is india’s public finance unsustainable? or, are the claims exaggerated?

Journal of Policy Modeling26 (2004) 401–420

Is India’s public finance unsustainable?Or, are the claims exaggerated?�

Rajan Goyal, J.K. Khundrakpam, Partha Ray∗

Department of Economic Analysis and Policy, Central Office,Reserve Bank of India, Fort, Mumbai 400001, India

Available online 26 April 2004

Abstract

Growing concerns about the sustainability of the fiscal policy in India have evokedserious concerns in the recent past. The present paper has assessed the Indian fiscal trendsin terms of inter-temporal budget constraint (IBC) for the Central and State Governmentsseparately and together by employing Gregory and Hansen [J. Econometrics 70 (1996)99] tests of co-integration with structural breaks. By addressing the issue of regime shift,this paper finds that while the fiscal stance of the Central and the State Government at theindividual level is unsustainable, it is weakly sustainable for the combined finances as itnets out inter-governmental financial flows. Thus, claims about sustainability of India’spublic finance, made on the basis of the assessment of individual finances and neglectinginter-governmental flows and the possibility of regime shifts seems exaggerated.© 2004 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved.

JEL classification: H620; H690; H770

Keywords: Deficit; Debt; Federalism; Revenue; Expenditure

1. Introduction

In the recent times, the issue of sustainability of public finance has drawn con-siderable attention. Significantly, this concern for fiscal sustainability is spread

� The views expressed in this paper are those of the authors and not of the Institution to which theybelong.

∗ Corresponding author. Tel.:+91-22-2266-0529; fax:+91-22-2263-3186.E-mail address: [email protected] (P. Ray).

0161-8938/$ – see front matter © 2004 Society for Policy Modeling.doi:10.1016/j.jpolmod.2004.03.004

402 R. Goyal et al. / Journal of Policy Modeling 26 (2004) 401–420

over both the developed and developing world. For example, while Maastrichttreaty required governments not to run budget deficit and debt–GDP ratio beyonda point (3% and 60% of GDP, respectively) as a precondition to enter the EuropeanMonetary Union, the recently passed Fiscal Responsibility and Budget Manage-ment Bill of India has envisaged complete elimination of revenue deficit in themedium term.

For a federal country like India having 28 States and 2 Union Territories (UTs),with the State Governments having independent executive, legislative and judicialwings, the issue of sustainability of public finance has an added dimension. Duringthe 1980s and 1990s, there has been a steep rise in the outstanding liabilities of theGovernment at both national and sub-national level. The liabilities of the CentralGovernment as a proportion to GDP increased from 41.4% as at end-March 1981to about 60% as at end-March 2002. Similarly the aggregate liabilities of the StateGovernments increased from 16.6% to 25.0% during this period.

These numbers indeed seem to be alarming, and in the recent times, seriousconcerns have been expressed about the sustainability of India’s public debt atvarious fora. Reports from the Comptroller and Auditor General of India, a statu-tory body that audits the finances of both the Central and the State Governments,expressed concern about the level of sustainability of the Central Government fi-nances. More recently, a World Bank Report found that the fiscal deficit of thegeneral government (i.e., Centre plus States) averaging over 9% over the past 6years to be alarming. This Report also found that about 60% of this deficit is at theCentre and 40% at the State level, where much of the recent fiscal situation hasoccurred (World Bank, 2003).

There are, thus, two sources to this crisis viz., the Central Government and theState Governments, and efforts to look into the empirics of debt sustainability inIndia by concentrating only on Central Government finances are at best partialand hence faulty. In this light, we, in the present paper, have looked into theramifications of the fiscal scenario from three standpoints, viz., Centre, States andcombined.

How do we measure sustainability? The term fiscal or debt sustainability per-haps implies a set of fiscal policies that could be continued unaltered withoutjeopardising the economic policy objectives such as economic growth, price sta-bility and external balance. Traditionally, the ability of the government to maintainits fiscal policies is measured in terms of debt–GDP ratio and a given fiscal stanceis considered sustainable if debt–GDP ratio does not grow to explosive proportionsover time. There are, however, alternative approaches to test the sustainability ofdebt. One approach, which is sometimes referred to as “Accounting Approach”,is that of satisfying the steady-state Domar condition in which rate of growthof income must exceed the interest rate on public debt, subject to the conditionthat primary balance is either positive or zero. As is evident, in this approach,income growth–interest rate differential is the key element for ensuring debt orfiscal sustainability. In case of rate of GDP growth less than the interest rate, freshborrowings due to rising debt-service obligations would grow more rapidly than

R. Goyal et al. / Journal of Policy Modeling 26 (2004) 401–420 403

the growth of GDP that the fiscal stance would be unsustainable as debt–GDPratio would be on an explosive growth path. Measuring the sustainability of deficitfrom the Domar condition could be naı̈ve; after all, a debt–GDP ratio might bestable at 200% — however, this may be quite unsustainable. In order to get rid ofsuch incredible outcomes, the standard approach in the literature is measuring theinter-temporal budget constraint (IBC).

In the IBC approach, a sustainable debt would require not only the debt–GDPratio to remain stable in the future, but it must also be ensured that the outstandingdebt is finally repaid. Formally, inter-temporal budgetary constraint requires thatdiscounted value of the expected future surpluses generated by the governmentare sufficient to repay the outstanding stock of government debt. Equivalently, thecondition requires that the present value of the government debt stock tends to zeroover time. Since, the IBC approach requires that the debt must be finally repaid, itis also termed as solvency condition for a sustainable debt. In econometric terms,the sustainability of public finance, thus, gets translated in terms of comovementbetween revenue and expenditure. In other words, non-existence of a co-integratedrelationship between revenue and expenditure could be interpreted as lack of sus-tainability. It is, however, well known that the powers of these co-integration testsare rather low. In particular, presence of structural breaks seriously affects thevalidity of co-integration test.

