year

    Source: Based on the authors’ own calculations from data published in NAS and the Handbook of Statistics on Indian Economy, RBI.

    pa during the recent decade.8 These trends indicate that there government consumption expenditure, however, accelerated from has been some disillusionment in the investment climate during 15.4 per cent to 16.3 per cent. Therefore, the deceleration in the post-1993 period in India. The reasons could be fall in demand government expenditure can solely be attributed to the deceland recessionary conditions in the Indian economy. eration in investment. These trends continued into the 1993-2003 period as well. Although the nominal government direct tax collection has accelerated, the total revenue seems to have

    Fiscal and Monetary Variables

    decelerated. Some fiscal prudence has led to deceleration in the

    In developing countries, the finances of the government play fiscal deficit over the years. In fact, fiscal deficit decelerated from an important role in the growth of the economy. Government 18.7 per cent in the 1980s to 15.8 per cent in the 1990s. However, total expenditure consists of current and capital expenditures. the government seems to have lost control over fiscal deficit again The nominal total government expenditure has decelerated from during 1993-2003. Money supply grew more or less steadily

    16.2 per cent in the 1980s to 14.1 per cent in the 1990s. The at about 17 per cent during the study period. Nominal interest

    Chart 3: Impact of 20 Per Cent Sustained Increase in Public Investment in Infrastructure on Select Macro Variables

    (In per cent)

    11 10 9 8 7 6

    40 35 30 25

    Nominal Fiscal Deficit Inflation

    9 8 7 6 5 4


    1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 Ye

    Year ar

    Nominal Interest Rate Imports

    14 13 12 11 10 9


    Exports Money Supply


    22 9.5 20 189 16


    14 8


    Year Year

    Rural Poverty Ratio Urban Poverty Ratio

    35 30 25 20




    Base simulation Policy simulation

    Source: Based on the authors’ own calculations from data published in NAS and the Handbook of Statistics on Indian Economy, RBI.

    rate grew marginally during the 1980s by 0.8 per cent pa, but picked up again (10.6 per cent) during 1993-2003. The unit value

    dropped significantly since then and the trend continued. of exports, proxy for export price, has increased slower than export quantity during most of the period except during the 1980s and much slower in the recent decade. The export competitiveness

    External Sector

    was facilitated by significant depreciation of Indian rupee (9.4 per

    Real export growth from the country has accelerated rapidly cent) against the US$, in addition to rise in unit value of exports. from 5.1 per cent in the 1980s to 10.8 per cent in the 1990s, Despite rupee depreciation, growth in real imports has accelerated with an overall growth of 9.5 per cent pa. Exports seems to have very rapidly from 7.3 per cent in the 1980s to 14.7 per cent in the 1990s, mainly due to higher demand. A substantial part of these imports could be oil imports, which have become essential both as inputs and final consumption goods. The import growth however seems to have slowed down to 7.8 per cent during 19932003. The nominal trade balance, as expected, has been negative and highly volatile, particularly during the 1990s and thereafter. The opening up of the economy must have been largely responsible for this.

    Poverty Ratios

    The data on the headcount (poverty) ratios, separately for rural and urban India, are taken from Radhakrishna et al (2004) and Panda (2006). The poverty estimates in these studies are obtained using data from the NSS, which are on calendar-year basis for some years and crop-year (July-June) for others. There are also gaps in the data for some years due to non-existence of NSS rounds. In order to match NSS rounds with NAS time series, a simple average of two adjacent years is used wherever necessary. For the purpose of estimating regressions, the data are interpolated for missing years. We know that this is not a very satisfactory method, but there is no other alternative. The poverty ratios show a declining trend, though with some fluctuations, in both rural and urban areas. The rate of decline also seems to have been slowed down in recent years. The fluctuations are more in the rural poverty estimates. The headcount ratio declined by about 3-5 per cent during the study period.

    In summary, the above trend analysis shows that the macroeconomy has been under severe stress with high volatility and slowing down of investment and economic growth during the mid-1990s and thereafter. However, the infrastructure and services sectors seem to hold some hope. This paper therefore tries to look at the potential of increasing public investment in the infrastructure sector as a vehicle for accelerating economic growth and reaching the much debated 10 per cent GDP growth in India.

    III Counterfactual Policy Simulations

    The estimated model is given in Appendix II. A brief discussion of the estimated model is also given there. Thus, from the signs, magnitudes, t-ratios of the coefficients and goodness of fit measures of all the equations in the model, we infer that there is considerable simultaneity in the macroeconomic relationships and the model is indeed a simultaneous system. Further, due to several endogenous lags, the model is truly dynamic in nature and impacts of any exogenous change will be spread over time. There will be both short- and long-run responses, which will enable us to analyse counter factual policy simulations. A brief discussion of the simulation methodology is also given in Appendix II.

    The main purpose of this paper is to analyse the impacts of hypothetical sustained9change(s) in public sector real investment in the infrastructure sector financed through borrowing from commercial banks. These changes are envisaged to be implemented from the year 1993-94. The policy simulation can be done for any sample period or even post-sample period. Here, the period 1993-94 to 2002-03 is chosen because it covers the first decade Sustained 20 per cent increase in public sector real investment in infrastructure sector financed through borrowing from commercial banks: It is hypothesised that the government will raise the necessary investment resources through borrowing from commercial banks. In this simulation therefore, both the exogenous variables, real public investment in infrastructure (PCFINF) and commercial bank credit to the government (BCG) are increased by 20 per cent of PCFINF each. Since BCG is in nominal terms, the amount of bank credit to the government is

    Table 1: Impacts and Dynamic Multipliers of 20 Per Cent Sustained Increase in Real Public Investment in Infrastructure Financed by Commercial Bank Credit

