Employment Effects of Technological Progress in U.S. Healthcare: Evidence from Listed Companies

Abstract

Employment is the foundation of social stability and a key factor for economic stability and sustainable development. With the rapid advancement of technology, the impact of technological progress on employment has become a focal point of academic attention. As an emerging industry, the healthcare sector has experienced rapid growth in recent years, driven by the widespread application of scientific and technological innovations. However, at the same time, these advancements have also exerted a significant influence on employment within the healthcare sector. To address this issue, this paper utilizes panel data from publicly listed healthcare firms in the United States between 2013 and 2023. It innovatively measures technological progress through Total Factor Productivity (TFP) and employs a two-way fixed effects model to empirically analyze the impact of technological progress on employment in the healthcare sector from a microeconomic perspective. The findings indicate that a 1% increase in technological progress in the healthcare sector leads to an average 0.116% rise in employment levels. This conclusion remains robust after conducting rigorous robustness checks and addressing endogeneity concerns, with the output effect playing a significant role in this process. Heterogeneity analysis indicates that technological progress significantly promotes employment across various sub-sectors, though the magnitude of this effect varies only slightly among industries. Furthermore, the employment-promoting effect of technological progress is more pronounced in larger firms and those with a higher proportion of fixed assets. Therefore, policies should actively support the improvement of technology levels and management efficiency within the healthcare sector, fully leveraging the potential of technological progress to promote employment, and achieve sustainable development for both the healthcare sector and societal employment.

1. Introduction

With the continuous increase in demand for health and the accelerating trend of population aging, the global health industry has developed rapidly, with the market size expanding and industry connotations diversifying. According to the Global Wellness Institute (GWI) 2019–2022 Global Wellness Economy Monitor, the global health industry market reached USD 5.6 trillion in 2022, accounting for 5.6% of global GDP. It is expected that by 2027, this market will grow to USD 8.5 trillion, representing 6.6% of global GDP. Especially after the COVID-19 pandemic, the health industry has seen rapid growth with an average annual growth rate of 12.1%, far exceeding global GDP growth, and has driven significant job creation, becoming a key driver of both global economic and employment growth. North America, particularly the United States, has long been the largest market for the global health industry, and its development deserves attention. According to the U.S. Bureau of Labor Statistics (BLS), in 2023, the healthcare sector added 654,000 jobs, nearly a quarter of all new jobs in the U.S., and it is projected that the sector will create about 45% of all job gains from 2022 to 2032. Healthcare staffing needs are expected to continue supporting overall employment growth in the near future.

On the one hand, according to the neoclassical growth model, technological progress is a crucial factor for economic growth [1], enhancing corporate revenue and promoting industrial development. In recent years, technological advancements have become the core driver of growth in the healthcare sector. Upgrades in medical devices and innovative applications of biomedicine have made the manufacturing of healthcare products more efficient, while technologies like big data and cloud computing have made healthcare services and management more convenient. These technological advancements have not only driven the intelligent and precise development of the healthcare sector but have also fostered deeper integration of the healthcare sector with industries such as culture and tourism, bringing new development opportunities. On the other hand, employment is the cornerstone of economic sustainability and social stability. According to Romer’s (1986) endogenous growth theory, technological progress is determined by internal factors within the economic system, such as human capital [2], and it also impacts employment. Automation technologies have helped the production sectors of the healthcare sector save labor, but digital health technologies have also created new occupations such as health managers and health analysts, compensating for job losses. Given the continuous rise in health demand and the rapid growth of the healthcare sector, how does technological progress affect employment in this sector? Does it promote or reduce employment? Addressing this issue will help us gain a better understanding of technological advancements in the healthcare sector, promote its sustainable development, and ensure the stable growth of employment, which holds significant practical importance.

Additionally, the academic community has yet to reach a consensus on the impact of technological progress on employment. Some scholars argue that technological progress leads to a reduction in jobs, while others believe it creates new employment opportunities and increases overall employment levels. Furthermore, some scholars point out that the impact of technological progress on employment varies across different industries. Despite recent debates in academia surrounding the relationship between technological progress and employment, especially with the rise in digital and AI technologies, research on this relationship in specific industries remains limited. The healthcare sector, with its large market, extensive employment, wide application of new technologies, and numerous sub-industries, is undoubtedly an excellent subject for study. Understanding the relationship between technological progress and employment in the healthcare sector, clarifying its mechanisms, and analyzing differences between sub-industries will provide new empirical insights into this topic, contributing significantly to theoretical research.

To answer these questions, this paper focuses on the U.S. healthcare sector and selects publicly listed companies as a sample to analyze the specific impact of technological progress on employment. Compared to previous studies, this paper narrows its focus to the healthcare sector, filling the gap in micro-level research on the relationship between technological progress and employment and enriching the empirical literature on the healthcare sector. This research is of great significance for understanding technological progress in the healthcare sector, promoting sustainable development of the industry, and ensuring stable employment growth. The structure of this paper is as follows: Section 2 provides a literature review on the impact of technological progress on employment and the healthcare sector; Section 3 introduces the theoretical framework and econometric models; Section 4 presents the data; Section 5 discusses the empirical results in detail; and Section 6 concludes with a summary of the findings and policy recommendations.

2. Literature Review

2.1. Research on Technological Progress and Employment

Research on the relationship between technological progress and employment has not yet reached a consensus in academia, but it is widely recognized that technological advancements have both substitution and compensation effects on employment. The net effect on employment depends on the relative magnitude of these two effects. On one hand, some scholars argue that technological progress generally leads to a decrease in employment, with the substitution effect being dominant. Technological changes and skill upgrades are considered major factors contributing to the decline in labor share [3]. In the early stages of technological advancement, high demand elasticity may stimulate job growth. However, as demand becomes saturated, demand elasticity decreases. Therefore, even though technological progress can enhance productivity and reduce prices, it may not lead to an increase in demand [4]. With the widespread application of technologies such as artificial intelligence and industrial robots, technological progress is increasingly seen as inhibiting employment growth [5,6], particularly in manufacturing, where technological advancements show a significant negative correlation with workforce size [7]. On the other hand, some scholars adopt an optimistic view, suggesting that technological progress can promote employment, with the compensation effect being dominant. Although technological change has replaced many routine tasks, it also generates compensatory effects through channels such as lowering product prices and increasing product demand, ultimately contributing to job growth [8]. While technological innovations (such as automation) directly lead to job losses in certain industries, their cross-sectoral effects and final demand effects have a positive impact on employment, helping to offset the negative direct effects and foster job creation [9,10]. Therefore, technological progress does not result in an overall contraction of the labor market but rather brings about structural transformation [11,12,13]. In particular, the increase in productivity can effectively mitigate the negative impacts of rising labor costs, and high-productivity firms may continue to hire more workers [14].

Technological progress not only affects the scale of employment but also has profound impacts on employment structure and quality, which has been extensively studied in empirical research. First, the impact of technological progress on labor with different skill levels varies. Technological progress is skill-biased [15,16], typically increasing the demand for high-skilled labor while reducing the demand for low-skilled labor [17]. During the process of technological advancement, highly educated, high-skilled workers are more likely to secure employment opportunities and high income [18,19,20,21], whereas low-skilled workers, particularly low-skilled male workers, experience a significant decline in real wages [22,23]. Goos et al. argue that the primary effect of technological progress is the reduction in medium-skilled jobs, as technological advancements tend to automate routine tasks. High-skilled workers in automation-related fields and low-skilled workers performing simple physical tasks are less likely to be replaced, while medium-skilled workers are more easily replaced by machines. This phenomenon, known as “job polarization”, is widespread in Europe, exhibiting both within-industry and between-industry polarization [24]. Secondly, technological progress affects different industries differently. Manufacturing tends to be capital-intensive with standardized and process-driven production [25,26], whereas service industries are labor-intensive with more flexible and interactive production processes [27]. In manufacturing, technological progress may lead to job reductions, while in the service sector, it may conversely stimulate employment [28]. Technological innovation will help achieve the green development of enterprises, which in turn will affect the employment structure [29]. The application of new technologies often results in a shift in employment from manufacturing to services, with complementary job roles such as managers and technical scientists benefiting, while routine task-based jobs, such as machine operators, face income declines [30]. Based on this, some scholars have further analyzed the specific occupations impacted by technological progress. Frey and Osborne, using machine learning and Gaussian processes, estimated the likelihood of 702 occupations being computerized in countries like China, India, and the United States [31]. West pointed out that the widespread application of robots and artificial intelligence will create numerous new jobs in fields such as technical support and data analysis [32]. Additionally, technological progress also affects employment quality. For example, the widespread use of computer technology has increased the number of low-skilled jobs, but has also led to a decline in job quality, particularly in terms of reduced autonomy and slightly increased work intensity [33]. Technological change may also lead to skill degradation among workers, affecting their interaction with customers and the openness of the work environment. Therefore, comprehensive measures should be adopted to balance technological efficiency with employee well-being in order to achieve sustainable improvements in job quality [34].

2.2. Research on Healthcare Sector

Technological progress has a significant impact on the development of the healthcare sector, and many scholars have explored this issue from a theoretical perspective. Technology is a driving force for the growth of the healthcare sector [35], with automation technologies helping hospitals improve operational efficiency by using network connectivity for real-time monitoring and adjustments, ensuring the effectiveness and precision of healthcare services [36,37]. Through technological innovation and process optimization, high costs have been effectively addressed, driving the transformation of the healthcare sector [38]. Additionally, technological advances have simplified daily administrative tasks, allowing doctors to rely more on non-medical professionals for preliminary diagnoses and routine care, thereby reducing their workload and enabling them to focus more on clinical practice [39].

While technological progress can improve the quality and efficiency of healthcare services, it may also lead to significant increases in expenditures, particularly in the use of high-cost medical equipment and pharmaceuticals [40]. As a result, many scholars have conducted cost-effectiveness analysis (CEA) and cost–benefit analysis (CBA) of technological applications [41,42], and some scholars have evaluated the effects of health policies [43], which constitutes the primary direction of current quantitative research in health economics [44]. Some scholars also point out that the risks associated with technological progress should not be ignored. Over-reliance on technology could affect the quality of healthcare services, exacerbate health inequalities, and raise concerns about data security [45].

Technological progress also has significant implications for employment in the healthcare sector. On one hand, technological advancements may replace some traditional low-skill jobs, particularly in areas such as nursing, patient handling, and pharmaceutical delivery. However, the demand for high-skill positions, such as medical robot operators, maintenance technicians, and developers, is also increasing, especially in hospitals and specialized treatment sectors [46,47]. Therefore, the adaptability of the labor market, skills training, and policy support will be critical to ensuring that technological transformation brings positive impacts to the development of the healthcare sector [48]. On the other hand, some scholars argue that healthcare-related work, due to its inherent need for interpersonal interaction, is difficult to outsource or automate. Low-skilled healthcare workers (such as nursing assistants and community health workers) remain an important part of health systems in many countries, and demand for these roles is expected to grow steadily in the future. Healthcare workers and primary care providers will continue to be vital pillars of health systems [49]. Hofmarcher further notes that, despite the continuous increase in healthcare sector jobs, improving labor productivity is key to ensuring the sustainable development of the healthcare sector [50].

2.3. Literature Summary

The review above indicates that although existing research on the relationship between technological progress and employment is relatively abundant, much of it focuses on macro-level areas such as national economies and manufacturing, with fewer studies addressing specific industries or sectors. Furthermore, most research suggests that the impact of technological progress on employment is complex and requires analysis based on the characteristics of different industries and sectors. It is also worth noting that quantitative studies on technological advancements in the healthcare sector mainly focus on benefit analysis, while research on the relationship between technological progress and employment remains largely theoretical, lacking empirical analysis of employment issues.

Building on the above analysis, this paper adopts a micro-level perspective of the healthcare sector, using publicly listed companies in the U.S. healthcare sector as a research sample to investigate the impact of technological progress on employment within the sector. Compared to existing research, the marginal contributions of this paper are threefold: First, it explores the specific impact of technological progress on employment from a micro perspective within the healthcare sector, filling the gap in the current literature regarding micro-level studies on the relationship between technological progress and employment, while enriching empirical research in the healthcare sector. Second, given the lack of empirical studies on the pathways through which technological progress affects employment, this paper combines output effects and capital deepening effects to validate two potential pathways. Third, this study further divides the healthcare sector into sub-sectors and conducts a heterogeneity analysis based on industry type, company size, and fixed asset ratios, revealing the differences in how various types of enterprises within the sector respond to technological progress in terms of employment.

3. Theory and Model

3.1. Theoretical Analysis

Neoclassical economics views technological progress as an important variable in the production function, believing that technological progress directly leads to economic growth, which in turn creates new employment opportunities and drives an increase in employment. Technological progress also leads to a decrease in price levels, raises real wage levels, and stimulates new consumer demand, thereby promoting employment. This paper primarily conducts a theoretical analysis of the impact of technological progress on employment based on the neoclassical growth model.

Firstly, following the approach adopted by most of the existing literature and based on the model fitting optimization with the data used in this study, this paper assumes that firms in the U.S. healthcare sector adhere to a Cobb–Douglas production function form:

$$Y=A{K}^{\alpha }{L}^{\beta }$$

(1)

In this equation, Y represents the output level of the firm, A denotes technological progress, L refers to labor input, and K represents capital stock. α and β are the elasticity coefficients of capital and labor inputs, respectively, which are generally stable and are often treated as constants.

Next, we introduce the firm’s profit function:

$$\pi =pY-wL-rK$$

(2)

In this equation, π represents the firm’s profit, while p, w, and r denote the price level, wage level, and interest rate level, respectively.

The first-order condition for profit maximization of the firm is as follows:

$$\partial Y/\partial L=w/p$$

(3)

Then, taking the partial derivative of Equation (1) with respect to L, we obtain

$$\partial Y/\partial L=A{K}^{\alpha }{L}^{\beta -1}$$

(4)

By simultaneously solving Equations (3) and (4), we can derive the expression for labor demand:

$$L={\displaystyle \frac{\beta Y}{w/p}}$$

(5)

From Equation (5), it is evident that labor demand can be expressed as a function of output level and wage level. Since the elasticity coefficient β is greater than 0, this implies that labor demand is positively correlated with output level and negatively correlated with wage level.

Finally, taking the natural logarithm of both sides of Equation (5) while holding the price level constant, we obtain the basic model of employment:

$$lnL={\beta }_0+{\beta }_{1}lnY+{\beta }_{2}lnw+\epsilon$$

(6)

As indicated by Equation (6), technological progress primarily affects employment through its influence on output levels. On the one hand, technological progress directly enhances production efficiency and output, which helps lower production costs and improves firm profitability, thereby encouraging firms to expand production and increase employment [51]. On the other hand, technological progress can create new consumer demand by improving the quality of existing products and by developing new products and services, which leads to the emergence of new business models and job opportunities [52]. Technological progress also stimulates consumer demand by reducing the price of individual products and increasing consumers’ real income, which facilitates the expansion of firm size and the creation of new employment opportunities [53]. Furthermore, according to endogenous growth theory, technological progress also affects the scale of capital, which indirectly influences output levels and, consequently, employment. Additionally, Equation (6) suggests that wage levels are an important factor affecting employment.

Given that this paper primarily focuses on the impact of technological progress in the healthcare sector on employment, in order to more directly examine the overall effect of technological progress on employment, we build upon the basic model in Equation (6). We introduce other relevant variables while excluding the output level from the basic model. The effect of output level on employment is treated as an intermediary effect of technological progress and will be discussed in the subsequent mechanism analysis.

3.2. Econometric Model

Since this paper focuses on the micro-level analysis of the healthcare sector, and given that the two-way fixed effects (TWFE) model is commonly used in microeconomic studies, it allows for the control of individual characteristics that do not change over time and time trends that do not vary across individuals. This approach effectively mitigates potential bias that might arise from omitting these factors, thus enabling a more accurate estimation of the net impact of technological progress on employment. Based on the previous analysis of the basic employment model, this paper ultimately adopts a model with two-way fixed effects (controlling for both individual and year effects) to examine the impact of technological progress on employment in the healthcare sector. The specific model is as follows:

$$ln{l}_{it}={\beta }_0+{\beta }_{1}lntf{p}_{it}+{\displaystyle \sum {\beta }_k}{X}_{ijt}+{\sigma }_i+{\lambda }_t+{\mu }_{it}$$

(7)

In this model, lntfp_it represents the core explanatory variable, which refers to technological progress in listed firms within the healthcare sector. lnl_it is the dependent variable, representing employment levels. X_ijt denotes the control variables, σ_i represents the individual firm effects, λ_t refers to the time fixed effects, and μ_it is the error term.

4. Data

4.1. Data Sources and Processing

Unlike previous studies that have primarily focused on the macroeconomic level or the manufacturing sector, this paper is the first to shift the research focus to the healthcare sector. To ensure data quality and reliability, and to enhance the robustness and credibility of the empirical results, it is crucial to select appropriate micro-level samples and perform corresponding data processing. The Osiris database provides detailed financial and operational information on publicly listed companies worldwide and is widely used in economic and management research. Given that the United States has been an early adopter of healthcare sector development and has a relatively well-established sector, it offers a comprehensive reflection of the overall landscape of the healthcare sector. Therefore, this paper selects the U.S. Health Care Sector under the Global Industry Classification Standard (GICS) from the Osiris database as the research subject, with publicly listed firms within this sector serving as the study sample. The specific selection and processing procedure is as follows:

Firstly, since the Osiris database suffers from significant missing data on employee numbers for listed companies prior to 2012, and to ensure the availability of the dependent variable—employee numbers for listed firms—this paper defines the sample period as 2013–2023. From this, 307 publicly listed firms with complete employee data and active operations were initially selected.

Secondly, after completing the above step, this study further removed firms with anomalous data, such as those with zero or negative revenue, firms with fewer than 10 employees, firms that were newly established or listed late in the sample period, and firms exhibiting large data fluctuations, which may indicate potential operational issues. After this cleaning process, 172 firms remained, forming the final sample for the panel data analysis.

Finally, in order to eliminate the impact of price level changes and more accurately reflect the relationship between healthcare sector data and employment, all data used in this study were deflated to 2013 prices. The deflation index used is the average quarterly GDP deflator for each year, published by the Bureau of Economic Analysis (BEA).

Additionally, for the data with significant missing values in the Osiris database, particularly for certain firms’ operating costs, including general, sales, and administrative expenses (SG&A) and R&D expenditures, this paper supplemented the missing information by referring to the financial statements of the companies published on the Securities and Exchange Commission (SEC) website.

Regarding the measurement of technological progress, this paper follows the methodology of Autor and Salomons and innovatively introduces Total Factor Productivity (TFP) into the empirical analysis at the micro-level of firms [9], using it as the metric for technological progress. The advantages of using firm-level TFP are as follows.

It can capture implicit technological progress and comprehensively reflect the contribution of various factors such as technological innovation, managerial efficiency, corporate governance, and scale effects to a firm’s productivity. Compared to traditional measures based solely on R&D expenditures or patent applications, TFP provides a more comprehensive and timely framework for measuring technological progress. This allows for a more complete capture of the impact of technological progress on employment at the firm level. Firm-level TFP, as a standardized indicator, can also facilitate cross-firm and cross-industry comparisons. In contrast, patents and R&D expenditures may be influenced by industry characteristics or firm size. By using firm-level TFP as the measure of technological progress, this study can explore the dynamic effects of technological progress in healthcare sector firms from a microeconomic perspective. This approach addresses gaps in the existing literature regarding the specific application of these measures within the sector.

Currently, there are two main approaches for calculating TFP: frontier analysis methods and non-frontier analysis methods. One representative of the frontier analysis method is Data Envelopment Analysis (DEA), which has the advantage of not relying on specific production function assumptions, making it suitable for evaluating relative efficiency. However, the limitation of the DEA method is that it cannot perform hypothesis testing because it does not set a production function and does not account for random errors. As a result, it can only provide estimates of relative efficiency and cannot reveal the absolute level of technological progress. In contrast, non-frontier analysis methods, such as the Solow residual method and its improvements, directly perform Ordinary Least Squares (OLS) estimation on the Solow model, treating the residuals as an indicator of technological progress. However, in micro-level studies, this approach is prone to severe endogeneity bias, which affects the reliability of the results.

To overcome these issues, the OP (Olley–Pakes) and LP (Levinsohn–Petrin) methods, as improved non-frontier analysis techniques, have gradually gained widespread use. The OP method overcomes the endogeneity bias problem of the Solow residual method by introducing firm investment as a proxy for unobservable productivity shocks. On the other hand, the LP method uses intermediate inputs as a proxy variable. While both methods offer significant advantages, Wooldridge and Ackerberg et al. argue that the OP method is superior to the LP method in addressing endogeneity issues in microeconomic data, particularly in the dynamic analysis of resource misallocation and capital inputs [54,55].

Therefore, this paper selects the OP method to calculate firm-level TFP and uses it as a proxy for technological progress. Specifically, the investment data required for the OP method comes from firms’ cash flow statements, particularly capital expenditures (purchases of fixed assets). This type of data is relatively complete for the sample period, with only a few firms experiencing missing data or zero investment in certain years. However, these issues do not lead to significant sample loss, making the use of the OP method to calculate TFP both feasible and reliable in this study.

4.2. Main Variables and Explanations

(1)

Dependent Variable

Employment level: The number of employees at the end of the fiscal year for listed companies is selected as the dependent variable, and the natural logarithm of this value is taken for analysis.

(2)

Independent Variable

Technological progress: Following the approach of Autor and Salomons (2018) [9], firm-level TFP is chosen as a proxy for technological progress, which is calculated using the OP method.

(3)

Control Variables

In studies on employment issues in enterprises, most scholars set similar control variables. This paper primarily references the research of Ni B and Obashi on enterprise employment [56] and incorporates the following control variables:

① Firm Size: Generally, the larger the firm, the more employees it can accommodate. Larger firms tend to have more resources to hire and sustain employment.

② Firm Age: The age of a firm reflects the length of time it has been in operation. Older firms often have a better reputation and more stability in the industry. Therefore, the older the firm, the more likely it is to attract employees and maintain employment opportunities.

③ Per Capita Operating Costs: The level of wages directly influences employment. However, most firms’ detailed data on employee salaries and benefits are not included in the database or financial statements. A commonly used proxy for wage levels—employee compensation—is also severely missing. Employee wages are generally included in operating costs under sales, general, and administrative expenses (SG&A). Since employee salaries are a significant part of operating costs, and from a cost control perspective, high operating costs, particularly high SG&A expenses, often become key factors in decisions like reducing wages or even laying off employees. Therefore, per capita operating costs are used as a proxy to describe the impact of costs, including wages, on employment.

④ Return on Assets (ROA): The return on assets is the ratio of net profit to total assets. It indicates the level of return a firm achieves from its total assets and is an important indicator of a firm’s profitability and growth prospects. A higher return on assets typically signals better performance and prospects, making it an important factor for attracting labor.

⑤ Debt-to-Asset Ratio: The debt-to-asset ratio reflects the proportion of a firm’s total assets that is financed through debt. An excessively high or low debt level can affect a company’s decision-making and operational performance, which in turn influences labor demand.

The descriptions of the main variables are shown in

Table 1. Description of main variables.

Types of Variables	Variable Names	Symbol	Representation
Dependent Variable	Employment	lnl	The number of employees at the end of the year is taken and log-transformed.
Independent Variable	Technological Progress	lntfp	The log-transformed value of the firm’s Total Factor Productivity.
Control Variables	Firm Size	lnas	The log-transformed value of the firm’s total assets.
	Firm Age	lnage	The log-transformed difference between the sample year and the firm’s founding year.
	Per Capita Operating Costs	lnaoc	The log-transformed ratio of the firm’s selling, general, and administrative expenses (SG&A) to the number of employees.
	Return on Assets	roa	The ratio of the firm’s net profit to total assets.
	Debt-to-Asset Ratio	lev	The ratio of the firm’s total liabilities to total assets.

below:

Table 2. Descriptive statistics of the key variables.

Variables	Obs	Mean	Sd	Median	Min	Max
lnl	1892	7.479	2.294	7.313	2.398	12.713
lntfp	1892	5.807	1.009	5.821	−0.182	9.676
lnas	1892	13.770	2.216	13.584	8.086	19.151
lnage	1892	3.157	0.573	3.135	0.000	4.812
lnaoc	1892	4.614	0.978	4.730	0.782	7.170
roa	1892	−0.097	0.383	0.023	−4.196	2.073
lev	1892	0.546	0.539	0.487	0.000	9.524

Note: The unit for employment is the number of employees, while the units for total assets, operating costs, research and development expenditures, and operating income are in thousands of US dollars.

presents the descriptive statistics for the key variables in this study. As shown in the table, the mean value of firm-level Total Factor Productivity (lntfp) is 5.807, with a median of 5.821. The close proximity of these values indicates that the average level of technological progress in the healthcare sector is relatively high and that the distribution of the sample is fairly symmetric. Similarly, the mean and median values for employment (lnl) and total assets (lnas) are also close, further suggesting that the sample exhibits a generally balanced distribution and a sound overall structure. However, the wide range observed in both lnl and lntfp indicates substantial variability in employment levels and technological progress across firms within the healthcare sector, highlighting significant individual effects. This finding supports the appropriateness of using micro-level data and a two-way fixed effects model, as the fixed effects model effectively controls for individual heterogeneity, thereby providing a more accurate estimation of the impact of technological progress on employment.

Furthermore, the maximum and minimum values for return on assets (roa) and leverage (lev) show considerable disparities. Notably, the leverage variable exhibits a minimum value of 0 and a maximum value of 9.524, suggesting the presence of firms with no debt and firms with a leverage ratio exceeding nine times their assets—indicative of “outlier” firms operating under abnormal conditions. These extreme values may result from the presence of a small number of anomalous observations in the data. To ensure the robustness of the results, these outliers will be addressed in subsequent robustness checks.

It is also noteworthy that the negative mean value of roa suggests that the average profitability in the healthcare sector is not particularly favorable. This could be due to the fact that, despite the relatively early development of the U.S. healthcare industry, a large influx of firms in the sector has only occurred after the 2008 financial crisis, with many of these emerging firms still in the early stages of operation, characterized by high investment and low returns.

5. Results and Analysis

5.1. Benchmark Regression Results

In

Table 3. Benchmark and robustness regression results for the impact of technological progress on employment in the healthcare sector.

Variables	lnl
	(1)	(2)	(3)	(4)
lntfp	0.098 **	0.116 ***		0.186 ***
	(0.045)	(0.028)		(0.035)
lnrd			0.218 ***
			(0.056)
lnas		0.608 ***	0.467 ***	0.623 ***
		(0.027)	(0.046)	(0.027)
lnage		0.022	0.051	0.040
		(0.119)	(0.135)	(0.107)
lnaoc		−0.501 ***	−0.409 ***	−0.530 ***
		(0.075)	(0.072)	(0.081)
roa		−0.280 ***	−0.137 ***	−0.492 ***
		(0.040)	(0.043)	(0.076)
lev		0.064 *	0.063 **	0.203 ***
		(0.034)	(0.030)	(0.054)
_cons	6.443 ***	0.477	−0.072	−0.115
	(0.273)	(0.605)	(0.655)	(0.599)
Firm fixed effects	control	control	control	control
Time fixed effects	control	control	control	control
Observations	1892	1892	1507	1848
Adj. R2	0.331	0.756	0.777	0.767

Note: The values in parentheses represent cluster-robust standard errors; ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.

, Columns (1) and (2) report the benchmark regression results. Column (1) includes only individual fixed effects and year fixed effects, presenting the estimates from a univariate regression. Column (2), on the other hand, considers various other factors and includes them as control variables in the regression. The regression results in Column (1) show that in the univariate regression, the coefficient of lntfp is significantly positive at the 5% statistical level, indicating that technological progress in the healthcare sector has a notable positive effect on employment. After incorporating control variables, the regression results, as shown in Column (2), indicate that the coefficient of lntfp increases and is significant at a higher statistical level of 1%. This suggests that technological progress in the healthcare sector can significantly promote employment. Specifically, a 1% increase in the TFP level is associated with a 0.116% increase in employment. This empirical finding addresses the research question posed earlier, fills the gap in the literature on the relationship between technological progress and employment in the healthcare sector, and provides new empirical evidence for the existing theories of technological progress and employment.

The benchmark regression results also show that the size of firms in the healthcare sector has a positive effect on employment, which is consistent with general experience and the findings of most studies—larger firms tend to have a higher demand for labor. Although the coefficient for firm age is positive, it is not statistically significant, which contrasts with certain theoretical expectations suggesting that older firms should have more resources to expand labor demand. A possible explanation for this is that the healthcare sector is still in a phase of continuous expansion, comprising both emerging and established firms, each with its own advantages in absorbing employment. Therefore, firm age may not be the key factor influencing employment. The coefficient for the log-transformed per capita operating costs (lnaoc) and the return on assets (roa) are both negative, suggesting that firms with higher operating costs and capital returns tend to reduce labor hiring. This phenomenon may be related to the fact that firms in the healthcare sector, when facing pressures to improve management efficiency or cut costs to enhance profitability, often achieve these goals by downsizing their workforce. The coefficient for the debt-to-asset ratio (lev) is positive, likely because most firms in the healthcare sector have a relatively reasonable asset structure (the average debt-to-asset ratio is 0.546, which is within a safe range). These firms can use debt financing to expand their scale and improve operational performance, which in turn promotes employment. However, a higher debt-to-asset ratio also implies greater financial risk and higher interest costs, which may force firms to cut expenses and improve efficiency. As a result, when the debt-to-asset ratio exceeds a certain threshold, its impact on employment may shift from promoting to suppressing it. The coefficient for lev is only significant at the 10% level, which may precisely be due to this reason.

5.2. Robustness Checks

To further ensure the robustness of the empirical results, this study conducts the following robustness checks.

5.2.1. Substitution of Core Explanatory Variable

In addition to using firm-level Total Factor Productivity (TFP) as a measure of technological progress, scholars often employ alternative indicators such as R&D expenditures [57,58], the number of patent applications [59,60], or the number of industrial robots per thousand workers [61,62,63] to capture technological advancement. Generally, the amount of R&D expenditure reflects the degree to which firms prioritize technological innovation and research and development. Changes in R&D expenditure can, to some extent, reflect changes in technological progress, as firms are more likely to increase their R&D investment to sustain the benefits derived from improvements in their technological capabilities. This relationship is evident in the empirical studies of many scholars. Using patent data to measure technological progress operates under a similar logic. Given this high correlation, and constrained by the availability of data on industrial robots, this study substitutes R&D expenditure (in logarithmic form, lnrd) for the core explanatory variable to test the robustness of the baseline regression results.

The reason why R&D expenditure is not used in the baseline regression in this study is twofold. On the one hand, R&D expenditure includes the wages of research and development personnel, who are also part of a firm’s workforce, leading to a certain degree of correlation with employment. On the other hand, R&D expenditure does not always directly translate into technological progress. The conversion of R&D expenditure into technological outcomes often involves a time lag and may not accurately reflect changes in technological progress in the current period. Moreover, technological progress in firms does not always arise from internal R&D innovation; it can also result from external technology acquisition, improved management practices, and other means.

The specific regression results are presented in Column (3) of Table 3. After replacing the core explanatory variable with lnrd, the coefficient for the impact of technological progress in the health sector on employment increases to 0.218 and is statistically significant at the 1% level. This suggests that although there are some differences between R&D expenditure (lnrd) and TFP (lntfp), in measuring technological progress, both indicators show a consistent direction of impact on employment, with similar statistical significance. This strengthens the reliability of the core conclusion of this paper, namely, that technological progress in the health sector can significantly promote employment growth. Additionally, the coefficients and significance levels of other control variables remain largely unchanged, further supporting the robustness of the baseline regression results.

5.2.2. Sample Reduction

To ensure the robustness of the empirical results, we conducted a sample reduction test by excluding certain outliers and observations that could potentially distort the results. Specifically, we removed firms with extreme values in key variables, such as firms with zero or negative revenue, firms with fewer than 10 employees, or those that experienced irregular fluctuations in operational data, such as large variations in annual revenue or expenditures. The adjusted sample was then reanalyzed using the same model specifications as in the baseline regression. The results of this analysis are presented in Column (4) of Table 3. Notably, the coefficient for the technological progress variable (lntfp) remains positive and statistically significant at the 1% level, with only a marginal change in the magnitude of the coefficient. This suggests that the baseline results are not significantly affected by the presence of extreme values or data anomalies.

The robustness check confirms the reliability of the original findings and supports the conclusion that technological progress in the health sector plays a significant role in promoting employment growth. Furthermore, the other control variables also exhibit consistent signs and statistical significance, reaffirming the stability of the model across different sample specifications.

5.2.3. Endogeneity Treatment

In the baseline regression analysis, we used a two-way fixed effects model to control for unobserved factors that do not vary over time or across individuals, thus mitigating potential endogeneity issues arising from omitted variables. However, the core explanatory variable in this study—firm-level Total Factor Productivity (TFP)—is calculated using current-period labor, capital, and output data. This may lead to endogeneity concerns such as simultaneity or reverse causality, potentially causing estimation bias. To address these concerns, we employed the instrumental variable (IV) method, which is effective in handling such endogeneity issues. The IV approach involves introducing an instrument for the endogenous explanatory variable, allowing us to obtain consistent estimates for the core variable and improving the reliability of the estimation results. A commonly used strategy is to use the lagged value of the explanatory variable as an instrument [64]. Accordingly, in this study, we treated the lagged value of the TFP (l_lntfp) as an instrument and conducted a two-stage least squares (2SLS) estimation to mitigate the endogeneity issue and ensure the robustness of the results.