Seen in this backdrop, the present paper is different from its predecessors fromtwo standpoints — one relating to coverage and the other relating to econometricmethodology. First, it explicitly looks into finances of the Central Government andState Governments, as well as their combined revenue and expenditure patterns.Secondly, in doing so, it explicitly looked for possible presence of any structuralbreaks in discerning co-integrating relationship(s) between revenue and expendi-ture.

We find that while the finances of Centre and State Governments are individ-ually unsustainable, the combined finances are sustainable provided, allowance ismade for structural breaks in the co-integrating relationship between revenue andexpenditure series. Further investigation, as to why combined finances are sus-tainable when individual finances are not, indicates that while inter-governmentaltransfers lead to overstatement/overestimation of the gap between revenue andexpenditure of governments at the individual levels; at the combined level, theimpact of these transfers is neutralised and the gap between revenue and expendi-ture gets narrowed down. In view of this, conclusions on sustainability of Indianpublic finances, based on the evidence at the individual levels of governments,could be biased. We, thus, conclude that any credible assessment of public debt inthe Indian context should necessarily be of the combined finances, which excludesthe impact of inter-governmental transfers emerging from financial intermediaryrole of the Central Government.

Rest of the paper is organised as follows:Section 2is a brief review of theliterature on measurement of sustainability of debt using the IBC approach. Thestylised facts on the trends in revenue and expenditure of the governments in India


are discussed inSection 3. The empirical estimates and the analysis of the resultsare then detailed inSection 4. Section 5concludes the paper.

2. Using the IBC approach to measure sustainability

2.1. The IBC approach: a digression

In discerning debt sustainability we employ the standard IBC approach. Follow-ing Hakkio and Rush (1991)andde Castro, Gonzalez-Paramo, and de Cos (2001),we give a brief description of the analytical framework, so as to bring out the em-pirically testable hypothesis from first principles. As the IBC approach requiresthat the current market value of debt be equal to the discounted sum of the expectedfuture primary surpluses, the starting point for deriving the inter-temporal budgetconstraint is to iterate forward the budget constraint at any single periodt, whichcan be expressed as:

�Bt = Gt − Rt + rtBt, (1)

whereBt is the government debt at the end oft, Gt is the government expenditureexcluding interest payments,Rt is the government revenue receipts (consisting ofboth tax and non-tax revenue including grants) andrt is the rate of interest payableon outstanding government debt att.1 The government always faces a similarconstraint for periodt + 1, t + 2 and so on. Expressing (1) as a ratio of GDP,

�bt = et − �t + �tbt−1, (2)

where�t = (rt − gt)/(1 + gt), et = Gt/Yt , �t = Rt/Yt .Assuming that�t is stationary around a mean�0, (2) can be expressed as,

�bt = e′t − �t + �tbt−1, (3)

wheree′t = et + (�t − �0)bt−1.

IteratingEq. (3)forward yields

bt =∞∑

j=0

�j+1(�t+j − e′t+j) + lim

j→∞�j+1bt+j+1, (4)

where�j+1 = (1 + �)−(j+1).Taking expectations in (4), the government inter-temporal borrowing constraint

can be expressed as,

bt = Et

∞∑

j=0

�j+1(�t+j − e′t+j). (5)

1 For simplicity of analysis, we assume that in each period the government budget deficit is financedthrough creation of market debt only.


Note that (5) is equivalent to the transversality condition,Et limj→∞ �j+1bt+j+1 =0. The economic interpretation of the above condition is that for a debt processto be sustainable, current debt must be equal to expected present value of futureprimary surpluses.

For the purpose of empirical testing in this exercise, the expression given in (4)is represented in terms of�bt to obtain,

�bt = eRt − τt =

∞∑

j=0

�j+1(��t+j − �e′t+j) + lim

j→∞�j+1bt+j+1, (6)

whereeRt = et+�tbt−1, i.e., denotes total expenditure including interest payments.

Accordingly, the transversality condition changes toEt limj→∞ �j+1�bt+j+1 =0

Thus, the inter-temporal budget constraint inEq. (4) shows that for the debtprocess to be sustainable, the current debt must be equal to the expected presentvalue of future primary surpluses. The constraint, as can be seen, simply imposesrestrictions on the time-series properties of government expenditure and revenue,which follow from the specification on the right hand side ofEq. (4). The debtprocess will be stationary as long as government expenditure, revenue and thestock of debt are stationary in first differences, which also implies thateR

t and�t on the left hand side cannot deviate from each other over time. Thus, the debtsustainability tests in the literature pay special attention to the integration orders ofdeficit and debt process, and the usual procedure of testing sustainability consistof testing the stationarity of�bt in various forms, or alternatively the stationarityof eR

t − �t if both areI(1). If both the individual serieseRt and tt are I(0), then

clearly it is a case of sustainable debt process since the right hand side ofEq. (4)would necessarily be stationary in its first difference.

An alternative procedure would be to test for the co-integration between therevenue–GDP ratio (�) and expenditure–GDP ratio (eR) of the following form:

�t = � + �eRt + εt . (7)

If � andeR are non-stationary, the null hypothesis for sustainability is thatβ = 1and� andeR are co-integrated.2 If we substitute (7) into (1), then the debt–GDPratio in first difference can be expressed as,

�bt+1 = (1 − �)eRt − � − εt . (8)

If eRt is I(1) and 0< � < 1, it could be observed fromEq. (8)that�bt would be

I(1). Deficit in this case is weakly sustainable. On the other hand, if� = 1 andrevenue and expenditure are co-integrated, i.e.,εt is I(0) then�bt would beI(0)and deficit is strongly sustainable. Ifεt is not I(0) and� = 1, this is again a caseof weak sustainability. On the contrary, if� = 0 then deficit is unsustainable.

