India post-independence is a period of time that deserves to be studied. Although many government parties took lead of the country from 1947 to now, they have preserved the Indian identity and pushed the economy up until gradually begin competitive with the most developed countries. In others words, following the forecasts of miscellaneous researchers, India will help fuel the world of tomorrow thanks to its powerful labour force, high growth development, etc.
Despite the fact that India went through several policy failures which damaged the population, its reforms keep moving forward and increasingly improve the welfare of its citizens. We can for instance show the relevance of the policies conducted by PV Narasimba Rao (BJP), Prime Minister from 1991 to 1996. When he took over the power, India was in a decline. For instance, 12% of inflation and 3.5% of the total GDP corresponded to the current account debt. He has undertaken stabilization and liberalization policies by borrowing founds from multiple organizations as the World Bank and the International Monetary Fund. Consequently, growth crashed into the negatives, finally reaching 5.2% annual growth almost 2 years later. This also gave rise to high, stable investment. By cautiously enforcing its reforms, India alleviated its socialist legacy and finally entered a sustainable growth path.
Moreover, since Narendra Modi’s election in 2014, India’s economy has moved faster than expected. In fact, the Prime Minister has led a large operation to reduce the threshold of the fraud that drains the country like a parasite. He has created new institutions to streamline tax collection and limited the amount of money lost to fraud as much as possible. In result of those reforms, the government has increased its investment capacity and has redistributed it to the public. Highways, telephone lines and education are examples of projects that have benefited from the government subsidies. In addition, a large initiative to census the billion residential Indians has been launched in order to provide them better access to services. Thus, the progressive approaches accompanied by nationalist reforms do not seem to be a drawback for the enhancement of society. There is no evidence whatsoever that proves the opposite.
However, those policies do not prevent Indian from suffering from unemployment, illiteracy and life below the poverty line. According to the data, approximately 13 million Indians are unemployed and another 393 million have a job qualified as “non-essential to the economy”; to cite a few, these are moonlight work, or cashier. Consequently, the biggest challenge for the government is to provide better job opportunities for its population, reduce the level of tax on raw materials, improve and access to health and education facilities, etc. Thus, although the overall situation of the economy is encouraging, the State has to become more responsive to its citizens.
Taking care of the population allows them to enforce appropriate policies and sharply increase their likelihood to be elected for another legislative term. Today, I am focusing on one election; precisely the Lok Sabha election that occurred in India in the May of 2019. The Lok Sabha or the ‘House of the People’ is one house that constitutes the Parliament; they are elected via an election where all Indian citizens can vote. The representatives are the voice of the population of the state from which they have been elected. In total, 543 seats have to be filled for a 5 year term. During their term, representatives must do their best to enact laws which contributes to civilian welfare. But let’s focus on the election process, for example, how many campaign promises dealt with either stabilizing or decreasing the domestic electricity or gas prices. Moreover, unemployment levels were a principal point and has been discussed many times.
This article is a summary of the 2019 Lok sabha election. I will try to identify some persistent factors, as well as variables which could have impacted the election, for instance, female attendance, the spread of political parties, etc. In addition to the econometric analysis, I will discuss the notion of power weight in Indian society through theoretical social choice methods.
Nevertheless, I want to specify that this analysis does not hold the intent to criticize, and rather only to underline the empirical aspects of the election and the conclusion which can be drawn from the data. In my first part, the paper contains data summarization and some econometric analysis; in the second part, the paper attempts to find a relation between party representation and power weight, while the final part concludes.
Part I: Statistics and Analysis
This sub-part provides a plain, fundamental analysis of the data. Firstly, support by the Young India Foundation Excel file, I gathered every piece of data from within a supportive database. This first set of variables constituted of 4 factors: candidate’s name, party’s name, gender and age. In complement, I have added seven other determinants, the endogenous characteristics of each Indian State: its population, GDP, GDP growing rate, the literacy rate, its widespread area and its localization. Incorporating those components in our database allowed me to widen the spectrum of actions I could take. I could manage the data in order to obtain descriptive statistics and drive econometrics regression. For a sake of clarification, all of this data is available on the India government website.