    Variable B a s e Multiplier (Per Cent)
    Simulation Impact Short- Medium- Long-
    Level* (1993-94) Term Term Term
    (1993-94) (1994-95) (1995-98) (1998
    Real sector
    Nominal income 816.87 -0.12 -0.08 0.27 0.41
    GDP deflator 1.02 -0.17 -0.31 -0.82 -1.41
    Agriculture 1.07 -0.14 -0.09 -0.02 0.01
    Manufacturing 1.00 -0.15 -0.27 -0.74 -1.22
    Infrastructure 1.02 -0.14 -0.78 -2.62 -3.89
    Real income 800.91 0.04 0.23 1.10 1.84
    Agriculture 241.80 0.00 0.00 0.58 1.23
    Manufacturing 150.98 0.19 0.36 0.75 1.03
    Infrastructure 114.83 0.00 0.77 3.16 4.51
    Services 293.31 0.02 0.14 0.84 1.50
    Real private
    investment 108.31 -0.07 2.50 3.60 3.51
    Agriculture 11.20 -0.04 0.70 1.62 1.82
    Manufacturing 46.10 -0.11 2.16 2.73 3.00
    Infrastructure 15.93 -0.07 10.40 8.29 7.31
    Services 35.08 -0.02 0.01 4.18 3.28
    Real private
    consumption 581.48 0.01 0.12 0.70 1.27
    Real personal
    disposable income 727.48 0.01 0.16 0.92 1.58
    Gross domestic
    savings (N) 191.71 3.36 4.06 4.42 3.59
    Gross investment (N) 196.42 3.49 4.33 4.68 3.99
    Headcount ratio-rural
    (per cent) 38.90 0.00 -0.05 -0.33 -0.67
    Headcount ratio-urban
    (per cent) 33.92 0.00 -0.04 -0.29 -0.58
    Fiscal sector
    consumption (N) 94.30 0.04 0.26 1.71 3.34
    Government total
    expenditure(N) 230.52 2.95 2.68 2.66 2.69
    revenue (N) 164.48 -0.03 0.18 0.96 1.35
    Direct taxes (N) 30.96 0.35 1.23 3.80 5.00
    Indirect taxes (N) 100.78 -0.11 -0.07 0.25 0.39
    Non-tax revenue (N) 32.74 -0.12 -0.08 0.28 0.41
    Fiscal deficit (N) 70.95 7.93 7.93 5.61 4.16
    Monetary sector
    Money supply 441.58 -0.09 0.22 0.72 0.92
    Price level 1.01 -0.16 -0.29 -0.79 -1.36
    Rate of inflation
    (per cent)# 8.51 -0.17 -0.15 -0.18 -0.17
    Rate of interest
    (per cent)# 11.78 -0.16 -0.17 -0.22 -0.29
    External sector
    Real exports demand 77.92 0.00 -0.03 -0.22 -0.30
    Real imports demand 89.48 0.49 0.65 0.78 0.74
    Unit value of exports 0.99 0.03 0.05 0.11 0.17
    Exchange rate (N) 30.45 0.03 0.02 -0.17 -0.45
    Trade balance (N)# -12.64 -0.41 -0.70 -1.52 -2.35

    after the implementation of economic reforms and their taking *: Rs ‘000 crore, except for GDP deflators, price level, rate of inflation, rate roots into the economy. The scenario results are presented in of interest, unit value of exports and exchange rate based on data from NAS and the Handbook of Statistics on Indian Economy, RBI.

    Table 1. The simulation impacts for a few important variables

    N: Nominal, i e, current prices.are also plotted (Charts 2 and 3). #: Changes in level.

    expressed in current prices using gross investment deflator (PGKE). Assuming competing needs for money, in other words “liquidity crunch”, the bank credit that was available to the commercial sector earlier (in base simulation) will be lesser in the policy simulation by the amount borrowed by the government for investment in the infrastructure sector. Such a policy will reduce the reserve bank credit to the government and thereby, reserve money and money supply. Changes in money supply will trigger several other changes in the economy. A sustained 20 per cent increase in public real investment in infrastructure,10 envisaged as above, has both short- and long-run effects on all the sectors of the Indian economy. The impacts and the dynamic multipliers are given in Table 1 and graphs comparing baseline and policy simulated values are given in Charts 2 and 3. Impacts: From the estimated model, it can be seen that public investment in infrastructure can affect private investment in that sector only with a one-year lag. This probably is due to gestation lags and delays. However, there is another important channel, namely the monetary (or interest rate) channel, which can bring about crowding-in or crowding-out depending on the sign and magnitude of the coefficient.11 Yet another channel is the output price channel, which is highly significant here. The 20 per cent increase in public investment in infrastructure in 1993-94 increased gross investment (3.5 per cent), thereby savings

    (3.4 per cent) and hence, the nominal interest rate fell (0.2 per cent). Although, this has no direct effect on private investment in infrastructure in 1993-94 due to lagged response, the fall in interest rate has a net negative effect on price of infrastructure goods through the monetary (price) channel and hence, on private investment in infrastructure, implying a very small (0.1 per cent) net crowding-out effect on private investment in that year.

    Similar is the case with private investment in all the three other sectors, wherein the interest rate channel is also present and reinforcing the monetary channel. But, the impacts are smaller is the case of agriculture and services sectors. The aggregate private investment has therefore decreased negligibly (0.1 per cent).

    Further, there are other macroeconomic effects. Due to increased public investment, government expenditure (3.0 per cent) and fiscal deficit (7.9 per cent) will rise. Since the government is envisaged to borrow the required funds from the commercial banks, the government may not require any support from the RBI. In fact, RBI credit to the government has fallen

    (1.3 per cent). This results in decline in reserve money (1.0 per cent), money supply (0.1 per cent) and prices (0.2 per cent).

    Due to one-period lag for the net capital stock variable in the production function for the infrastructure sector, the output will increase only with a lag. Due to an increase in investment, aggregate demand (absorption) in the economy will increase, thereby increasing total output negligibly (0.04 per cent), mainly due to small output growth in manufacturing (0.2 per cent) and services (0.02 per cent) sectors. There will be a small decrease in the GDP deflator (0.2 per cent), leaving a decrease of 0.1 per cent in nominal income. Nominal gross investment seems to increase by 3.5 per cent, exceeding marginally the growth in nominal domestic savings (3.4 per cent), necessitating adjustment with current account balance from the external sector.

    On the fiscal side, the impacts in 1993-94 are small, except for government expenditure and fiscal deficit. Higher public investment will increase government expenditure (3 per cent). Due to decline in nominal income, there will be a small fall in revenue from indirect taxes (0.1 per cent) and non-tax revenue (0.1 per cent) of the government, leaving a large uncovered fiscal deficit (7.9 per cent). Demand (supply) for Indian exports will rise (0.0 per cent – these figures are very small and have been rounded off to one decimal place) due to negligible rise in relative export price (0.03 per cent). Also, real imports into the country will rise (0.5 per cent) due to cheaper import prices and higher absorption. The Indian rupee depreciates marginally (0.03 per cent) against the US$. As expected, nominal trade balance and balance of payments will worsen (0.4 per cent).

    Since the head count ratio is inversely related to per capita real income, the former declines negligibly (0.0 per cent) due to similar increase in the latter in both rural and urban areas in 1993-94, the year of 20 per cent increase in public investment in infrastructure. Thus, growth in income leads to decline in poverty instantaneously, though very small in magnitude.