2 However, the condition that� = 1, is merely a sufficient condition for the government budget tohold.


Table 1Sustainability of Indian public debt: results from some existing studies

Authors Period Method Governments Conclusion

Buiter and Patel (1992) 1971–1989 Discounted series(IBC)

Public sector Unsustainable

Rajaraman andMukhopadhyay (2000)

1952–1998 Undiscounted series(trend analysis)

CombinedGovernment

Unsustainable

Olekalns and Cashin (2000) 1952–1998 Co-integration withbreaks

Centre Unsustainable

Jha and Sharma (2001) 1952–1998 Co-integration withbreaks

Centre Unsustainable

In any country, over a period of time, there are number of fiscal regimes, leadingto structural breaks in the trends of revenue or expenditure or in both. When suchregime shifts take place, the usual test for stationarity and co-integration withouttaking into account of these regime shifts could give rise to misleading results.In other words, a series which is shown to be non-stationary by usual tests, mayturn out to be stationary after allowing for the structural breaks and vice versa.Similar is the case in the co-integrating relationship when allowed for possiblestructural breaks. Thus, in some of the recent literature, the test for stationarityand co-integration between revenue and expenditure have been extended by incor-porating structural breaks both in the individual series and in the co-integratingrelationships.3

2.2. Sustainability of Indian public finance using IBC

In the Indian context there are a number of studies that have tested the sustain-ability of public debt in terms of inter-temporal budgetary constraint (Table 1). Inone of the initial studies,Buiter and Patel (1992)tested for the sustainability of debtthrough a stationarity test on the discounted public debt process for 1971–1989.Discounting various public sector debt (including Centre, State Governments andPSUs) by various alternative measures of interest rates, they found that irrespectiveof the alternative discount rate, Indian public debt was unsustainable as all the dis-counted debt series turned out to be non-stationary. Subsequently,Rajaraman andMukhopadhyay (2000), performing the test for stationarity on the aggregate publicdebt series of the Central and State Governments during the period 1952–1998,showed that debt–GDP ratio is on an unsustainable path as the series was found tobe non-stationary. However, unlikeBuiter and Patel (1992), they did not discountthe series on public debt.

Olekalns and Cashin (2000)andJha and Sharma (2001)examined the alter-native hypotheses of testing the presence of co-integration between revenue and

3 See for example,Olekalns and Cashin (2000)andde Castro et al. (2001).


expenditure, rather than stationarity test on the discounted series of public debt.Both the studies found that revenue and expenditure of Central Government fol-lowed anI(1) process during the periods 1952–1998.Jha and Sharma (2001)foundthe evidence of co-integration but since� has been found to be less than one theyconcluded that public debt do not satisfy the sustainability restriction. This is con-trary toQuintos (1995)or de Castro et al. (2001)who have shown that 0< � < 1also guarantees the sustainability but in weak form. On the other hand,Olekalnsand Cashin (2000)did not find any evidence for the presence of co-integration be-tween revenue and expenditure even after considering the possibility for structuralbreaks. They, therefore, concluded that the Central Government debt is unsustain-able.

A look at these works hints about the lack of the sustainability of Indian fiscalpolicy. This, coupled with the narrative evidence about various ratios, raises se-rious concern about the state of public finance in India. We shall argue later theorigin of the conclusion on lack of sustainability due to either of the two omissions,viz., non-consideration of Central and State Government together, and negligenceof the structural breaks in the co-integrating relation between revenue and expen-diture. But before we do so, let us have a review of the trends on Indian fiscalposition.

3. Some trends in Indian fiscal position

In view of the prominent role assigned to the growth of public sector in thedevelopment planning in the Indian Economy, Government expenditure exhibiteda steady rise over the years since the early 1950s. However, given the constraints inraising resources through tax and non-tax sources, a significant proportion of thepublic expenditure had to be met through borrowings, which resulted in a steadyrise in public debt over the years.

Public expenditure of the combined Government as a proportion to GDP showeda rapid growth from 11.7% in the 1950s to 19.3% in the 1970s and further to25.9% in the 1980s. However, the pattern of expenditure has been highly skewedtowards current expenditure and there has been a persistent decline in the share ofdevelopmental expenditure. Growth of revenue remained relatively sluggish fromthe 1950s through the 1970s, followed by deterioration in the mid-1990s. Grossrevenue of the Government as a proportion of GDP rose from 8.6% in the 1950sto 14.8% in the 1970s, further to 17.8% in the 1980s and remained more or lessaround the same level in the 1990s (Table 2).

In conformity with the conventional paradigm that revenue account should gen-erate surplus and borrowings should be used only to finance capital expenditure,during the first three decades of post-Independence, the Central and State Gov-ernments maintained by and large surpluses in their revenue accounts. In fact,it is observed that most part of the period through the early 1950s to the 1970swas a period when the growth in public expenditure was relatively more aligned


Table 2Revenue and expenditure of the Centre and the States (as a percentage of GDP)