In addition to the conventional approach, we innovatively used a combination of firm stock-based compensation and R&D expenditures, normalized by the firm’s income level, as an alternative instrument. R&D expenditure is widely regarded as a key indicator of technological progress, and it is frequently used in empirical studies on technological advancement, thus satisfying the conditions for a valid instrument. Stock-based compensation, on the other hand, serves as an incentive for aligning the interests of employees, especially top management, R&D personnel, and other skilled workers, with the long-term goals of the firm. Stock-based compensation is often employed by firms to improve operational efficiency, reduce costs, and optimize cash flow, making it an effective tool for promoting organizational development. As such, the level of stock compensation can reflect the overall technological progress of a firm, aligning closely with the broad concept of technological progress as measured by TFP. Moreover, in order to facilitate comparisons across firms of different sizes, this study uses the output share of both stock-based compensation (stc) and R&D investment (rd) as an instrumental variable (IV) for the 2SLS regression, denoted as emp, where $emp=(stc+rd)/y$.

The regression results are presented in

Table 4. IV 2SLS regression results for the impact of technological progress on employment in the healthcare sector.

Variables	(1)	(2)	(3)	(4)
	First Stage	Second Stage	First Stage	Second Stage
lntfp		0.142 ***		0.057 ***
		(0.047)		(0.012)
IV	0.470 ***		−0.020 ***
	(0.053)		(0.002)
Controls	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Observations	1720	1720	1505	1505
Adj. R2	0.737		0.760
Kleibergen–Paap rk LM statistic	10.405 ***		4.118 **
Kleibergen–Paap rk Wald F statistic	77.892		163.68

Note: The values in parentheses represent cluster-robust standard errors; ***, ** indicate significance at the 1%, 5% levels, respectively.

.

The results from the 2SLS regressions with the instrumental variable (IV) are presented in Table 4. Columns (1) and (2) report the 2SLS regression results using l_lntfp as the instrumental variable. As shown in the table, the KP LM statistic and KP F statistic are 10.405 and 77.892, respectively, both of which reject the null hypothesis of instrument irrelevance and weak instruments at the 1% significance level. This indicates that l_lntfp is a valid instrument. Columns (3) and (4) of the table report the 2SLS regression results using emp as the instrument. Similarly, both the KP LM statistic and KP F statistic reject the null hypothesis of instrument irrelevance and weak instruments, confirming that emp is also a valid instrument.

The results from both 2SLS regressions indicate that, after addressing potential endogeneity issues, the coefficient of the core explanatory variable (lntfp) remains significantly positive, albeit with slight differences compared to the baseline regression model. This finding confirms that the primary conclusion of this study—that technological progress in the healthcare sector significantly promotes employment—still holds true.

5.3. Mechanism Analysis

The analysis above indicates that technological progress in the healthcare sector significantly increases employment within the industry. However, existing research has rarely explored this phenomenon from the perspective of its underlying mechanisms. To further clarify the specific pathways through which technological progress in the healthcare sector leads to employment growth, and to contribute empirical analysis of the mediating mechanisms to the existing literature on the relationship between technological progress and employment, this study will conduct a deeper discussion on the possible pathways of influence.

Drawing primarily on the mediation effect testing method proposed by Baron and Kenny [65], which is effective in revealing potential mediating mechanisms of the impact of technological progress on employment, this study will construct the following models. Through a stepwise regression analysis, we aim to examine the mediation effects:

$$ln{l}_{it}={\beta }_0+{\beta }_{1}lntf{p}_{it}+{\displaystyle \sum {\beta }_k}{X}_{ijt}+{\sigma }_i+{\lambda }_t+{\mu }_{it}$$

(8)

$$M_{it}={\beta }_0+{\beta }_{1}lntf{p}_{it}+{\displaystyle \sum {\beta }_k}{X}_{ijt}+{\sigma }_i+{\lambda }_t+{\mu }_{it}$$

(9)

$$ln{l}_{it}={\beta }_0+{\beta }_{1}lntf{p}_{it}+{\beta }_2{M}_{it}+{\displaystyle \sum {\beta }_k}{X}_{ijt}+{\sigma }_i+{\lambda }_t+{\mu }_{it}$$

(10)

Consistent with the previous baseline model, lnl_it, lntfp_it, X_ijt and μ_it represent the dependent variable, the core explanatory variable, the control variables, and the residual term respectively, while M_it is the mediating variable. This model also controls for individual fixed effects (σ_i) and time fixed effects (λ_t). Equation (8) represents the impact of technological progress on employment in the healthcare sector without considering the mediating effect, which corresponds to the baseline model in this paper. Equation (9) assesses the influence of technological progress on the mediating variable, while Equation (10) examines the effect of technological progress on employment with the mediating variable included. Additionally, to verify the reliability and magnitude of the mediation effect, this study conducts a Sobel test on the mediating variable. The mediation effect coefficient and its corresponding significance are also reported in the table.

5.3.1. Output Mechanism

In the previous literature review in Chapter 2 and the theoretical analysis in Chapter 3, the output effect of technological progress was discussed, which refers to how technological advancements lead to increased employment by enhancing a firm’s output. This output effect can even offset the disruptive effects brought about by new technologies (Carnoy, 2013) [66]. Next, we select firm revenue (lny) as a mediator variable to test the mediating role of the output mechanism in the relationship between technological progress and employment in the healthcare sector. Columns (1) to (3) in

Table 5. Examination of the mechanisms through which technological progress in the healthcare sector affects employment.

Variables	(1)	(2)	(3)	(4)	(5)
	lnl	lny	lnl	clr	lnl
lntfp	0.116 ***	1.022 ***	−0.930 ***	0.004 ***	0.103 ***
	(0.028)	(0.009)	(0.069)	(0.001)	(0.028)
clr					3.526 **
					(1.388)
lny			1.023 ***
			(0.061)
Controls	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes
Observations	1892	1892	1892	1892	1892
Adj. R2	0.756	0.938	0.897	0.162	0.762
Sobel statistic		1.578 ***		0.024 ***

Note: The values in parentheses represent cluster-robust standard errors; ***, ** indicate significance at the 1%, 5% levels, respectively.

report these results, with Column (1) providing the baseline regression result for comparison and analysis.