Let’s start with the analysis; I will be primarily focus on the data from a national level. We can observe 6390 candidates spread in 29 states and hundreds of parties in total. Within that set of candidates, 5511 are men and 879 are women, thus, women are under-represented in this election. Even though their place in the Indian society is properly defined, I wondered what caused a gap so large to emerge between men and women candidatures. One answer is noticeable by looking retrospectively into the Indian cultural history. The multiple scandals about widow remarriages, dowry and even female infanticide are some aspects which suggest that the gender equality is completely present in the system. Even if I can’t conclude that these events are the cause of the lack of women candidatures, I can nevertheless assume that they could have negatively impacted the choice of becoming a candidate in states where women’s rights are not a concern of the population. The overall data shows that women represent 10–15% of the total number of candidatures in each state. However, both states Maharashtra and Madhya Pradesh do not fit with these statistics, in fact, 50.4% of the candidates in Maharashtra are women while that figure reach 91.5% in the latter. These examples prove that balance equity between genders is not a impossible dream, but rather, only a matter of time.
Still focusing on gender, let’s take a look of the proportion of young males and females in our database. Supported by data, there are 100 women that are competing for a representative seat over 879, therefore, the new generation barely represents 11% of the number of competitor. In the same way, there were 448 males that competed for the LS election over 5511, and so the percentage of representation falls thus to 8%.
If I do not split the population by gender, the percentage of youth participation increases slightly, to 8.5%. Similarly to the first discussion, what could have caused this pattern? According to the data delivered by the Indian government, 600 million young people are below the age of 25 in India and almost 70% of the country’s demographic is below the age of 30. So, there is a conflict between young representation and their percentage in the population. However, I cannot conclude that all of these people are potential candidates; in fact, we have to take into account that the highest birth rates in India are often in remote areas where lower economics classes live. Those people may not have had the same quality of teaching at school or the same life expectations; thus, we can consider that politics and much of the related branch are not their principal concern.
However, we can discuss in a deeper way the contradiction between old versus young. Precisely, what could be the advantages or drawbacks to be an old or young representative? In most case, in favor of the oldest, people argue that they have more experience than the youngest; moreover, that their age grants them more maturity. Firstly, this idea is not totally true as being old does not necessarily correlate with a greater amount of life experiences—we all know somebody who has spent his entire life in the same area. Hence, even if a politician has been a state representative for a long time, he could have been elected several times because of few challenging oppositions, and not necessarily based on his skill. Regarding maturity, this argument varies a lot following the point of view, for instance, I really appreciate candidates with a modern approach, so I think they are more accurate and mature for filling in all of the duties imposed by their position. On the other hand, if I felt that electing an older candidate would be more comforting and would contribute to my welfare, I would consider those people more effective and mature for the vacancy. Consequently, maturity cannot be discussed as fundamental factor for determining my vote.
Two thirds of the Indian population is young, so, why are they not legitimized for helping rule the country? This receivable argument is correct but not totally true. We have several examples of countries which, while ruled by the majority, became dictatorial for the minority. Some historical events have demonstrated , in a few extreme cases, that cohabitation turned into mass killing. I would not say that this situation may happen in India but would merely warn you about these examples where the majority surpassed their rights. On the other hand, young people are often more aware of major issues that their country faces and can ask the right question. For instance: how can we create millions of valuable jobs in India? Would a universal health-care system be sustainable? Who can pay for the retirement benefits of 700 millions Indians? I am not alleging that old representatives do not think about those issues; rather, I am just informing you that people usually feel more concerned about the potential problems that could directly impact their life.
Thus, in both cases, age could have an impact on people’s thinking. Consequently, a government where the population is fairly represented could be the solution. The possibility of having multiple interactions between both old and young side seems to be relevant and benefit for all.
Now, I will try to emphasise relations, which could link the number of candidates and their age with endogenous constituents of the state they are living in.
In the first step, I want to determine the potential correlation between the factors identified. Although an econometric study cannot rely solely on the correlation parameter, it nevertheless contributes to form our model, which will be helpful to underline how impactful our endogenous variables are on the number of candidates. For the sake of comprehension, let’s illustrate that with the actual wage gap problem: we can easily deduct that being a man has positive impacts on your wage; thus gender is highly correlated with salary. A correlation coefficient is always between -1 and 1. -1 means a perfect negative correlation between the two factors; namely, when x increases by one, y decreases by one unit. When the coefficient is 1, however, there is a perfect positive correlation, where when x increases by one unit, y also increases by one unit. The last case is when the coefficient equals 0, which does not mean independence necessarily; rather, it is fair to say that there is no evidence of a linear correlation between the two factors; nevertheless, a discontinuous linear relation is still possible.