Period Centre State Combined

Expenditure Revenue Expenditure Revenue Expenditure Revenue

1950s 6.03 4.55 6.46 4.74 11.70 8.631960s 11.96 7.20 9.01 6.50 16.80 12.051970s 12.05 8.56 10.55 8.52 19.34 14.821980s 16.44 9.62 13.85 10.98 25.87 17.751990s 15.75 9.40 14.13 11.15 25.59 17.50

with the revenue flows. Significantly, in case of Central Government, there hasbeen notable rise in expenditure in the early 1960s, followed by a sharp declinein the latter part of the 1960s. Increased expenditure during the first half of the1960s seem to have been caused by increased defence expenditure in view ofconflict with China and Pakistan. In addition to this, Central transfer to Stateswitnessed unprecedented rise since the mid-1960s. In fact, trends since 1951show that the share of net lending to States by the Central Government as pro-portion to total expenditure have been highest during the second half of 1960sand the first half of the 1970s. During the 1980s, there was an explosive growthin both the expenditure and revenue. Revenue growth could not keep pace withthe rise in expenditure and the fiscal gap steadily increased during the periodnecessitating heavy borrowing requirements. The growth rate of borrowing re-quirements of the government increased from around 17% in the 1970s to 21%in the 1980s. Consequently, the aggregate public debt of the combined Gov-ernment as a proportion to GDP rose from 22.9% in 1952 to as high as 61.3%in 1991.

In 1991 India experienced a Balance of Payments crisis and took recourse toa standby arrangement of the International Monetary Fund. Following the crisis areform process was initiated. Fiscal adjustments and consolidation measures werepart of this reform process, which brought some improvement in the fiscal sce-nario. The impact, however, remained restricted mainly to the Central Governmentfinances. Improvement in fiscal–monetary coordination, elimination of automaticmonetisation of fiscal deficit, introduction of market oriented interest rates andtax reforms brought about a material change in the resource gap incurred by theGovernment and in the mode of its financing. During the first half of the 1990s,Central Government finances witnessed a narrowing down of the fiscal gap. Thiswas achieved largely by a cut in expenditure as revenue generation continued to besluggish, as across the board cut in tax rates did not generate the desired level of taxbuoyancy. The structural transformation in the economy also impacted upon therevenue flows. Although the services sector contributes about 50% to the GDP andhas been growing at a relatively faster rate, it largely remains outside the purviewof taxation. Similarly, agricultural sector, a potential area for revenue generation,remains unexploited.


However, the fiscal correction could not be sustained for long as the Governmenthas been unable to curtail the rise in consumption expenditure, and in particular,the interest payments and spending on wages and pension. Fiscal deficit–GDPratio, which declined in the immediate aftermath of fiscal adjustment programmeinitiated in 1991, started to pick up in the latter half of the 1990s for the CentralGovernment and by 1993–1994 in the case of State Governments. As a result, thecombined debt of the Central and State Governments as a ratio to GDP, whichdeclined from a peak level of 61.7% in 1990–1991 to 56.8% in 1996–1997, startedrising and reached a level of 63.7% in 2000–2001.

The notion that deterioration of fiscal situation in India is of recent origin hasbeen questioned. Explaining for the progressive deterioration on the fiscal frontover the years,Lahiri (2000)point out that a progressive relaxation of appropriatebudgetary checks was already evident from the mid-1950s.4

By the nature of the budgetary operations ingrained in the structure of thefederal finance in India, the issue of sustainability of Central Government debtand State Governments debt diverge significantly at least in one important respect.As far as the market permits, Central Government has the unlimited power toborrow from the market to finance its deficit, and also until very recently, it couldmonetize its deficit by borrowing from the RBI at sub-market interest rates.5 TheState Government, on the other hand, operates under a hard budget constraint.As the States are indebted to the Central Government, by the mandate of IndianConstitution, the States are required to obtain prior consent of the Centre forborrowing from the market. They also do not have the facility of direct recourse toborrowing from the RBI. In view of these unequal powers between the two layers ofGovernment, while the fiscal deficit and debt of the Central Government may growwithout constraint, there is an implicit cap in the growth of deficit and debt of StateGovernments. Thus, the fiscal deficit of the Centre and that of the fiscal deficit ofthe States and UTs combined, which were close to each other until the mid-1970sdiverged since then. The fiscal deficit to GDP ratio of Central Government rose byabout three times from 3.08% to 8.47% during 1970–1971 and 1986–1987. Aftera gradual fall till about the middle of the 1990s, it has once again picked up. On theother hand, the combined deficit of States and UTs rose by only about half timesduring the comparable period rising from 2.14% in 1970–1971 to about 3.25% inthe mid-1990s. Consequently, the debt of States grew relatively slower than thatof the Centre.

4 This is evident from of the report of the Fourth Finance Commission (1965) which noted, “In1955 the Government of India advised the State Governments that all expenditure on capital assets, thatis durable or fixed though not necessarily productive or self-liquidating assets, should be held eligiblefor being serviced out of loans, and the amortisation of such loans need not be treated as a charge onthe revenue except to the extent that the State Governments were bound to provide in accordance withany law or any specific undertaking given in the case of any loan” (paragraph 148).

5 Since 1994, the process of automatic monetization has been replaced by a system of Ways andMeans Advance from the Reserve Bank Of India.


The relative stability of fiscal deficit and debt of States as against that of Centreis a reflection of the harder budget constraints of the former. However, with the per-sistent upward pressures on revenue expenditure emanating from increasing costof borrowing and other current expenditure such as wages and salary, the hard bud-get constraint has led to a continuous squeeze on capital expenditure of the States.Social spending being primarily the responsibility of the State Governments, thishas seriously hampered the provision of social and physical infrastructure by theStates thereby threatening macroeconomic stability and economic growth (Rao,2002).

There are substantial inter-State differences in the degree of financial imbal-ances with high- and middle-income States being generally in a relatively betterposition than the rest. The overall trend, however, has been decline in the growthof own funds and sluggishness in Central transfers that necessitated increase inthe share of borrowings in total resources required to meet steady increase ingovernment expenditure. A number of factors have been identified for the rapiddeterioration in the finances of States. Few of them may be mentioned. First, StateGovernment employees have not only grown but at regular intervals the wages andsalaries have also been raised. For some States it has come to a situation wherethe entire revenues are not adequate even to meet wages of their employees.6 Theimpact of large employment is also visible in growing pension bill.7 Second, thereare huge explicit and implicit subsidies on non-merit goods reflected in the verylow recovery rate. In 1994–1995, the recovery rate on services by the States is esti-mated at 9.3%. Third, a number of State Government run enterprises (called PublicSector Enterprises, or PSEs), particularly, the State Electricity Boards (SEBs) andState Road Transport Corporations (SRTCs), are making huge losses due to un-economic pricing and inefficiencies. As a result, instead of revenue flowing to theState budget by way of profits and dividends from the PSEs, there is massive drainfrom the State budget in the form of loan and grants to these PSEs. Fourth, theinterest payments have risen enormously due to rising debt and cost of borrowing.8

Fifth, tax revenue has stagnated due to inefficient tax machinery, complicated taxsystem allowing number of exemptions leading to large-scale tax evasion, tax warbetween States, etc. Sixth, the gap filling approach in the devolution of resourceshas discouraged fiscal prudence in the States. Seventh, in the recent years the lowbuoyancy of Central transfers and spillover of Central pay revisions have had themost adverse impact on State finances.

6 Kurian (1999)noted, “Several State Governments are indeed becoming ‘governments of theemployees, by the employees and for the employees’ only”.

7 In some States, pension payments have been indexed to wages as a result with each pay revisionthere is a concomitant rise in pension liabilities.

8 Although there is a distinct decline in the cost of borrowing in the recent years, size of interestpayments continue to grow due to overhang effect of high cost borrowings of the past and, ever-wideningfiscal gap.


A mention could be made of Central transfers to States which essentially reflectthe financial intermediary role of the Union Government and have an impact on thefinances of both Central and State Governments. There are at least three channelsthrough which Centre transfer funds to the State Governments viz., on the rec-ommendations of the Finance Commission, Planning Commission and on accountof Centrally Sponsored Schemes.9 The total loans and advances from the CentralGovernment to the States as a proportion to aggregate expenditure (combined)grew sharply from 15.7% in the 1950s to 22.1% in the 1960s. Since then there isa steady fall in the quantum of central transfer as a proportion to total expenditureto just about 9% during the 1990s.

Given the increasing demand for expenditure, the States have soften their hardbudget constraint through several ways, which only further contributes to their de-teriorating fiscal situation. First, public account has no separate cash balances fromthat of the consolidated fund. Funds mobilised through various small savings andpublic provident funds schemes operated by both Central and State Governments,accrue to this account. The Central Government do not have any control even onfunds mobilised through its own small saving schemes as the entire accretion to theschemes are now passed on to the States as per the preset formula.10 Second, thereare Ways and Means Advances (WMA) from the RBI, which are meant to meetonly the temporary mismatch between receipts and expenditure. These advancesare repayable not later than 3 months from the date of availing that advance. Sinceit is not required to vacate these advances at the end of fiscal year, they becomea regular source of financing deficit of the States. Third, guarantees act as a sub-stitute for government expenditure in some cases as it enable PSEs of doubtfulcommercial viability to raise funds from the market instead of a grant and loanfrom the budget. Fourth, States also adjust accounts with PSEs by way of loanrepayment received in the budget. Given the financial conditions, these PSEs are,however, in no position to make the loan repayments.Rao (2002)also points outother ways of softening the hard budget constraints faced by the States. It worksin terms of advance tax collections and by keeping the contractor’s bill pending.

Thus, despite the apparent hard budget constraint, the deficits and debts of theState Governments have grown enormously over the years. With some reductionin the deficit of Central Government, particularly during the first half of the 1990s,the contribution of the State Government deficit in the combined deficit of thegeneral Government has increased. In fact, the responsibility of reducing fiscalimbalances of the general Government in the country to a large extent lies with theStates and a major effort on their part is called for in order to retrieve the situation(Bajpai & Sachs, 1999).

9 Finance commission is a statutory body that is set up every 5 years to recommend policy orformula to vertical and horizontal distribution of revenue collected. Planning commission is a permanentnon-statutory body set up by the executive.

10 Prior to 2002–2003, 75% of the total mobilisation under small savings schemes was being dis-tributed among the States.


Table 3Co-integration analysis and degree of sustainability

� Co-integration �bt Sustainability

1 Yes I(0) Strong>0 and<1 Yes I(0) Weak– No I(1) Unsustainable

In this backdrop of the State Government finances, we now move to investigatethe nature of debt dynamics by testing for the solvency of public finance in India.

4. Empirical results 11

4.1. How do we infer about sustainability of public finance?

As discussed earlier we adopt the inter-temporal budget constraint approach toaccounting approach in assessing the sustainability of Indian public debt. Withinthe approach, we adopt the concept of testing the co-integration of revenue andexpenditure series, if they follow anI(1) process.

In view of the relationships derived above, we adopt the following procedure inorder to test the strong and weak sustainability of Government debt in the Indiancontext. As a first step, orders of integration of the variables ‘e’ and ‘t’ would bedetermined. If they areI(1), test for presence of co-integration would be conductedas perEq. (7), i.e.,�t = � + �eR

t + εt and null H0: � = 0 against the alternativeHa: � > 0 would be tested. If H0 is accepted, the deficit is unsustainable, and ifrejected, the null H0: � = 1 against the alternative Ha: � < 1 would be tested. Ifthe null is rejected, it would imply 0< � < 1. After obtaining the results from thisprocedure, implications for strong and weak sustainability would be determinedas perTable 3.

