Federalism in India: An Economic Analysis

Abstract

The outcomes of federalism have played out in very different manners in various societies that have chosen to adopt this design of organising themselves. The Indian context is particularly interesting because of how Indian states have formed, evolved or have carved out of one another into existence. In this paper, I explore whether smaller states could perform better on governance outcomes. The measure of governance is legislative activity in Indian state parliaments. The results indicate that as states become smaller units to govern, the legislative in activity in their respective parliaments does increase.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction and Literature Review

An important classical argument favouring federalism is laid out by Friedrich Hayek. His thesis is that in a heterogeneous society, apart from for truly national public goods such as defence or energy, the central government does not possess relevant information to frame policies that are suitable for all (Hayek 1948). However, when I observe how federalism has played out in various parts of the world, I see vastly contrasting outcomes.

The United States of America, which is a federal state, is one of the wealthiest and least corrupt nations in the world, while countries such as India, Mexico and Argentina, which are all federal states, have governments plagued with corruption and poor economies (Parikh and Weingast 1997). Comparing federal countries with other non-federal systems may not allow for accounting of several unobservable factors that vary between countries. In this paper, I focus within India’s federal structure, hence mitigating this problem of unobservable discrepancies. I am looking at different regions within a country, thus balancing the need for sufficient variation as well as the ability to control for regional idiosyncrasies and time-fixed effects.

Over time since independence, Indian states have been splitting into smaller and more homogenous units. The purpose of this paper is to understand whether these splits have resulted in better outcomes for governance. I measure governance using a new data set I collected on the number of bills and amendments passed in each state parliament(s) between 1956 and 2014. In view of state splits, amendments may be seen as the refinement of older existing laws to tailor to the needs of a newer, more homogenous society. Laws may be viewed as fresh legislative activity that actually results in new governance outcomes.

The Indian context is particularly unique because the idea of federalism can be tested in a relatively exogenous sense, i.e. a central government ensuring similar economic climate in all states, but also sufficient diversity among states. Also, there have been numerous state splits over time.

The key contribution of this paper is that so far, nobody has been looked at this issue through an empirical lens. There has been no effort in the Indian context, to test systematically the effect of state splits on economic and governance outcomes. This paper is a first step towards getting a better understanding in this field.

I find that a split causes legislative activity to increase by 12 bills (laws and amendments) approximately. The effect of a split on state domestic product is positive, but not significant. However, when I measure the effect of the split on state domestic product after 1990 (when the liberalisation reforms were implemented at the central level), I find that state domestic product increases by almost 33 lakh rupees. The value added in manufacturing units/factories reduces by 752 crore rupees after a state split.

The vast body of work in the area of fiscal federalism has been segregated into first- and second-generation fiscal federalism. The major difference being in an assumption made about public officials—the former treats them as benevolent, while latter treats them as working for their own good but who are held publicly accountable for their actions.

Second-generation fiscal federalism stresses on the importance of incentivising lower governments with sufficient tax revenues such that they provide ‘market-enhancing public goods’. This leads us to examine the concept of an ‘ideal’ form of federalism which is ‘market preserving’. See Weingast (2007) for a survey of literature on first- and second-generation fiscal federalism.

Weingast (2009) embarks on a comparative study across various federal units to understand the necessary elements to result in this ideal form of federalism. His paper discusses the importance of inter-jurisdictional competition to reap the benefits of federalism. Parikh and Weingast (1997) also present arguments for federalism that prevents different ethnic/religious groups from fighting one another over heatedly debated policy concerns. In an ethnically diverse country like India, this feature is beneficial for a functioning democracy. It can thus be seen in the several splits that are caused for reasons such as religion, language, ethnicity, as discussed in the background section below.

Drawing from the axioms of market-preserving federalism which forms a good base for any further comparative studies, I look at the data to see whether state splits are, in fact, resulting in governance improving outcomes and improvements in economic indicators. The main idea I test in this paper is whether the splitting of states into smaller federal units has any impact on governance and whether the outcomes of better governance reflect in economic variables indicators. The exploratory study’s findings are elaborated upon in the following sections.