Now, in our model, I am trying to prove that the GDP, population, unemployment rate, literacy rate, gender and area of a state can result in noticeable fluctuations in the number of candidates. Thus, in this first sub-part, I will test the reliance of these factors. Primarily, there is evident correlations between factors, for instance, the total GDP of one state is tremendously correlated with the population, and where the population rises, the relation between GDP and population becomes increasingly stronger.
I have realised this matrix of correlation via STATA, but what is the matrix showing us? Firstly, we can see by neglecting the number of candidates variable that most of the results are within the -0.5/ 0.5 interval; namely, there are mostly lightly positive or negative correlations between factors. For instance, the population, the GDP and so on do not impact the age of the candidates much. Few variables are considered strongly correlated, except for those which were more obvious, like those between the GDP and the area. Finally, as expected, most of the interactions are linked to the number of candidates. The population, the GDP, the GDP growth rate and the unemployment rate have strong relations with the number of participants, while that number is also sometimes impacted by the other characteristics.
Thus, these results are comforting because those endogenous variables interfere with our dependent variable. We can finally envisage a model where most of the factors that cause fluctuations are explicitly used and, thus, exploit results for the final analysis part.
Regression results and analysis
Now, I can undertake the final part of the analysis, as all of the previous results have suggested to us that the number of candidates would fluctuate in consequence of variations in other factors. In the two first regressions, I want to distinguish candidates using their gender. I have previously deduced that male and female were not equal at all in representation, thus, comparing both of their results can partially explain where those differences are coming from. Before launching the regression model, I had to convert all the values into the same units; some of them were in percentage while others were in points of percentage—this conversion could entrain bias in my analysis. Moreover, certain factors had wide sets of values, which started from negative values to large positive ones. For sack of simplicity, I have used a log-linearization method in order to linearize my values and obtain results exploitable in percentage.
The equation used is compose as following:
- Y = Number of candidates (dependent variable)
- Seven explicative variables (for each state):
- X = Population
- W = Area
- Z = Unemployment rate
- A = GDP
- B = GDP growth rate
- C = Literacy rate
- D = Age
I obtained the following expression:
Then, I tried to determine, with as much precision as I could, the coefficient of each explicative variable. By using a regression method in STATA, I have obtained this chart:
This first regression is that which refers to the female sample. What does this chart show us? Firstly, R-squared is closer to 1, which means the model explains most of the components which interact with the number of candidates. Secondly, almost all the coefficients are significant at the 5% significance level. Moreover, as both independent and explicative variables are in logarithm, I could interpret these coefficients in percentage. Thus, when the population increased by 1%, the number of female candidates increased by 0.65%; this result sounds evident; indeed, with many more people in the state, there is a higher likelihood of having additional candidates. However, let’s focus on the results which are not so easy to determine, for instance, when the unemployment rate increased by 1%, the female participation rose by 7.8%. This result is very strange because it suggests that women trend to be more involved in the case where their welfare is being diminished. In my opinion, I think that it could be due to the negative, rolling impacts of job loss. All the family is affected by this loss, and so the intrinsic behavior of each individual to protect their family can push them to compete in politics in order to enact relevant legislature.
Moreover, the impact of the literacy rate also deserves to be analysed. When it rises by 1%, female participation falls by 5.67%. A potential explanation would be that being literate is strongly correlated with being educated. Thus, in such circumstances, we can assume that households in high literacy level states benefit from better revenues, hence, because men are generally the breadwinners of the family in India, women have fewer incentives to compete in politics because their life-conditions are already satisfactory. However, context can interfere with this logic; indeed, within states where both the GDP and the literacy rate are high, we observe that increasingly, women compete in politics. The fact that women have higher chance there to work for a living and make their own choice can incentivize them to be much more independent than in other states. That independence is strongly correlated with their desire to resolve issues themselves and thus, positively related to competing in a national election.
Moreover, we have to keep in mind that the correlation between the number of candidates, the literacy rate and the unemployment level is low. Thus, even though these coefficients are important, it is far-fetched considering them as fully meaningful over the number of female candidatures.
Now, take a look at the male sample regression. Coefficient results are close to those found for the female sample, however, the intensity of the coefficient is always inferior to the female finding. For instance, an increase by 1% in the level of unemployment rate would result in a rise by 4.71% for the number of men candidatures. Similarly, the effect of the literacy rate does not go beyond 2%. The regression suggest that the male implication is similar regardless of factors; even with endogenous perturbation, these numbers do not vary much. This could be explained by society’s mindset of conferring higher responsibilities to men.
I have obtained the following chart: