TL;DR
This study explores how the language used in hospital CEO messages on their websites correlates with financial performance. By combining text-mining with the balanced scorecard framework, researchers found that specific word choices can indeed predict financial indicators like liquidity and reserves, offering a novel way to assess hospital health beyond traditional financial statements.
What Problem Does This Paper Address?
Hospitals, especially in countries like Korea, operate under a public good mandate, often facing financial challenges due to high costs and regulatory limitations. This creates a delicate balancing act: they need to demonstrate public value without necessarily revealing financial weaknesses. CEOs, therefore, have an incentive to strategically use their website's introductory messages (like the "CEO Greeting" or "About Us" section) to shape public perception. This paper addresses the question of whether the qualitative information presented in these CEO messages can serve as a meaningful indicator of a hospital's underlying financial status, a connection that has been underexplored.
Methods
The researchers employed a two-pronged approach:
- Text Mining: They used a text-mining technique to extract keywords from the "About Us" or "CEO Greeting" sections of 65 Korean hospital websites. Specifically, they calculated the Term Frequency-Inverse Document Frequency (TF-IDF) for each word. TF-IDF is a standard metric that highlights words that are frequent within a specific document (hospital's message) but relatively rare across a larger corpus (all hospital messages), thus indicating their importance and distinctiveness in that particular context. They focused on nouns to ensure lexical meaning.
- Balanced Scorecard Framework: The extracted keywords were then categorized using a modified balanced scorecard framework. Instead of the traditional four perspectives (finance, customer, internal process, learning & growth), this study adapted it for non-profit hospitals by replacing "finance" with "public" (PB). The remaining perspectives were "patient" (PT), "internal process" (IP), and "learning & growth" (LG). Words that didn't fit into these categories were classified as "neutral" (NT) and excluded from the main analysis.
Finally, a t-test was used to compare the mean TF-IDF scores for words within each perspective between "high" and "low" groups of hospitals, defined by their financial ratios. This allowed them to assess if the frequency of certain types of words was significantly associated with specific financial outcomes.
Datasets
The study utilized data from 65 senior public general hospitals in Korea, designated as such by the Ministry of Health and Welfare as of 2018.
- Text Data: Introductory text from the "About Us" or "CEO Greeting" sections of these hospitals' websites.
- Financial Data: Financial statements were collected from the Hospital Accounting Standard System provided by the Korean Health Industry Development Institute.
The paper does not introduce a new benchmark dataset but rather uses publicly available website text and financial data from a specific set of hospitals.
Key Insights and Ideas
- CEO Messages as Financial Proxies: The core idea is that the language used by hospital CEOs isn't just fluff; it can be a quantifiable indicator of financial health. This opens up new avenues for analyzing organizational performance beyond traditional financial metrics.
- Strategic Language Use: Hospitals with different financial profiles seem to strategically emphasize certain aspects in their CEO messages. For instance, those with lower financial liquidity might focus more on "public value" to deflect attention from financial concerns.
- Balanced Scorecard Adaptation: Modifying the balanced scorecard for non-profit hospitals by prioritizing "public" over "financial" is a crucial adaptation that reflects the unique operational context of healthcare institutions.
- Text Mining for Qualitative Insights: The study demonstrates the power of text mining to extract meaningful, quantifiable insights from unstructured qualitative data, bridging the gap between text and numbers.
Key Results
The study found several significant associations:
- Public Perspective (PB): Hospitals using more "public-related" words showed significantly higher liquidity (current ratio). This suggests that emphasizing public value might be a strategy for hospitals with strong liquid assets.
- Patient Perspective (PT): A higher emphasis on "patient-related" words was significantly associated with a higher accumulating reserves for proper business purposes ratio. This implies that hospitals planning for future investment might highlight patient satisfaction.
- Internal Process Perspective (IP): Interestingly, a higher use of "internal process" words was negatively associated with hospital size. This suggests that smaller hospitals might use language emphasizing operational efficiency to compensate for their size.
- Learning & Growth Perspective (LG): Hospitals using more "learning and growth" words showed significantly lower ratios for debt ratio, fixed-asset turnover ratio, asset turnover ratio, and accumulating reserves for proper business purposes. This indicates that hospitals facing financial challenges (e.g., low leverage, inefficient asset use) might focus on future growth narratives.
Limitations and Future Work
- Sample Size: The authors acknowledge that the sample size of 65 hospitals is relatively small, which may limit the generalizability of the findings.
- Hospital Type: The study focused only on senior general hospitals, meaning the results might differ for other types of healthcare facilities.
- Future Work: The authors suggest that future research could expand the analysis to include smaller hospitals and explore the relationship between CEO messages and financial performance in greater detail. They also highlight the need for further research into the text-mining methodology within the hospital management domain and emphasize the importance of qualitative data disclosure online.
Summary updated