    Short-run Effects

    The impacts get strengthened by 1994-95 and subsequent years. Due to crowding-in effect, 20 per cent increase in public sector investment in infrastructure in 1993-94 encourages private real investment in infrastructure by 10.4 per cent in 1994-95, a significant positive (lagged) response of private sector. This implies a net (total) elasticity of 0.52 for private investment with respect to public investment in this sector. This value, incidentally, is very close to the partial elasticity (0.48) given in Appendix II, equation 21. Due to increase in real gross (and net) capital stock in infrastructure in 1993-94, there will be increase in infrastructure output (0.8 per cent) this year. It is very interesting to note that private investment responds positively in all the other three sectors of the Indian economy, with lead role from the manufacturing sector (2.2 per cent) followed by agriculture (0.7 per cent) and services (0.0 per cent) in that order.

    The aggregate real private investment is expected to rise by

    2.5 per cent and output (real income) by 0.2 per cent in 1994-95. The nominal income will however decline (0.1 per cent) due to a steeper fall in the GDP deflator (0.3 per cent). This sets in other macroeconomic effects. Prominent among these are increases in government expenditure (2.7 per cent), revenue (0.2 per cent), fiscal deficit (7.9 per cent), money supply (0.2 per cent) and imports (0.6 per cent). Important variables which fell are the GDP deflator and price level (0.3 per cent), real exports (0.0 per cent) and trade balance (0.7 per cent). Growth in gross domestic savings

    (4.1 per cent) continues to lag behind gross investment (4.3 per cent), the gap bridged by current account balance.

    By 1994-95, the decline in poverty gained momentum in both rural and urban areas. Due to larger increase in per capita real income, the headcount (poverty) ratio declined by nearly 0.1 per cent in both the areas. This implies that the percentage decline in poverty is roughly half the percentage increase in aggregate real income (GDP). Long-run effects: As expected, all these effects strengthen further over time (since the policy is a sustained change) and lead to significant and wide spread real benefits to the economy. For example, after 10 years (long-term), real gross capital stock in agricultural sector and thereby real agricultural income is expected to increase by a sizeable 1.2 per cent, real aggregate income by 1.8 per cent, with a moderate increase in money supply

    (0.9 per cent). Therefore, general price level is expected to fall by 1.4 per cent and rate of inflation by 0.2 per cent.

    Real exports will continue to decline (0.3 per cent) and imports will increase (0.7 per cent), resulting in a moderate deterioration in nominal trade balance (2.4 per cent) and balance of payments. The current account balance is also expected to fall by the same extent. The Indian rupee will appreciate by 0.4 per cent against the US $. However, due to significant fall in prices (and GDP deflator), the nominal income increases by only 0.4 per cent.

    Two alternative simulations were also attempted to raise the necessary resources for public investment through utilising (a) the foreign exchange assets (reserves) of the RBI, and (b) the accumulated foreign capital inflows (capital account balance of BOP). The long-run effects of these two scenarios are also found to be quite similar, the second alternative indicating a slightly higher GDP growth (2.0 per cent) and money supply (1.2 per cent).12 Since the required legal apparatus for the utilisation of RBI foreign assets and more so for foreign capital inflows by the government appears not in place yet, probably, it may be easier for the government to borrow the required funds from the commercial banks by selling the conventional government security bonds. Thus, sustained public investment in infrastructure can provide the necessary push to the higher growth path of the Indian economy.

    The approach paper by the Planning Commission for the Eleventh Five-Year Plan documented that the Indian economy had registered an average 7 per cent real GDP growth during the first four years of the current Tenth Five-Year Plan (2002-03 to 2005-06) and indicated the potential for achieving 9 per cent real GDP growth. This study confirms such a scenario provided the necessary infrastructural investments are made. If the more recent estimate of 8 per cent or even higher GDP growth was true and sustainable, then our scenario projection will make it nearly 10 per cent pa. Further, it is interesting to note that in the long run, the headcount (poverty) ratio declined by 0.7 per cent in rural and 0.6 per cent in urban areas of India. This is a very significant result and offers credence to policy initiatives aimed at reducing poverty through economic growth.13

    IV Summary and Conclusions

    This study has analysed the likely macroeconomic effects of changes in public investment in infrastructure in India. The quantified effects include the allocative and dynamic responses of the chosen policy change on important macroeconomic variables relating to four broad sectors – real, fiscal, monetary and external sectors of the Indian economy. The real sector is further decomposed into four sub-sectors, agriculture, manufacturing, infrastructure and services. The sign and magnitude of the effects vary over time – immediate to long run.

    Briefly, the estimated model indicated a significant crowdingin effect between private and public sector investment in all the four sub-sectors of the real economy. This has important consequences for investment/disinvestment policies of the government in each of these sectors. Sustained increase in public investment in infrastructure was found to stimulate sizeable increase in private investment in all the sectors. Such a policy is expected to result in widespread changes in the fiscal and monetary sectors of the economy. Thus, public sector investment in infrastructure has the potential to provide the much-needed push and accelerate the growth process of the Indian economy.

    A 20 per cent sustained increase in public sector investment in infrastructure (about Rs 6,900-7,500 crore pa at 1993-94 prices) will enable the Indian economy to grow at an additional

    1.8 per cent and achieve the much debated 10 per cent aggregate real GDP growth per annum in the medium- to long-run.14

    Appendix I Table 1: Annual Average Compound Growth Rates (Per Cent) of Important Variables Used in the Model

    Variable Annual Compound Growth Rate (Per Cent) during (1980-89) (1990-99) (1980-03) (1993-03)

    Real sector

    Nominal income 13.9 15.2 14.5 12.4

    GDP deflator 8.1 8.5 8.4 6.1 Agriculture 8.1 9.4 9.0 6.5 Manufacturing 6.8 7.1 7.2 5.0 Infrastructure 10.9 9.3 9.4 5.7 Services 7.7 8.1 8.0 6.4

    Real income 5.4 6.2 5.7 6.0 Agriculture 3.0 3.2 3.0 2.2 Manufacturing 7.3 6.9 6.6 5.9 Infrastructure 5.4 6.8 6.5 8.0 Services 7.1 7.9 7.2 7.9

    Real income per capita 3.1 4.1 3.6 4.0 Real private consumption 4.1 5.0 4.5 5.2 Real personal disposable income 6.6 7.0 6.5 7.1 Headcount ratio – rural (per cent) -4.3 -2.7 -2.7 -5.1 Headcount ratio – urban (per cent) -3.1 -4.3 -3.2 -4.7 Gross domestic savings (N) 16.2 15.4 16.2 12.7 Gross investment (N) 16.8 16.1 15.6 11.7

