4 min read.Updated: 18 Jan 2022, 01:25 AM ISTVidya Mahambare,Sowmya Dhanaraj,Akash Gupta
Varied ministerial styles and preferences are reflected in their speeches but the impact of tonality on audiences awaits research
The annual budget speeches by finance ministers provide a summary of a country’s economic health, its prospects and the government’s vision on public policies that would drive the economy. Since 2017, the Indian finance minister has delivered the budget speech on 1 February in the lower House of Parliament. Earlier, it used to be delivered on the last day of February; 2017 also saw the merging of India’s general and railways budgets. The budget speech is keenly followed by media, analysts and citizens, and is also an opportunity for the incumbent government to impress the broader electorate with its social and economic policies.
Drawing on a part of the analysis from a broader ongoing research project related to economic and social priorities as reflected in budget speeches over time, this article reports insights drawn from an analysis of sentiments in budget speeches since 2000-01. Applying advances in the field of text analysis to the official speech transcripts on the government’s website, we filtered out the most important sentiment signals using the Loughran sentiment lexicon. We first created word lists using this technique and its tools, including cosine similarity, based on historical budget speeches, and then used it to train our algorithm, which identified sentiment-expressing words and phrases within each document.
We derive and divide sentiment from each budget sentence, based on the tone expressed in its words, into either positive, negative, mixed (with both positive and negative words) or neutral sentiment. We then divide sentences for each sentiment by total sentences in the speech to arrive at the proportion of each sentiment. We exclude two interim budgets presented during 2014-15 and 2019-20 to maintain the consistency of analysis. The sentiment analysis presented here is limited to the speech transcripts and does not include the analysis of body language, hand gestures and emotions during the speech delivery.
The accompanying graph plots negative, positive and mixed sentiments as a proportion of the total speech for every budget announced in February of the calendar year before the start of the fiscal year. We find that all six budget speeches of former finance minister P. Chidambaram, most of which were delivered during India’s high-growth phase (2004-05 to 2008-09), were the lowest on sentiments, but also had the least negative tone among all speeches. In contrast, Pranab Mukherjee’s budget speeches as finance minister (2009-10 to 2012-13) and Arun Jaitley’s (2014-15 to 2018-19) were the highest on sentiments. Jaitley’s budgets as finance minister had announcements of several new economic and welfare programmes, such as the smart cities project, Atal Pension Yojna, the Make in India initiative, and in 2017, the goods and services tax (GST).
The budget speeches of 2018-19 and 2019-20, although delivered by two different finance ministers, were almost identical in each tone category and also had the highest positive sentiments among all speeches. Somewhat surprisingly, the February 2009 speech of Mukherjee, delivered at the height of the global financial crisis that hit the world economy in September 2008, was found to have high positive sentiments.
We find that the most balanced speeches, accounting for both positive and negative tones, were by Jaitley (2014-15 and 2016-17) and Yashwant Sinha (2000-01 to 2002-03). In a slight departure, the last budget speech given by Jaitley (2018-19) had the highest net positive sentiment, along with the first two speeches of India’s current finance minister Nirmala Sitharaman.
The net positive tone of the budget speeches in recent years, despite a deceleration in economic growth, might reflect the government’s optimism over its reform process to push the economy onto a high-growth path. However, in the last budget speech, delivered nearly a year after the covid pandemic hit India, the positive tone fell and negative sentiment clearly rose, leading to a much lower net positive sentiment. The rise in negative sentiment was primarily on account of the frequency of sentences related to a contraction in the global and domestic economy and uncertainties and losses due to the pandemic.
We also calculate the complexity metric for each speech using the Flesch readability ease, a widely deployed metric for this purpose. The test score ranges from 0 to 100, and the higher the score, the easier it is considered to read.
The most complex speeches, implying those which are difficult to read and understand, were three consecutive speeches from 2009-10 to 2011-12 by Pranab Mukherjee, with the 2009-10 speech being the most complex with a score of just 11. This speech was made just four months after the collapse of Lehman Brothers in the US, and its heavy use of financial terminology, coupled with complicated sentence framing, contributes significantly to the complexity score of this document.
It should be noted that as India’s budget speech is delivered prior to the start of the financial year it relates to, the impact of unforeseen domestic and global shocks that hit the economy during the financial year can be captured only in the next year’s speech. Further, the sentiment tone of a budget speech is an outcome of a combination of factors such as linguistic preferences that are typically governed by the style and personality of the finance minister and party in power, the state of the economy and its prospects, and the conviction of the government in its policy agenda and reform process.
Overall, however, our analysis points towards the tone of Indian budget speeches reflecting more of the individual style and preferences of a particular finance minister.
It is a matter of further research whether differences in budget-speech tones have any association with the way in which analysts, citizens and markets respond to the speech.
Vidya Mahambare, Sowmya Dhanaraj & Akash Gupta are, respectively, professor of economics, Great Lakes Institute of Management, Chennai; and assistant professor and postgraduate student at Madras School of Economics