5 Apr, 2023
Analyzing Sentiment in Quarterly Earnings Calls Q1 2023
In the below, we used Transcript Sentiment Scores to analyze the performance of the S&P 500 and its constituents. Transcript Sentiment Scores use natural language processing to provide a way to look at earnings call transcripts in a quantitative fashion. S&P Capital IQ Pro provides net positivity, numeric transparency, language complexity, and analyst selectivity ratio metrics for transcripts at the total, speaker, and component level.
Global stock markets had a good start to Q1* 2023 but this was jolted by the news of Silicon Valley Bank’s collapse and the impending effect on the wider global banking industry. This analysis looks at earnings calls that happened prior to the SVB news.
*Q1 refers to the date the earnings call transcript was released not the earnings period it represents.
Net Positivity
The net positivity score is based on the ratio of positive to negative words from the Loughran & McDonald’s (LM) Sentiment Word Lists and compares that to the total number of words.
In Q1 2023, the S&P 500 had a net positivity score of 1.08%, up from the previous quarter score of 0.96%. Over the last few quarters, we’ve seen this score gradually fall but this quarter’s performance was a bump back above the trailing 4 quarter average of 1.05%. Improving sentiment was widespread with 9 of the 11 GICS sectors seeing an improving score from the previous quarter. Consumer Staples and Consumer Discretionary posted the highest net positivity scores at 1.37%, but it is interesting to note that Consumer Staples stocks are seeing in improving trend to their scores over the last few quarters versus Consumer Discretionary scores that have been trending lower. Given the backdrop of increased recession fears, this would seem to be reasonable. The lowest sector scores were Real Estate at 0.83% and Financials at 0.75%. However, both sectors did improve their score from the previous quarter. Looking at some notable individual stocks, Adobe (Information Technology) and IDEXX Laboratories (Health Care) had the highest positivity in the S&P 500 with scores of 2.55% and 2.51% respectively. The Allstate Corporation and PNC Financial Services (both Financials) were the weakest with scores of -0.60% and -0.44% respectively.
Numeric Transparency
The numeric transparency score is the ratio of numbers to words. A higher value means more use of numbers relative to words, which signifies a higher level of transparency. This is considered more objective and precise and thus, more favorable.
Overall, in Q1 2023, the S&P 500 had a numeric transparency score of 2.64%, improving from its previous quarter score of 2.25%. This current score is the highest of any of the last four quarters which has an average score of 2.27%. A higher score means more transparency so it’s nice to see this number trending in the right direction and with all sectors improving their score from the previous quarter. Looking at the individual sectors, Utilities received the highest score for numeric transparency with 3.16%, better than its previous quarter score of 2.81% and above the previous four quarter average of 2.68%. Communication Services had the weakest score of 2.01%, but again, this was an improvement from the previous quarter (1.97%) and also higher than the four quarter average (1.94%). Looking at some notable individual stocks, Loews Corporation (Financials) and Monster Beverage Corporation (Consumer Discretionary) had the highest positivity in the S&P 500 with scores of 6.21% and 6.08% respectively. The Hershey Company and The Clorox Company (both Consumer Staples) were the weakest with scores of 0.74% and 0.88% respectively.
Language Complexity
We use the Gunning Fog Index as a proxy for language complexity. Each score can be interpreted as the number of years of formal education a person needs to understand the text on the first reading. A lower score denotes simpler language and is viewed favorably.
In Q1 2023, the S&P 500 had a language complexity score of 12.44, up slightly from the previous quarter score of 12.36 and also higher than the previous four quarter average of 12.35. While there is not much variance between the different sectors, Industrials had the lowest score (more favorable) of 12.00 with the Utilities sector having the highest score (less favorable) of 13.11. Looking at some notable individual stocks, DISH Network Corporation (Communication Services) and DXC Technology Services (Information Technology) had the lowest language complexity scores of 9.46 and 9.68 respectively. American Water Works Company and NextEra Energy (both Utilities) had the highest language complexity with scores of 16.10 and 15.99 respectively.
Analyst Selectivity Ratio
An analyst selectivity ratio is the percent of active analysts covering the stock that are allowed to ask questions during the call. A higher value is viewed favorably, and scores will range from 0% to 100%.
Overall, for Q1 2023, the S&P 500 had an analyst selectivity ratio of 42.27%, down from the previous quarter (42.43%), and below the previous four quarter average of 42.46%. For the individual sectors, Real Estate had the highest percentage of active analysts allowed to ask questions during earnings calls with 51.24%, which was significantly higher than its previous quarter score of 45.87%, ranking 6th best out of the 11 sectors. After Industrials, Materials was the next highest at 50.36%. The sector with the lowest analyst selectivity ratio was Communication Services at 27.85%. This was a slight improvement from last quarter (26.72%) but still below its previous four quarter average (29.58%). Looking at some notable individual stocks, UDR (Real Estate) and Gartner (Information Technology) had the highest analyst selectivity scores of 100% and 89% respectively. Amgen (Health Care) and Etsy (Consumer Discretionary) had the lowest analyst selectivity with both companies having a score of 0% indicating that active analysts did not ask questions on the call.
Source: S&P Capital IQ Pro. Data as of March 27, 2023.
S&P Capital IQ Pro provides Net Positivity, Numeric Transparency, Language Complexity, and Analyst Selectivity Ratio metrics for transcripts at the: 1) Total level, 2) Speaker level, such as Executive or Analyst, and 3) Component level, such as Presentation Operator message, Presenter speech, Question, or Answer. Scores for companies are based on the most recent transcript of an earnings call in the calendar quarter that the event occurs. Scores are typically available 3 hours after a transcript is published to CIQ Pro. Scores are most effective when used to determine trends in sentiment by comparing current scores for a company vs. the prior quarters’ scores of the same company.
References:
Zhao, F. "Natural Language Processing – Part II: Stock Selection" (September 2018). Natural Language Processing – Part II: Stock Selection | S&P Global Market Intelligence (spglobal.com)
Zhao, F."Natural Language Processing – Part III: Feature Engineering" (January 2020). Natural Language Processing – Part III: Feature Engineering | S&P Global Market Intelligence (spglobal.com)