    Fiscal sector

    Government consumption (N) 7.7 6.4 5.9 7.0 Government total expenditure (N) 16.2 14.1 14.3 13.8 Government revenue (N) 15.9 13.6 14.1 12.1 Direct taxes (N) 14.5 18.9 17.2 15.2 Indirect taxes (N) 16.5 12.1 13.4 11.1 Non-tax revenue (N) 14.7 14.2 13.8 12.2 Fiscal deficit (N) 18.7 15.8 15.4 17.2

    Government non-market borrowings (N) 19.1 15.0 14.9 19.3

    Monetary sector

    Money supply 17.3 17.4 17.2 16.6 Price level 6.6 7.8 7.7 5.5 Rate of inflation (per cent) -4.9 -12.7 -3.0 -10.3 Rate of interest (per cent) 0.8 -1.7 -0.8 -7.5 External sector Real exports demand 5.1 10.8 9.5 10.6 Real imports demand 7.3 14.7 9.3 7.8 Unit value of exports 9.7 7.5 9.2 3.6 Exchange rate (N, Rs/$) 7.6 9.1 9.4 5.7 Trade balance (N) # 9.0 26.9 13.2 13.1 Real total investment 4.9 6.0 4.8 1.8

    Public investment 4.5 2.2 2.5 1.1 Agriculture -3.9 -0.1 -2.1 -0.8 Manufacturing 7.3 0.1 -0.1 -4.7 Infrastructure 6.4 1.8 3.9 1.9 Services 3.3 5.1 3.7 3.6 Private investment 5.3 8.2 6.3 2.3 Agriculture 2.6 3.5 4.2 4.8 Manufacturing 6.0 11.7 7.0 0.8 Infrastructure 5.3 5.2 5.9 2.0 Services 5.6 4.8 6.2 4.0

    Real GDP share (per cent)

    Agriculture 36.4 29.1 31.5 26.5 Manufacturing 17.6 19.4 18.6 19.6 Infrastructure 13.7 14.5 14.4 15.1 Services 32.3 37.0 35.5 38.8

    Notes: The annual average compound growth rate is computed using semilogarithmic regression over time for each variable based on data from NAS and the Handbook of Statistics on Indian Economy, RBI.

    N: Nominal, i e, current prices. #: In absolute value.

    Further, such growth is non-inflationary and welfare improving 50 YAR: Real output in agriculture, forestry and fishing (industry group 1 of NAS)

    through higher government revenue and 0.7 per cent reduction

    5 1 YINFR: Real output in infrastructure incl electricity, gas, water

    in poverty in rural and 0.6 per cent in urban areas. The additional supply; construction; transport, storage and communication expenditure is about 0.5 per cent of the GDP and 2.7 per cent (industry groups 4, 5 and 7 of NAS) of the government total revenue in 2002-03. We believe 5 2 YM: Gross domestic product at market prices (nominal) 5 3 YMNR: Real output in manufacturing including mining and quarrying

    that such investment is quite feasible and cost effective.

    (industry groups 2 and 3 of NAS)An alternative simulation wherein the government utilises 5 4 YNAR: Real output in non-agriculture sector (=YMNR+YINFR+YSRR) accumulated capital inflows instead of borrowing from com-55 YSRR: Real output in services including all others (industry groups 6,

    8 and 9 of NAS)

    mercial banks, gave similar results, with few changes in external

    56 YR: Real output at factor cost

    and monetary sectors. It must be mentioned that the major limitation of the study is its aggregative nature – both sectoral Exogenous variables and spatial (all India). A more disaggregated model may give 1 AREA: Index of gross cropped area (1993-94=1.0)

    2 BCG: Commercial bank credit to government (nominal)

    better insights into the process of the working of the Indian

    3 DNB: Non-market borrowings of both central and state


    governments (nominal) 4 CAPB: Net capital account in the balance of payments incl. errors and omissions (nominal)

    Appendix II

    5 CAPTR: Capital transfers to government (nominal)

    Description of Variables Used in the Analysis

    6 DUMMY1: Dummy for sharp increase in output of Infrastructure (1993-98)

    (Rs ‘000 crore)

    7 DUMMY2: Dummy for post-reform period (1991-92 onwards) 8 DUMMY3: Dummy for sharp decline in Inflation (post 1990s)

    Endogenous variables

    9 DUMMY4: Dummy for sharp increase in exports (1999 onwards)

    1 ABSP: Real private absorption

    10 DUMMY5: Dummy for sharp increases in private investment in

    2 AD: Real aggregate absorption

    manufacturing sector

    3 ADD: Real aggregate demand for domestically produced goods

    1 1 DUMMY6: Dummy for sharp increases in gross fiscal deficit (1998

    4 BCP: Bank credit to commercial sector (nominal)


    5 BOP: Balance of payments (nominal)

    1 2 EB: External borrowings by the governmenmt (nominal)

    6 CAB: Current account balance (nominal)

    13 INVISB: Invisibles in current account balance (nominal)

    7 CONS: Real government consumption expenditure

    14 GCL: Government current liabilities to the public (nominal)

    8 DEPAG: Real depreciation in agriculture

    15 MISCR: Other components of RBI credit to government (nominal)

    9 DEPINF: Real depreciation in infrastructure

    16 MISL: Miscellaneous components of reserve money (nominal)

    1 0 DEPMN: Real depreciation in manufacturing

    17 PCFAG: Real gross public investment in agriculture

    11 DEPSR: Real depreciation in services

    18 PCFINF: Real gross public investment in infrastructure

    12 DIT: Indirect taxes of both central and state govts (nominal)

    1 9 PCFMN: Real gross public investment in manufacturing

    13 DT: Direct taxes of both central and state govts (nominal)

    20 PCFSR: Real gross public investment in services

    1 4 EXPT: Real exports

    21 PCFSAV: Gross public sector savings (nominal)

    1 5 EXR: Exchange rate against US $ (nominal, Rs/$)

    22 PGKE: Gross investment deflator (1993-94=1.0)

    1 6 FD: Gross fiscal deficit of both central and state govts (nominal)

    23 PRSR: Price deflator for services including all others (industry

    17 GCFADJ: Gross domestic capital formation, adjusted series (nominal)

    groups 6, 8 and 9 of NAS)

    18 GXP: Total expenditure of both central and state govts (nominal)

    24 PYDIFF: Difference between income at market prices and factor cost

    19 HCRRUR: Headcount ratio in rural areas (per cent)


    20 HCRURB: Head count ratio in urban areas (per cent)

    2 5 RAIN: Annual rainfall (mm)

    2 1 IB: Nominal interest rate (per cent) on three-year bank deposits

    26 RBCS: RBI credit to the commercial sector (nominal)

    22 IMPT: Real imports

    27 RBFA: Net foreign exchange assets of RBI (nominal)

    23 KAGR: Real net capital stock in agriculture

    28 RES: Residual components of Bank credit to commercial sector

    2 4 KMNR: Real net capital stock in manufacturing

    2 9 RNML: RBI’s net non-monitory liabilities (nominal)

    25 KINFR: Real net capital stock in infrastructure

    30 TRP: Transfer payments (nominal)

    26 KSRR: Real net capital stock in services

    31 UVII: Unit value of imports (1993-94=1.0)

    27 M3: Money supply (nominal)

    32 WPEXP: World price index (1993-94=1.0)

    28 NTX: Non-tax revenue of both central and state govts (nominal)

    33 WYR: Real world income

    2 9 P : Wholesale price index (1993-94=1.0)

    3 4 NTOT: Aggregate population (millions)

    30 PC: Real private consumption

    3 5 TREND: Time trend variable with its value as unity for 1978-79.

    31 PCFTOT: Real aggregate public investment 32 PITOT: Real aggregate private investment 33 PGDP: GDP deflator (1993-94=1.0)

    Estimated model 3 4 PIAG: Real gross private investment in agriculture

    The proposed macroeconometric model consists of 4 blocks – real, fiscal, 3 5 PIINF: Real gross private investment in infrastructure monetary and external sectors. It has 56 endogenous variables (31 equations 3 6 PIMN: Real gross private investment in manufacturing and 25 identities) and 35 exogenous variables. For convenience of estimation 37 PISR: Real gross private investment in services and future improvements, the model is estimated in three separate modules (I, 38 PRAG: Price deflator for agriculture, forestry and fishing (industry II and III) using three stage least squares (3SLS) method for each module.

    group 1 of NAS) The module I contains all the macroeconomic relationships except the real 39 PRINF: Price deflator for infrastructure incl electricity, gas, water sector equations, which are put into module II. Module III has only two supply; construction; transport, storage and communication equations representing rural and urban head count (poverty) ratios. Due to (industry groups 4, 5 and 7 of NAS) lags and use of rate of change in some variables, the actual estimation uses 40 PRMN: Price deflator for manufacturing including mining and data for 1981-82 to 2002-03. quarrying (industry groups 2 and 3 of NAS) While estimating the model, a TREND variable is included in some equations 41 PYDR: Real personal disposable income to capture the autonomous time related changes in the endogenous variables. 42 PYD: Personal disposable income (nominal) Dummy variables are included in the model to separate the pre- and post43 RCG: Reserve bank credit to the government (nominal) liberalisation (1991-92 onwards) effects (Dummy 2) and also to capture the 44 RM: Reserve money (nominal) abnormal fluctuations in the data for certain variables (Dummy 1, Dummy 3, 45 SAV: Gross domestic savings (nominal) Dummy 4, Dummy 5 and Dummy 6). The choice of the equations was guided 46 TB: Trade balance (nominal) by expected sign as well as statistical significance for the coefficients and high 47 TR: Current revenue of both central and state govts (nominal) goodness-of-fit, including absence of serial correlation for residuals. It may be 48 UVIX: Unit value of exports (1993-94=1.0) mentioned that the choice of lag length for various determinants was also 49 Y: Output at factor cost (nominal) guided by expected sign and significance. It involved careful search process.

    The finally selected model is given below:

    Estimated model: Period: 1981-82 to 2002-03 Method: 3SLS

    Module I:

    Fiscal sector:

    1 DT = -36.518 + 0.177 YNAR – 31.083 P (-15.96) (8.85, 2.31) (-2.78, -0.71) ⎯R2= 0.98DW = 0.49 2 DIT = 11.358 + 0.109 Y + 0.400 AR (1)

    (4.71) (52.85, 0.94) (3.15) ⎯R2= 0.99DW = 1.78 3 NTX = 0.036 YM + 0.357 AR (1) (55.26, 1.01) (2.52) ⎯R2= 0.99DW = 1.96 4 CONS = 0.119 (YM/P) + 0.407 CONS-1 + 1.075 AR (1) (6.80, 1.06) (4.58) (55.78)

    ⎯R2= 0.99DW = 2.06 5 PC = 164.923 + 0.573 PYDR + 0.594 AR (1)

    (22.68) (65.11, 0.77) (4.92) ⎯R2= 0.99DW = 1.71

    6 FD = 0.817 GXP – 0.786 TR – 16.904 ((P-P-1)/P-1) (39.30, 2.60) (25.10, -1.62) (-2.13, -0.01)

    + 8.537 DUMMY6 – 0.610 AR(2)

    (5.41) (-5.10)

    ⎯R2= 0.99DW = 1.88

    Monetary sector:

    7 P = -0.0002 YR+0.000098 M3 +0.009 IB + 0.017 TREND + 0.823 P-1

    (-3.00, -0.18) (2.53, 0.07) (3.43, 0.07) (4.81) (12.04)

    + 0.348 AR (1)

    (3.04) ⎯R2= 0.99 DW = 2.33 8M3 = -67.265 + 0.297 RM – 17.321 TREND + 1.122 AR (1) (-2.02) (2.04, 0.08) (-3.19) (123.24)

    ⎯R2= 0.99 DW = 2.56 9 IB = 7.901+0.018 (Δ(BCP) +CAPB) – 6.410 ((P-P-1)/P-1) – 0.010 SAV

    (16.55) (2.82, 0.19) (-2.31, -0.04) (-4.52, -0.40)

    + 2.559DUMMY2 + 2.861DUMMY4

    (9.34) (9.12)

    ⎯R2= 0.86 DW = 1.40 External sector: 1 0 EXPT = 190.908 – 3453.455 (UVIX/EXR/WPEXP) + 0.0004 WYR

    (9.21) (-7.50, -0.67) (3.84, 0.30) -33.227 DUMMY4 (-5.85)

    ⎯R2= 0.93 DW = 1.04 11 UVIX = – 4.867 (P/EXR)+9.14E-06 WYR+0.560 WPEXP –0.002 EXPT-1 (-8.16, -0.14) (22.77, 0.69) (17.07, 0.60) (-5.09) ⎯R2= 0.99 DW = 1.96 1 2 IMPT = 0.082 AD – 2.270 (UVII*EXR/P) + 0.316 (TREND*TREND) (3.98, 0.58) (-5.03, -0.55) (5.75)

    + 0.821 AR (1)

    (7.77) ⎯R2= 0.99 DW = 1.20 13 EXR = 21.700 P – 0.107 CAB + 0.141Δ RBFA + 4.006 DUMMY2 (30.14, 0.75) (-3.31, 0.02) (6.23, 0.12) (7.61)

    ⎯R2= 0.99 DW = 1.73

    Link equation:

    1 4 PGDP = -0.058 + 1.063 P + 0.616 AR (1) (-2.80) (57.48, 1.04) (5.98) ⎯R2= 0.99 DW = 1.50

    Module II:

    Real sector:

    15 YAR = -261.002 + 0.035 RAIN + 222.162 AREA + 0.664 KAGR-1

    (-1646) (5.73, 0.11) (10.54, 0.81) (21.08, 0.84) +0.360 YINFR-1 (12.61, 0.19)

    – 0.501 AR (1) (-5.18) ⎯R2= 0.99 DW = 2.06

    1 6 YMNR = 0.045 ADD + 0.072 KMNR + 4.202 TREND + 0.622 AR (1) (2.59, 0.27) (3.12, 0.32) (3.92) (5.38) ⎯R2= 0.99 DW = 1.45 1 7 YINFR = 0.132 KIFNR-1 + 0.153 (M3-1/P-1) + 0.812 AR(1) (12.85, 0.41) (21.17, 0.59) (13.37) ⎯R2= 0.99 DW = 2.09 1 8 YSRR = -156.810 + 0.462 KSRR-1 + 0.171 (M3/P) + 0.559 AR(1) (-5.69) (8.13, 1.09) (4.21, 0.28) (4.72) ⎯R2= 0.99 DW = 1.91

    19 PIAG = 0.029YAR-1+0.136 PCFAG-1+ 0.060 PIINF-1 (6.66, 0.57) (1.60, 0.05) (2.28, 0.09) –0.054(IB-((P-P-1)*100/P-1)) (-1.89, -0.01) –0.012 (PCFSAV-1/PGKE-1 – PCFTOT-1) +2.26 PRAG (-2.63, 07) (3.63, 0.24) ⎯R2= 0.95 DW = 2.95 20 PIMN = –46.941+3.075 PCFMN+0.979 PIINF–1.275 (IB-((P-P-1)*100/P-1)) (-9.33) (11.76, 0.50) (2.65, 0.22) (-4.40, -0.05) +27.333 PRMN –10.398 DUMMY5+0.513 PIMN-1 – 0.718 AR (1) (4.51, 0.40) (-7.19) (8.20) (-7.16) ⎯R2= 0.90 DW = 2.12 2 1 PIINF = 0.279 PCFINF-1 + 7.445 PRINF (3.68, 0.48) (3.46, 0.51) ⎯R2= 0.74 DW = 1.74 22 PISR = 0.711 PCFSR+0.994 PIINF-1– 0.716 (IB-((P-P-1)*100/P-1)) (5.18, 0.36) (3.79, 0.44) (-2.84, -0.06) +7.739 PRSR (1.69, 0.24) ⎯R2= 0.90 DW = 1.50 2 3 DEPAG= -7.128 + 0.057 KAGR-1 (-2.66) (6.56, 1.57) ⎯R2= 0.61 DW = 1.59 24 DEPMN= 15.516 + 0.037 KMNR-1 + 0.175 AR (1)

    (4.75) (7.64, 0.68) (5.01) ⎯R2= 0.74 DW = 2.06 2 5 DEPINF= – 4.362 + 0.079 KINFR-1 – 0.146 AR (1) (-3.89) (28.02, 1.12) (-2.18) ⎯R2= 0.95 DW = 1.89 26 DEPSR= –5.960 + 0.034 KSRR-1 (-3.62) (16.69, 1.24) ⎯R2= 0.90 DW = 1.23 2 7 PRAG= -0.003 YAR + 0.002 PYDR + 1.033 P + 1.094 AR (1) (-6.93, -0.57) (6.64, 1.22) (6.54, 0.98) (78.82)

    ⎯R2= 0.99 DW = 1.84 28 PRMN = 0.070 + 0.916 P + 0.502 AR (1)

    (6.69) (97.83, 0.95) (4.90) ⎯R2= 0.99 DW = 1.76 2 9 PRINF = -0.170 – 0.008 YINFR + 0.001 PYDR + 0.988 P (-15.57) (-14.30, -0.89) (14.93, 1.04) (27.76, 0.97)

    ⎯R2= 0.99 DW = 2.08 Module III: Poverty ratios: 30 HCRRUR= 64.760 – 36.774 (PYDR/NTOT) + 4.132 DUMMY2 + 0.425 AR (1)

    (19.27) (-8.06, -1.13) (2.48) (2.40)

    ⎯R2= 0.89 DW = 2.32 31 HCRURB= 59.998 – 31.975 (PYDR/NTOT)

    (51.46) (-22.74, -1.14) ⎯R2= 0.96 DW = 1.64

    Note: The t-ratios are given in parenthesis. For important variables, the shortrun mean partial elasticity is also given adjacent to the t-ratio.

    A perusal of the estimated model indicates that the model is estimated quite well. Almost all the regression coefficients, except a few (four to be precise out of 124 coefficients) are significant at 5 per cent or less. The signs of the coefficients also look appropriate, a priori. However, despite our best efforts, some of the equations still seem to suffer from serial correlation. In order to understand the direction and relative magnitude of response of each determinant on the dependent variable, the estimated mean partial elasticities are also given in each equation. It is important to note however that the direction and size of response implied by these mean partial elasticities is only indicative and the net impacts measured through policy simulations are likely to be different from these mean partial elasticities. For this reason, the interpretation of the individual coefficients may be of less importance except making few observations on the implied incremental capital-output ratios (ICOR) for different sectors and the direction of association between some important variables in the model.

    From the coefficient of the net capital stock variable in the agricultural production function, the implied ICOR in agriculture is low at 1.5. Thus, there exists significant (nearly unitary) output response in Indian agriculture with respect to capital stock. It is interesting to note that there is a significant complementarity between outputs of agriculture and infrastructure, the latter acting as an essential input to the former. The other two sectors, manufacturing and services, do not exhibit this feature. In the manufacturing sector, which includes mining and quarrying, the implied ICOR is very high at 13.8, indicative of low productivity of capital or high capital intensity.15 For the infrastructure sector, the implied ICOR of 7.6 is somewhat high and reflect the relatively high capital intensity of this sector. The implied ICOR of the services sector is low at 2.2. The real balances (credit) variable seems to play an important and positive role in the production of both infrastructure and services sectors. This confirms our main proposition that changes in money supply affects both output and prices, which in turn causes several macroeconomic effects.

    In all the four sectors, the public investment variable has a positive coefficient in the respective private investment equations and sets the stage for crowding-in effect between public and private investments. The resourcegap variable also seems to contribute to this phenomenon in the agriculture sector alone. With the exception of infrastructure sector, rather surprisingly, the real interest rate (current or lagged) seems to be significant despite it being regulated by the central bank until recently. It is interesting to notice significant cross complementarity between private investments in infrastructure and all the other three sectors as well. This is contrary to the belief that private sector is less enthusiastic in investing in infrastructure and expects the government to invest first.

    From the estimated general price equation, with increase in money stock (and also interest rate), the whole sale price index will go up by a negligible percentage. An increase in real aggregate output, ceteris paribus, will decrease the whole sale price index by a small magnitude. Assuming demand-supply equilibrium (market clearance)16 for three sectoral outputs namely agriculture, manufacturing and infrastructure, we use inverse demand functions to estimate their output prices. Based on these, sectoral outputs of both agriculture and infrastructure seem to exert a larger pressure on their respective output prices (deflators). A surprising exception is the manufacturing sector, where output has no significant effect on the sectoral price. Perhaps, mark-up pricing, rather than demand-supply, may be appropriate for this sector. For all the above three sectoral prices, wholesale price index exhibits (positive) near unitary elasticity; while real personal disposable income has elastic positive response for only agriculture and infrastructure sectors. Manufacturing price seems independent of real personal disposable income, again a bit puzzling.

    Government nominal revenue from direct taxes and indirect taxes as well as non-tax revenue seem to increase with income. Government consumption expenditure also increases with income. The export, import demand functions and nominal bilateral exchange rate equations have expected signs for their determinants.

    As expected, the headcount (poverty) ratio is inversely related to per capita real income in both rural and urban areas. This seems to be the broad linkage between economic growth and poverty reduction. It underlies the familiar “trickle down” hypothesis, with all its limitations. In reality, the nature and extent of (absolute) poverty depends on several socio-economic factors, real income being only one of them.

    Simulation methodology

    To assess the empirical adequacy of the full model in describing the historical data, EViews package was employed to solve the 56 relations together iteratively for each year with deterministic simulation and dynamic solving options for the entire sample period, 1981-82 to 2002-03. The simulated values for the above period are also called the “base simulation” values. Assessment of the full model was done by (a) comparing the time series plots of actual and base simulation values, and (b) computing the summary measures, mean absolute percentage error (MAPE) and root mean square percentage error (RMPE). Based on all these three criteria, the base simulation was found to trace the historical data quite well. Due to limited space, these details are omitted here.

    The allocative and dynamic effects due to the above exogenous/policy change are quantified as percentage changes, also known as multipliers, with reference to base simulation values. They are reported only at four points of time, namely response in the same year of exogenous change (immediate or instantaneous or impact), response after one year (short-term), response between three and five years (medium term) and response between six and ten years (long-term). Since the responses change each year rather slowly, the medium-term and the long-term responses are simple averages of the respective time periods. In the case of headcount ratio, rate of inflation, rate of interest and trade balance, the impacts are changes in level, not rates of change. It may be mentioned that these percentage responses are contemporaneous in nature (policy simulation vs. base simulation) and should not be treated as usual percentage rate of change over time. These responses therefore are likely to be different from the direct responses (both partial and net) implied by the estimated equations. The results of counterfactual simulations are discussed in the main Section III.




    [This paper is an abridged version of a working paper with the same title, which is available at A copy of the working paper can also be obtained from the authors on request. This research is partly funded by the UGC Special Grant called ‘University with Potential for Excellence’ being implemented at the Department of Economics, University of Hyderabad. The authors would like to thank R Radhakrishna and K Krishnamurty for their helpful discussions while formulating the model. The helpful comments of D Narasimha Reddy have also improved the content. S L Shetty and D Anjaneyulu also gave a lot of insights into the data generation mechanism of the NAS and the RBI on certain variables. The comments of the journal referee are also gratefully acknowledged. The usual disclaimer remains.]

    1 While browsing through the literature on the infrastructure sector in India, we came across the most comprehensive study by the “expert group on the commercialisation of infrastructure projects” with Rakesh Mohan as chairman. Their report, submitted in 1996, has examined in detail all aspects of this most crucial sector and made policy recommendations relating to its commercialisation, role of capital markets, the necessary regulatory framework, fiscal reform, and sub-sector specific issues for the post-reform period. We are thankful to Dinesh Singh for bringing this reference to our notice.

    2 A comprehensive review of macroeconometric models and policy modelling for India can be found in Krishnamurty (2001), Pandit and Krishnamurty (2004) and Bhattacharya and Kar (2005).

    3 A good review of monetary sector models was provided by Jadhav (1990).

    4 The explanatory variables given in each equation are those actually found to be empirically suitable after a careful search process during estimation. It is therefore more appropriate to call the given model as “selected model” instead of “proposed model”.

    5 The underlying equations are somewhat modified production functions in the sense that some other related variables, viz, infrastructure output appears as “intermediate input” in the production of agriculture, while the aggregate demand variable is included in manufacturing sector.

    6 Due to volatility in the data for certain variables, the compound growth rates for the sub-periods do not match well with that of the entire period. To avoid this, some analysts recommend smoothing of the series using a moving average method before computing growth rates. This has not been done here.

    7 Perhaps anticipating this, the “expert group” has made projections of yearly investment requirements during 1996-2006 in order to achieve the desired 8.5 per cent GDP growth in India by 2005-06. The required total investment in infrastructure over the 10-year period 1996-2006 is estimated at Rs 7,50,000 crore, with a break-up of 85 per cent from domestic and 15 per cent external sources. The share of infrastructure investment in GDP is projected to increase from 5.5 per cent in 1994-95 to 7 per cent in 2000-01 and 8 per cent in 2005-06. However, in retrospect, we notice that the share of infrastructure investment (out of GDP) declined to merely

    3.5 per cent in 2002-03. The desired GDP growth seems to have been achieved despite this decline.

    8 Notwithstanding this deceleration in domestic savings (and investment), there are serious criticisms about the overestimation of the rate of domestic savings during recent years by the CSO [Shetty 2005, 2006]. Shetty puts the extent of overestimation in the savings rate around 3-4 per cent during 2000-03.

    9 Some analysts prefer to hypothesise one-period or shock-type exogenous changes. If the underlying estimated model is dynamically stable, the impacts of any one-period exogenous change should decay over time and all the endogenous variables return to base simulation levels. In other words, shock-type simulations are inappropriate for studying long-term policy effects. The present model confirmed this property.