2 Institutional Background

Understanding India’s political economy order is a colossal task and can be done through the lenses of various disciplines. In this paper, I wish to understand certain fundamental factors motivating the further fragmentation of India’s existing federal units. Before diving further into our research, I will use this section to provide some context on what motivates such a discussion.

Article 3 of the Indian Constitution sets out the right of Parliament of the Union government to create new states and change boundaries if a Parliamentary majority is achieved. As difficult as this may be to achieve, it is quite unique to India’s federal system as opposed to other forms of federal governments which allow the states greater protection of their rights in such issues (Singh and Pani 2012).

In the United States of America for example, for a new state to be formed, the consent of the Legislature(s) of the concerned states as well as the Congress is necessary, without which a new state cannot be formed ‘within an existing State’ or ‘by the Junction of two or more States’ (Constitution of the United States of America, Section 3).

The current 29 Indian states (as the federal units are called) all have state parliaments that come together for sessions three times a year. Each state parliament amends existing bills or passes new laws in areas that they have jurisdiction over. Upon gaining independence in 1947, state reorganisation was implemented to establish new boundaries of states and alter existing boundaries established in British India.

At the time of independence, the subcontinent was divided into 550 princely states and 14 provinces (Singh and Pani 2012). The Nehru administration set up a State Reorganisation Commission (SRC) whose recommendations were actualised in 1956 through the workings of The State Reorganisation Act, 1956 (Sarangi 2011).

Following Independence, federal organisation occurred on the basis of languages, and the first of such cases was Andhra Pradesh and then part of a larger Madras province. Upon the death of Potti Sriramulu, a well-known Gandhian following a 50-day hunger strike, the central government was forced to concede to the formation of the first linguistic state, Andhra Pradesh, in 1953 (Mawdsley 2002).

This marked the beginning of the development of political experiments, alliances, unrest and insurgency: all outcomes of passing of the Act which recognised fourteen states and six union territories (Sarangi 2011). Post-independence, the major task for the government was to integrate princely states into existing provinces so as to make the country a single geographical unit (Sharma 1967).

In order to be part of the Union of India, states had to sign an ‘Instrument of Accession’ (Sharma 1967). As Deputy Prime Minister, Sardar Vallabhai Patel managed to accomplish complete territorial integration of the then fourteen states of India into the constitutional set-up as of 1950 (Sharma 1967).

The State Reorganisation Committee (SRC) was later set up by the Congress to further organise the nation into efficient administrative units. This process was propelled by language as a basis of division. Several members of the Congress rejected the notion of ‘one language, one state’, still traumatised by the aftermath of the Partition. Punjab and Bombay provinces were left as bilingual states, as examples of ‘unity in diversity’ states. As of 1956, the original lines along which states were divided were linguistic as per the recommendations of the SRC.

Since then, there have been increasing demands for new states that are motivated by several other reasons. For example, some of these groups believe that resources of the state government are not shared equally among all regions within a state (Singh and Pani 2012).

The struggle for Telangana was motivated by such a discussion and resulted in a success in 2014. There are still ongoing struggles for separate states in Coorg, Karnataka and Vidarbha in Maharashtra.¹

3 Data

The main idea I test through this study is whether smaller states resulting from carving out of larger states are governed better. Governance being a term that can be understood in several ways requires us to impose a restriction on its interpretation for the purpose of this paper.

I measure governance quantitatively by using the number of bills passed/amended in state parliaments. Article 245 of the Indian Constitution lays out the extent of lawmaking power of the Parliament and State legislatures.

Article 246 of the Indian constitution governs subject matter of the laws made by Parliament and State legislatures. When a new state is carved out, the laws of the state it was created from continue to stay in force, unless explicitly repealed/amended.