4.2. Results from co-integration testes

In context of the above framework, we now proceed to see the sustainability (orits lack) of government finance in India via the existence of co-integration betweenthe revenue–GDP ratio and expenditure–GDP ratio. As set out earlier, the presentpaper undertakes investigation of debt dynamics in case of Central Government,State Governments and combined finances of Center and the States. As a first step,we began by ascertaining the degree of integration for revenue and expenditureseries in each case. All the series are found to beI(1).12

11 Period of study extends from 1951 to 2000 and the data contained inLong-term fiscal trends inIndia—A conspectus, National Institute of Public Finance and Policy, New Delhi has been used for thepurpose of this study.

12 Annex I reports the unit root results from Dickey–Fuller and Phillip–Perron tests.


Table 4Long-run relationship between revenue–GDP ratio and expenditure–GDP ratio

Parameter and test estimates Centre States Combined

Engle–Granger test� 0.47 0.82 0.60ADF −2.61 −2.41 −2.29PP test −2.84 −2.10 −2.27

Phillips–Hansen test� 0.49 0.84 0.61ADF −2.57 −2.57 −2.30PP test −2.88 −1.92 −2.33

Fully modified Wald test � = 0 (154.2)∗ � = 0 (580.4)∗ � = 0 (258.2)∗� = 1 (178.6)∗; � = 1 (21.42)∗ � = 1 (102.0)∗

∗ Rejection of null at 5% significance level.

In view of the fact that all the series areI(1), we now proceed to the testing forthe presence of co-integration between revenue–GDP ratio and expenditure–GDPin each case following bothEngle and Granger (1987)andPhillips and Hansen(1990)procedure.13

The results indicated inTable 4reject the presence of any long-run relationshipbetween revenue and expenditure for all the levels of the governments indicatingunsustainable debts.

4.3. Co-integration relationship under the presence of structural breaks

How far are these results robust? It has been shown that the power of theco-integration tests diminishes in the presence of structural breaks (Gregory &Hansen, 1996). Since it is unlikely that such break points will be knowna pri-ori, Gregory and Hansen (1996)proposes a statistic that attempts to test the nullhypothesis of no co-integration against the alternative co-integration with a struc-tural break at an unknown point of time. They suggested testing of the followinghypotheses to account for three types of possible structural breaks:

Level shift : y1t = �1 + �2Dt� + �y2t + et, (9)

Level shift with trend : y1t = �1 + �2Dt� + �t + �y2t + et, (10)

Regime shift : y1t = �1 + �2Dt� + �1y2t + �2y2tDt� + et, (11)

whereDt� = 0 for t ≤ �T , Dt� = 1 for t > �T , � ∈ (0, 1).As suggested inGregory and Hansen (1996), these models are estimated recur-

sively by OLS for all possible break points in the interval� ∈ (0.15, 0.85). We

13 Similar results are obtained with Johansen’s maximum likelihood procedure. Annex II reportsthese results.


Table 5Long-run relationship between revenue–GDP ratio and expenditure–GDP ratio in presence of an un-known structural breaks (results from Gregory–Hansen test)

Level shift Level shift with trend Regime shift

Year Test statistic Year Test statistic Year Test statistic

Centre 1962 −4.16 1962 −4.18 1969 −4.55States 1970 −3.17 1994 −3.45 1994 −4.45Combined 1965 −4.89∗ 1965 −4.93∗∗ 1965 −4.84∗∗

∗ Significant at 5%.∗∗ Significant at 10%.

computed the modified ADF∗ statistic as, inf�∈T ADF(�) the observation associ-ated with the maximum statistic is taken as the most probable break point.14

Allowing for possible structural breaks, the G–H tests, shown inTable 5, in-dicate that while the debt of Central and State Governments individually are stillunsustainable, the combined Government debt is sustainable once allowed forstructural break in the co-integrating relationship. It can also be seen inTable 5that irrespective of the model chosen, the structural breaks occurs in 1965.

Thus, while the revenue and expenditure series of Central and State Govern-ments taken individually diverge from each other, taken together, combined rev-enue and expenditure series converge that the limit term inEq. (6)tends to zero.Hence the fiscal position of Central and State Governments are individually un-sustainable, whereas taken together they are sustainable.

4.4. Why are the combined finances sustainable? A conjecture

When the individual finances are unsustainable, how could the combined fi-nances be sustainable? To seek reasons for such observations, one need to examinethe adjustments involved in arriving at the combined series from the individual se-ries. Relationship between combined and individual Gross Fiscal Deficit (GFD),revenue and expenditure series can be depicted in the form of the following iden-tities:

Combined GFD = (GFD of Centre+ GFD of States)

− net lending from Centre to States,

Combined revenue receipts = (revenue receipts of Centre

+ revenue receipts of States)

− (interest receipts of Centre from States

+ grants received by States from Centre),

14 Gregory and Hansen (1996)have shown that this ADF∗ statistics is equivalent to theZt statistics.


Combined expenditure = (expenditure of Centre

+ expenditure of States)

− (net lending by Centre to States

+ interest payments by States to Centre

+ grants made by Centre to States).

It can be seen from the above identities that for arriving at combined finances,all intra-governmental transfers involved in the individual finances need to beexcluded from individual series of the Central and the State Governments. There arethree intra-governmental flows in the above identities viz., net lending of the Centreto States, interest receipts of Centre from States and grants-in-aid by the Centreto States. Of these, two items viz., interest receipts (revenue) and grants-in-aid(expenditure) by the Centre, are essentially the contra items and correspondingentries appear on both the expenditure and revenue sides of the State Governmentbudgets. On the other hand, the third item, i.e., ‘net lending’ by the Centre toStates appears as an expenditure item in Central Government budget and there isno corresponding entry in the State Government revenues. Obviously, being loans,net lending do not form part of non-debt receipts of the State Governments. Thus,in the process of computing combined numbers, interest receipts and grants-in-aidare knocked off from both aggregate revenue and expenditure series of Centre andthe State Governments, whereas net lending is excluded only from the aggregateexpenditure.

Thus, it is the exclusion of ‘net lending’ from the Centre to States in the com-bined finances, which probably alters the time-series properties of the combinedexpenditure series and also narrows down its gap from the revenue series. It alsoimplies that it is probably the ‘net lending’ from Centre to States, which is the ele-ment responsible for divergence of Central Government’s expenditure series fromits revenue series. If this is true then it should logically imply that there should bea long-run relationship between the revenue and expenditure series pertaining toCentral Government once corrected for ‘net lending operations’.

With a view to confirm the presumption that combined finances is sustainablenecessarily on account of exclusion of ‘net lending by Centre to State Governments’,we analysed the issue of debt sustainability further by testing the adjusted expendi-ture series of Central Government.15 Deriving the expenditure series by excluding

15 There are two components of ‘net lending’ viz. ‘loans and advances’ by Centre to States and‘Recovery of loans’ from States to Centre. Thus, for arriving at adjusted numbers for the CentralGovernment, there could be two procedures. First, net lending could be straight away subtracted fromthe expenditure side. Alternatively, loans and advances could be subtracted from the expenditure sideand recovery of loans could be added on the revenue side. Although, the size of revenue and expenditurewould vary depending upon the choice of adjustment procedure, obviously, the gap in revenue andexpenditure series would remain same in either case. We also conducted the G–H test with both thetype of adjustments; results turned out to be similar.


Table 6Gregory–Hansen test for structural breaks in the co-integrating relationship — finances of CentralGovernment adjusted series

Level shift Level shift with trend Regime shift

Year Test statistic Year Test statistic Year Test statistic

1973 −5.55∗ 1973 −5.69∗ 1974 −3.33

∗ Significant at 5%.

net loans and advances to the States from the Centre, we conducted the G–H testfor co-integration with structural break. Results are as discussed below.

Null for the absence of long-run relationship was rejected, as was the case withcombined finances inTable 6. Thus, with adjusted expenditure series, there is theevidence that revenue and expenditure are co-integrated and Central Governmentfinances are sustainable. It implies that assessment of sustainability of CentralGovernment debt without excluding its operations relating to financial intermedi-ation could give biased results. Thus, it would be appropriate to assess combinedfinances, which would bring out the true position of the government finances.

4.5. Measuring the degree of sustainability

Having obtained the evidence about sustainability we now move to ascertain thedegree of sustainability. As indicated earlier, mere presence of long-run relation-ship is not the sufficient condition for a strong sustainability. The speed at whichinter-temporal budget constraint is satisfied is also critical. Speed of adjustmentcould be measured by� values as estimated inEq. (7). In the present exercise, betavalues are relevant only in case of central finances with adjusted series and for thecombined finances where there is evidence for long-run relationship. The resultsare as shown inTable 7.

As 0 < � < 1, it implies that speed at which inter-temporal budgetary con-straint is satisfied is lower in case of both central finance (adjusted) as well as thecombined finances. Thus, both are the cases of weak sustainability. Put differently,even if there is no immediate problem of fiscal sustainability, there are possibilitiesof future problems. This is because the speed at which the inter-temporal borrow-ing constraint is satisfied is quite slow and is likely to result in higher growthrate of debt. Higher growth of debt could result in rise in interest rate, which

Table 7Speed of adjustment to inter-temporal budget constraint

Item �

Central finances (adjusted) 0.41Combined finances of Central and State Governments 0.48


might eventually create problems for marketing or rolling over of public debt infuture.

5. Conclusions

The present paper primarily focuses on three issues in its attempt to extend theexisting analysis on sustainability of Indian public debt. First, whether India’s pub-lic debt is sustainable if structural changes on account of regime changes are takeninto account. Second, whether financial intermediary role of the union governmentis a significant factor influencing the sustainability of the public debt. Third, if thepublic debt is found to be sustainable, what is the degree of sustainability?

From the series of tests conducted on Central, States and combined finances it isfound that while the finances of both Central and State Governments are unsustain-able individually (whether or not the allowance is made for structural break), theircombined finances are sustainable when structural break is taken into account. In-vestigation as to why combined finances are sustainable when individual financesare unsustainable brings out the importance of financial intermediary role of theCentral Government. In India, to bring about vertical as well as horizontal eq-uity across the government finances, there are sizeable statutory and non-statutorytransfers from the Centre to the State Governments. These inter-governmentalflows lead to overstatement/overestimation of the gap between revenue and ex-penditure of the Central Government. At the combined level, with the netting outof inter-governmental flows, the gap between revenue and expenditure becomesnarrower. To that extent, conclusions regarding sustainability of Indian public debtbased on the individual level of governments are biased. It emerges that any cred-ible assessment of public debt in the Indian context should necessarily be of thecombined finances, which excludes the inter-governmental flows. Furthermore, wefind that public debt of the combined Government sector is weakly sustainable, im-plying that the speed at which the inter-temporal borrowing constraint is satisfiedis quite slow and is likely to result in higher growth rate of debt. Higher growth ofdebt could result in rise in interest rate, which might eventually create problems formarketing or rolling over of public debt in future. In other words, while fiscal stanceof Government is sustainable at least in the short-run, for long-run sustainabilitygovernment needs to alter its fiscal policies to prevent any adverse repercussions.

Appendix A. Unit root tests

As a first step, the order of integration of the revenue and expenditure series wasdetermined by conducting some of the standard unit root tests viz., Dickey–Fuller,Augmented Fuller–Fuller and Phillips–Perron (with and without trends). Results ofalternative unit root tests conducted for the Central Government, State Governmentand the combined Government finances are presented in theTable A.1.


Table A.1Unit root tests for government expenditure and revenue

Government Alternative Series DF test ADF test Phillips–Perron test

LevelsCentre Without trend Expenditure −2.47 −2.42 −1.95

Revenue −2.39 −2.36 −1.87With trend Expenditure −2.16 −2.22 −1.76

Revenue −1.96 −1.92 −2.65States Without trend Expenditure −1.18 −1.19 −2.34


Revenue −1.46 −0.89 −1.72Combined Without trend Expenditure −1.66 −1.69 −2.34


Revenue −1.82 −1.09 −2.32

First differenceCentre Without trend Expenditure −6.83∗ −5.14∗ −8.28∗

Revenue −6.50∗ −5.11∗ −7.41∗With trend Expenditure −6.85∗ −5.19∗ −7.36∗

Revenue −6.66∗ −5.39∗ −9.35∗States Without trend Expenditure −7.47∗ −6.21∗ −9.03∗


Revenue −7.50∗ −6.65∗ −9.22∗Combined Without trend Expenditure −8.95∗ −5.53∗ −7.94∗


Revenue −8.47∗ −6.60∗ −8.95∗

Note: The lag length in the ADF tests are chosen based on the Schwarz Bayesian Criterion (SBC).The PP-tests are for truncation lag of 1.

∗ Rejection of the null at least at the 5% significance level.

Table A.2Unit root tests for Central Government finances –adjusted revenue and expenditure series

Series DF test ADF test Phillips–Perron test

Level First difference Level First difference Level First difference

Without trendExpenditure −1.15 −5.85∗ −1.58 −4.41∗ −1.14 −5.76∗Revenue −1.43 −7.46∗ −1.47 −3.65∗ −1.39 −7.45∗

With trendExpenditure −2.21 −5.80∗ −2.21 −4.53∗ −2.45 −5.68∗Revenue −2.58 −7.37∗ −2.27 −3.72∗ −2.80 −7.37∗

Note: The lag length in the ADF tests are chosen based on the Schwarz Bayesian Criterion (SBC).The PP-tests are for truncation lag of 1.

∗ Rejection of the null at least at the 5% significance level.


Table B.1Long-run relationship between revenue receipts and expenditure (results from Johansen–Jusileius max-imum likelihood procedure)

Trace Eigenvalue

H0: r = 0against H1:r = 1




Centre 11.35 3.24 8.12 3.24States 8.13 2.84 5.29 2.84Combined 15.0 2.70 12.30 2.70

95% critical values 17.86 8.07 14.88 9.07

Note: The reported statistics are for the test with unrestricted intercept and no trend, though all thefive alternative specifications of testing co-integrating vector in Johansen’s procedure do not reject thenull of no co-integrating vectors.

From Table A.1, it is clearly seen that all the series are non-stationary in thelevel form and stationary in the first difference, implying that all the series followI(1) process.

The adjusted series was tested for the presence of unit root and results as inTable A.2.

Results inTable A.2reject the null for absence of unit root and show that boththe series areI(1).

Appendix B. Co-integration results from Johansen–Jusileius procedure

SeeTable B.1.

References

Bajpai, N., & Jeffrey, D. S. (1999). Fiscal policy in India’s economic reforms. In D. S. Jeffrey, A.Varshney, & N. Bajpai (Eds.),India in the era of economic reforms. New Delhi: Oxford UniversityPress.

Buiter, W. H., & Patel, U. R. (1992). Debt, deficits and inflation: An application to the public financesof India.Journal of Public Economics, 47(2), 171–205.

de Castro, F., Gonzalez-Paramo, J. M., & de Cos, P. H. (2001). Evaluating the dynamics of fiscal policyin Spain: Patterns of interdependence and consistency of public expenditure and revenues. Bancode Espana.

Engle, R., & Granger, C. (1987). Cointegration and error correction: Representation.Econometrica,55, 251–276.

Gregory, A. W., & Hansen, B. A. (1996). Residual — based tests for cointegration in models withregime shifts.Journal of Econometrics, 70, 99–126.

Hakkio, C., & Rush, M. (1991). Is the budget deficit “too large?”.Economic Inquiry, XXIX, 429–445.Jha, R., & Sharma, A. (2001).Structural breaks and unit roots: A further test of the sustainability of

the Indian fiscal deficit (ASARC Working Paper).


Kurian, N. J. (1999). State Government finances: A survey of recent trends.Economic and PoliticalWeekly, 34(19), 1115–1134.

Lahiri, A. (2000). Subnational public finance in India.Economic and Political Weekly, 35(18), 1539–1549.

Olekalns, N., & Cashin, P. (2000).An examination of the sustainability of Indian fiscal policy. Australia:Department of Economics, University of Melbourne.

Phillips, P. C. B., & Hansen, B. E. (1990). Statistical inference in instrumental variables regressionwith I(1) process.Review of Economic Studies, 57, 99–125.

Quintos, C. E. (1995). Sustainability of the deficit process with the structural shifts.Journal of Business& Economic Statistics, 13, 409–413.

Rajaraman, I., & Mukhopadhyay, A. (2000).Sustainability of public domestic debt in India (WorkingPaper). National Institute of Public Finance and Policy.

Rao, M. G. (2002). State finances in India: Issues and challenges.Economic and Political Weekly,37(32), 3261–3271.

World Bank (2003).India: Sustaining reforms, reducing poverty. Washington DC: World Bank.

is india’s public finance unsustainable? or, are the claims exaggerated?

Documents