    10 This constitutes Rs 6,927 crore in 1993-94 and Rs 7,494 crore in 200203 at 1993-94 prices. These expenditures, in nominal terms, are 4.5 per cent and 2.7 per cent of government total revenue; 0.9 per cent and 0.5 per cent of GDP in respective years. From past experience, during 199303, both public and private investments in infrastructure have grown at 2 per cent pa. The average investment growth was higher at 3.9 per cent and 5.9 per cent during 1980-2003 in the public and private sectors. The investment projections in infrastructure made by the “expert group” for public and private sectors are much higher than what we are postulating.

    However, some analysts [Sastry et al 2003] believe that sustained public investment may not be possible under the present circumstances of resource crunch in the economy.

    11 But, rather surprisingly, the real interest rate channel is inoperative only for this infrastructure sector as the variable dropped out of the private investment equation due to statistical insignificance though it had correct (negative) sign.

    12 Due to space limitation, the details are not included here. Interested readers can refer to the working paper at or write to the authors for a hard copy.

    13 Some recent studies [Himanshu 2006] aimed at decomposing the rate of decline in poverty into growth, inequality and population components indicate that the economic growth is the largest contributor to decline in poverty in urban India. However, some other analysts [Panda 2006] argue that growth may only be a necessary but not sufficient condition for poverty reduction.

    14 Shetty 2001 suggests that the banking system can provide additional resources to the extent of Rs 15,000-16,000 crore pa for infrastructure development in specific projects without causing inflation.

    15 The estimated coefficient (and hence the ICOR) of the net capital stock variable in the production of manufacturing sector seems some what sensitive to the inclusion of time trend variable in the regression. For manufacturing sector, the ICOR is inflated due to inclusion of time trend variable. In general, the time trend variable in a production function is expected to account for all omitted variables including changes in technology.

    16 It may be mentioned that the output equations are “production”, but not “supply” functions and therefore it is not conventional “market equilibrium”. Further, the output price of services sector is assumed exogenous to the model keeping in mind the increasing share of IT, banking, insurance and other services after globalisation. There exists large heterogeneity in the constituents of this sector and endogenous determination of its price in a simple demand-supply framework may be difficult to justify.


    Bhattacharya, B B and S Kar (2005): ‘Shocks and Long Run Growth in Agriculture: A Macroeconomic Analysis’, Development Planning Centre, Institute of Economic Growth, University Enclave, Delhi.

    Himanshu (2006): ‘Urban Poverty in India by Size Class of Towns: Level, Trends and Characteristics’, paper presented at the seminar ‘The Multidimensions of Urban Poverty in India’, Indira Gandhi Institute of Development Research, Mumbai.

    IEG-DSE Research Team (1999): ‘Policies for Stability and Growth: Experiments with a Large and Comprehensive Structural Model for India’, Journal of Quantitative Economics, Special Issue on Policy Modelling, Vol 15, pp 25-109.

    Jadhav, N (1990): ‘Monetary Modelling of the Indian Economy: A Survey, Reserve Bank of India’, Occasional Paper, Vol 11 (2), June.

    Krishnamurty, K (2001): ‘Macroeconometric Models for India: Past, Present and Prospects’, Economic and Political Weekly, October 19.

    Krishnamurty, K and V N Pandit (1985): Macroeconometric Modelling of the Indian Economy, Hindustan Publishing House, Delhi.

  • (1996): ‘Exchange Rate, Tariffs and Trade Flows: Alternative Policy Scenarios for India’, Indian Economic Review, Vol 31, pp 57-89.
  • Murty, K N and A Soumya (2006a): ‘Macroeconomic Effects of Public Investment in Infrastructure in India’, Indira Gandhi Institute of Development Research, WP-2006-003, Mumbai (web://
  • (2006b): ‘Effects of Public Investment in Infrastructure on Growth and Poverty in India’, Indira Gandhi Institute of Development Research, WP-2006, Mumbai (web://
  • Panda, M (2006): ‘Poverty Reduction and Macroeconomic Developments in India: Growth and Equity Considerations’, paper presented at the silver jubilee seminar of Centre for Economic and Social Studies, Hyderabad.

    Pandit, V N and K Krishnamurty (2004): Economic Policy Modelling for India, Oxford University Press.

    Radhakrishna, R et al (2004): ‘Chronic Poverty and Malnutrition in 1990s’, Economic and Political Weekly, Vol 39, No 28, pp 3121-30.

    Rakesh Mohan (1996): ‘The India Infrastructure Report: Policy Imperatives for Growth and Welfare’, report of the expert group on the commercialisation of infrastructure projects, Vols 1-3, NCAER, New Delhi.

    Rangarajan, C (2000): ‘Monetary Policy: Objectives and Targets’ in C Rangarajan (ed), Perspectives on Indian Economy, Academic Foundation, New Delhi.

    Rangarajan, C and M S Mohanty (1997): ‘Fiscal Deficit, External Balance and Monetary Growth: A Study of Indian Economy’, Reserve Bank of India, Occasional Papers, reprinted in A Vasudevan (ed), Money and Banking: Select Research Papers by the Economists of the Reserve Bank of India, Academic Foundation, New Delhi, 2003.

    Sastry, D V S et al (2003): ‘Sectoral Linkages and Growth Prospects: Reflections on the Indian Economy’, Economic and Political Weekly, June 14, pp 2390-97.

    Shetty, S L (2001): ‘Reviving the Economy: Some Explorations’, Economic and Political Weekly, July 28, pp 606-10.

  • (2005): ‘Savings and Investment Estimates: Time to Take a Fresh Look’, Economic and Political Weekly, February 12, pp 606-10.
  • (2006): ‘Revision of the National Accounts Statistics: A Welcome Step, Good in Parts’, Economic and Political Weekly, June 10, pp 2317-26.


    Infant Survival: A Political Challenge —Shantha Sinha Decentralised Childcare Services: The SEWA Experience —Mirai Chatterjee Food Dole or Health, Nutrition and Development Programme? —Shanti Ghosh Infant and Young Child Feeding: An ‘Optimal’ Approach —Arun Gupta Hidden Hunger: The Problem and Possible Interventions —Tara Gopaldas Universalisation of ICDS and Community Health Worker Programmes: Lessons from Chhattisgarh —T Sundararaman Implementation of ICDS in Bihar and Jharkhand —Nandini Nayak, Naresh C Saxena Tamil Nadu: ICDS with a Difference —Anuradha Khati Rajivan Rethinking ICDS: A Rights Based Perspective —Dipa Sinha Chhattisgarh: Grassroot Mobilisation for Children’s Nutrition Rights —Samir Garg

    For copies write to Circulation Manager

    Economic and Political Weekly

    Hitkari House, 284, Shahid Bhagatsingh Road, Mumbai 400 001 email:

    Dear Reader,

    To continue reading, become a subscriber.

    Explore our attractive subscription offers.

    Click here

    Back to Top