The reason of choice is as follows: the more active a state legislature is, the more laws it will enact and the state will effectively be better governed. In order to test this, I utilise state-level panel data made publicly available by Timothy Besley and Robin Burgess of the London School of Economics. Below is a table of summary statistics of the variables used:

Table of Summary Statistics

	(1) Mean	(2) Std. dev.	(3) Min	(4) Max	(5) Observations
Per capita income	188.4492	199.0094	19	620	N = 1062
Factory value added	205.2627	322.243	19	888	N = 1062
State population	42643.25	29449.52	3555.474	177869.3	N = 572
Laws	41.64501	15.86403	19	62	N = 1062
Amendments	67.98495	56.78084	0	393	N = 1062
Split	0.1374765	0.3445	0	1	N = 1062

This paper makes use of acts and bills data available on two major sources—PRS ‘Laws of India’ database and Manupatra. The merging of the two data sets was performed on MATLAB by using the unique state IDs to ensure a 100% match of all variables. The fourteen states reorganised as of 1956 have now become twenty-nine through several carving out events. In order to get a clear picture of the effect of state splits, I need a way to capture differential effect across a broad cross section.

Thus, upon importing to stata, combination of states that were initially one at the time of the States Reorganisation Act were assigned a new unique region ID. This acts as an identifier for a ‘greater’ state region. The data are collapsed into 19 regions within which the splits occur. All state variables are cumulated into the regions. The picture below depicts this:

Since the sample period is 1956–2014 (both inclusive), we have 59 * 19, i.e. 1,121 observations or lesser when there are missing data points. The split is accounted for by using a dummy variable which takes value ‘1’ starting from the year that the new state was carved out and otherwise remains ‘0’. For example, in the image above, Region 1 will take value 1 starting from the year 1963 because State 1 splits in that year.

Laws and amendments have been separated into two separate variables in order to test whether there are any differences: this separation was done by writing a program on MATLAB. Laws are completely fresh legislation representing the policy objectives that are ‘new’, whereas amendments are simply to modify existing legislation to adjust to changing policy needs rather than changing the direction of policy itself.

Three dummy variables were created to represent the number of times a state was split. The maximum number of times a state has been reorganised in India is three times. Until the year of the respective time(s) of the split, the observations take the value 0 and then take the value 1. States not split always take the value 0. The dummy variable representing the second split is used as a control variable in the regressions.

4 Results

This exploratory activity to understand how governance is affected by state splits can be summed up in the following way:

$$ {\text{Y}}_{\text{it}} =\upalpha_{\text{i}} +\updelta_{\text{t}} +\upbeta_{1} {\text{split}}_{\text{it}} +\upbeta_{2} {\text{x}}_{\text{it}} +\upvarepsilon_{\text{it}} $$

(1)

Equation (1): ‘y’ represents laws, amendments or a cumulation of both in Tables 1, 2, 3 and 4. ‘Split’ is the dummy variable which takes the value ‘0’ if a state is never split and the value ‘1’ if a state is split, starting from the year when the split occurred.

Table 1 Baseline results—laws

Table 2 Baseline results—amendments

Table 3 Baseline results—total governance activity

Table 4 Post-reform(s) effect on economic variables

‘xi_t’ represents the vector of control variables, ‘δ_t’ represents year-fixed effects and ‘α_i’ represents region-fixed effects. The control variables used are:

• 1. dummy variable taking the value ‘1’ if a state is split twice and ‘0’ otherwise

• 2. region’s population.

4.1 Governance

The baseline results are presented in Tables 1, 2 and 3. The estimates show that the number of ‘fresh’ laws decreases by 6 bills when time-fixed effects are not controlled for. But, we see that the number of amendments increases by 18 bills without controlling for time-fixed effects. The overall outcome is that total legislative activity increases by 12 bills with state splits, thus showing that the effect of amendments activity overpowers that of fresh laws (Tables 4, 5 and 6).

Table 5 Shift to services economy

These results are in line with the idea of market-preserving federalism which says that a state government has greater incentive to implement policies supporting economic activity when they can reap larger proportions of the rewards. The chances to reap large portions of fiscal rewards are higher when a state is smaller and more homogenous.²

4.2 Economic Outcomes

This exploratory activity to understand how economic outcomes are affected by state splits can be summed up in the following way:

$$ {\text{Y}}_{\text{it}} =\upalpha_{\text{i}} +\updelta_{\text{t}} +\upbeta_{1} {\text{split}}_{\text{it}} +\upbeta_{2} {\text{split}}_{\text{it}} \text{ * }{\text{post}}_{1990} +\upbeta_{2} {\text{x}}_{\text{it}} +\upvarepsilon_{\text{it}} $$

(2)

Equation (2): ‘y’ in Tables 4 and 5 represents various economic indicators, i.e. net state domestic product per capita, value added in factories, number of man days lost in industrial disputes and proportion of state domestic product coming from service sector activities. The interaction term is between ‘split’ and a time dummy taking the value 0 before 1990 and 1 after 1990. This was generated because liberalisation reforms in India were implemented after the year 1990.

A potential problem with the results in the previous section is that there could be factors endogenous to each state that has resulted in positive governance outcomes post-split. Therefore, I use the interaction term which accounts for the liberalisation reforms implemented at a country level. I use Eq. (2) for the following reasons. First, I expect the effects of a split to be magnified since states have more opportunities to expand their domestic businesses. Second, this central-level nature of reforms resolves the endogeneity concern.

The reason for using economic indicators as dependent variables is that successful market-preserving federalism results in more tax revenue for governments which therefore promotes business activity in these states (Parikh and Weingast 1997). Drawing from this, I can say that governance matters more when there are market opportunities. No states face particular disadvantages because of the reforms (enacted at a central level), it can be observed in Tables 4 and 5 that the coefficients confirm this story.

I observe that state domestic product increases by 28 lakh rupees after the reforms are implemented, if a state is split. The interaction term and the per capita state domestic product are thus showing how impact of post-liberalisation reforms magnified the effect of increased governance activity on economic outcomes.

The interaction term provides for exogenous variation. The liberalisation reforms after 1990 were implemented at the central level and were not within the decision-making power of the state governments. I can therefore say that the opportunity for economic development was impartially given to all states.

Interaction term coefficients in columns (1) and (2) of Table 5 indicate that liberalisation does indeed shift economic activity towards the service sector and the benefits of a state split accrue at an increasing rate. There is a positive and significant relationship between the proportion of net domestic product coming from service sectors (banking, insurance, communication, trade, hotels and restaurants), while I observe contrasting results without controlling for reforms. The state domestic product coming from the services sector increases by 2 lakh rupees, post-reform, if the state has been split.

A possible reason could be that the Indian economy shifted away from being manufacturing-oriented to being service-oriented. Seeing that there was a competitive advantage in the services sector, business ventures focused their efforts there.

5 Further Work

The measure of governance that I am currently using is very preliminary. An important concern is that these laws are simply a number which may not be representative of how impactful this activity may be at all. The nature of such laws cannot be determined beyond simply segregating them into fresh laws or amendments. However, this basic study is an important first step towards the direction of in-depth, data-driven study of the Indian federal system. An important caveat to keep in mind is the political structure in India.

The context of these laws depends heavily on the political parties in power in the parliament. Accounting for this will need strong methods to eliminate endogeneity because each state faces its own political challenges. Thus, for the purpose of this paper, I will not focus on political issues that may affect the overall outcome. Taking this study forward would include textual analysis of the content of these laws to determine whether these state splits would result in better governance outcomes.

Finally, as discussed in appendix, the factors leading to demands for statehood are various and idiosyncratic. It is likely that splits are endogenous to economic and governance outcomes. It would be interesting to explore the factors that lead to splits to come to a more comprehensive understanding of both: what causes an increase in federalism and how federalism affects outcomes.

• A. Overview on states

See Table 6.

Table 6 Summary of state split dates and causes

• B. Historical context of state splits—details

At the time of enactment of the Act, because of Nehru’s rejection of ‘one language, one state’ idea, the Marathi- and Gujarati-speaking regions were clubbed into the bilingual Bombay state. The administration, however, had to give into the demands for two separate states following Satyagraha and occasional violent outbursts in Maharashtra. Thus, following the electoral debacle of the Congress, Maharashtra was carved out of Gujarat on 1 May 1960, with Mumbai being included, much to the displeasure of Gujarat (Thakkar and Sanghavi 2011).

Following the partition, the Indian part of Punjab lost most of its agricultural territory to Pakistan, thus rendering majority of Sikh peasants to suffer economically (Kumar 2011a). From here on stemmed the dissatisfaction among the Sikh minority scattered all over a unified Punjab from various factors such as the government’s refusal to recognise Sikhism as a separate religion, not a sect under Hinduism (Kumar 2011a).

After prolonged violence and protests in the region led by Akali Dal (a Sikh-dominated political party), the Punjab Reorganisation Act was passed on 18 September 1966: Section 3 resulted in the formation of Haryana, Section 5 transferred Punjabi territory to Himachal Pradesh and Section 4 declared Chandigarh the capital for both Punjab and Haryana. Himachal Pradesh was also carved out of Punjab at this point, although it only attained full State recognition in 1971 Mawdsley (2002).

Nagaland was formed in 1963 by the government of India following armed conflict in the region in the 1950s Baruah (2003). The demands for division of states in the North-Eastern areas were rejected by the SRC because their formations were not based on the linguistic criterion that they promoted Dhar (2011).

The North-Eastern Areas Reorganisation Act passed in the Indian Parliament in 1971 led to the reorganisation of the region into Assam, Manipur, Meghalaya and Tripura. Arunachal Pradesh was initially a Union Territory and then become a State in 1987. The North-Eastern people, having been isolated since colonial times, were unsure as to why they had to join the large bloc of Indian states Dhar (2011).

In 2000, the North-central region of India experienced the formation of three new states of Chhattisgarh, Jharkhand and Uttarakhand out of three old states Madhya Pradesh, Bihar and Uttar Pradesh, respectively. Mawdsley (2002) argues that the support for these regional movements comes from BJP’s attraction towards electoral pay-offs: BJP won a majority in following elections.

The movement in Chhattisgarh was motivated by demands of local elites who sought privileges through geographical rearrangement Kumar (2011b). They believed that the distance from Bhopal’s administrative centre was too large. However, what truly mattered for the carving out of this state were the political numbers. When BJP was contesting for the 2000 elections, A. B. Vajpayee’s campaign included a promise to pass the bill allowed by Article 3: BJP won eight seats, and the majority passed in favour of formation of Chhattisgarh (Kumar 2011b).

In Uttarakhand, the majority of the population are upper-caste Brahmins or Rajputs accounting to between 80 and 85% of the population in the region (Singh and Pani 2012). The Government of India allows for reservation of seats in government jobs, universities, etc., based on national statistics of Scheduled Castes (SCs), Scheduled Tribes (STs) and Other Backward Classes (OBCs) (Mawdsley 1997). When the government under V. P. Singh raised the reservation of seats to almost 50%, students in these hilly areas who were of the upper castes began to protest. Eventually, this led to the creation of the State of Uttarakhand in 2000 (Mawdsley 1997).

The formation of the State of Jharkhand in India was the result of due recognition given to tribal identity in the region (Prakash 2011). Jharkhand was separated from Bihar for ethnocultural reasons: the area was predominantly occupied by ‘Scheduled Tribes’ (Mawdsley 2002). A resource heavy state, Jharkhand, is paradoxically one of the most poverty-stricken areas in India: 85% villages did not have electricity, only 54% were literate and about 56% of the population lived below poverty line (Prakash 2011). Obviously, socio-economic development took centre stage in the establishment of tribal rights to be materialised in the carving out of Jharkhand from Bihar in 2000.

An important case rejecting the linguistic division framework is the struggle for the formation of Telangana in 2014 (Srinivasulu 2011). The Congress hoped that by including Telangana into Andhra Pradesh on linguistic grounds, all the injustices to a backward Telangana would be forgotten. However, the protests went on due to several frustrations such as unemployment of educated youth, privatisation of the Public Sector Enterprises (PSEs), farmers’ suicides in Telangana and many other tragic triggers (Srinivasulu 2011). The core of statehood movement in Telangana was inspired by the cultural sphere where people from Telangana felt that their dialect and folk culture were being sidelined by the process of Andhraisation Srinivasulu (2011).

• C. Alternate Specifications: Ordinary Least Squares (OLS) and Negative Binomial

See Tables 7 and 8.

Table 7 Baseline with poisson estimation

Table 8 Baseline with negative binomial estimation