As shown in Column (2) of Table 5, the coefficient of lntfp is significantly positive, indicating that technological progress has a significant positive effect on output levels, which is consistent with the basic theory. After introducing the mediator variable, the output level (lny), the coefficient of the core explanatory variable lntfp in Column (3) decreases noticeably compared to the baseline regression result in Column (1) and even becomes negative (−0.930). On the other hand, the coefficient of lny is significantly positive (1.023). This suggests that the output level mediates the effect of technological progress on employment, implying that technological advancements promote employment indirectly by enhancing output levels.

Further analysis reveals that the direct effect of technological progress on employment in the healthcare sector may be negative (for every 1% increase in technological progress, employment decreases by 0.930%). However, the substantial indirect effect through the output mechanism counterbalances this substitution effect. Specifically, for every 1% increase in technological progress, the output level increases by approximately 1.022%, which in turn drives employment growth by about (1.022*1.023)%. After combining both effects, the net positive effect of technological progress on employment is 0.116%. And the results of the Sobel test indicate that the mediating effect of the output mechanism on employment is statistically significant and relatively strong.

5.3.2. Capital Deepening Mechanism

Capital deepening refers to the capital–labor ratio, where the degree of capital deepening within a firm reflects the extent to which its production model depends on capital. Firms with a higher capital–labor ratio rely more heavily on capital equipment for production, while firms with a lower ratio depend more on labor. Generally, technological progress drives capital deepening. The higher the degree of capital deepening (i.e., the higher the capital–labor ratio), the more likely capital will substitute labor, thereby increasing the capital share and decreasing the labor share. As a result, the firm’s capacity to absorb employment weakens [67,68]. However, some scholars argue that technological progress and capital deepening not only enhance productivity but also increase employment opportunities [69]. Previous studies often remain at the theoretical level, suggesting that technological progress, through driving capital deepening, either substitutes or complements labor. Yet, empirical analyses of this pathway are relatively scarce. Therefore, this study innovatively uses the ratio of a firm’s fixed assets (k) to its number of employees (l) as a measure of capital deepening (clr) and examines its mediating role in the effect of technological progress on employment in the healthcare sector.

The regression results are presented in Columns (1), (4), and (5) of Table 5. First, as shown in Column (4), the coefficient of lntfp is significantly positive, indicating that technological progress in the healthcare sector drives capital deepening, i.e., technological advancements encourage firms to increase the share of capital in their production inputs. This finding is consistent with the results of most studies and confirms the theoretical hypothesis that technological progress promotes capital deepening. Additionally, the coefficient of lntfp in Column (5) is lower than the coefficient in the baseline regression in Column (1), while the coefficient of clr is significantly positive, suggesting that capital deepening plays a mediating role. Specifically, technological progress indirectly promotes employment growth by increasing the capital–labor ratio in firms. This finding differs from general empirical conclusions, but from a theoretical perspective, the result is reasonable: an increase in capital deepening means a relative increase in the share of capital in the allocation of production factors, with a corresponding decrease in the share of labor. However, this does not necessarily lead to a reduction in the absolute value of labor input. A possible explanation is that technological progress in the healthcare sector, by promoting capital deepening, increases the demand for skilled labor, thereby compensating for the reduction in employment. The effect of this mechanism is relatively limited, as indicated by the coefficient of lntfp in Column (5) of Table 4, which is only 0.004. Furthermore, the results of the Sobel test show that the mediating effect of capital deepening on employment is statistically significant.

In summary, unlike the existing literature that has paid limited attention to mediating effects, the empirical results and analysis in this study indicate that the output mechanism and capital deepening mechanism are two possible pathways through which technological progress in the healthcare sector promotes employment. This finding not only validates the relevant theories but also provides a new perspective for understanding the pathways through which technological progress in the healthcare sector impacts employment.

5.4. Heterogeneity Analysis

The impact of technological progress in the healthcare sector on employment may vary depending on the nature of the industry and individual firm characteristics. Therefore, this study conducts a heterogeneity analysis from three dimensions: industry type, firm size, and the proportion of fixed assets within firms.

5.4.1. Heterogeneity Analysis Between Sub-Industries

The healthcare sector encompasses multiple sub-industries, which may exhibit significant differences in terms of technological characteristics and employment structures. For example, the medical manufacturing industry typically has a higher capital intensity and technological complexity, where technological progress may be more reflected in an increased demand for high-skilled labor. In contrast, the healthcare services industry relies more heavily on human capital, and technological advancements may indirectly influence employment by enhancing service efficiency. Additionally, in the biotechnology and pharmaceutical industry, technological progress may be concentrated in the research and development (R&D) phase, leading to a higher demand for R&D personnel, while the demand for other types of labor remains relatively low. These differences cause the effects of technological progress on employment to vary across sub-industries, highlighting the importance of examining the heterogeneity within the healthcare sector.

Therefore, this study innovatively classifies the sample of publicly listed companies in the healthcare sector based on the GICS classification into four groups: Medical Manufacturing, Health Care Services, Bio-pharmaceuticals, and Others. This classification allows for a more detailed analysis of the industry-specific differences in the impact of technological progress on employment, providing a fresh perspective on the relationship between technological progress and employment within the healthcare sector. Through subgroup testing, this study further enhances the innovation of the research framework and the richness of the findings, offering more targeted insights for policymakers. The specific classification is shown in

Table 6. Industry classifications of the healthcare sector and their GICS codes.

Name	Included Sub-Sectors and Their GICS Codes
Medical Manufacturing	Health Care Equipment (35101010), Health Care Supplies (35101020)
Health Care Services	Health Care Services (35102015), Health Care Facilities (35102020)
Bio-pharmaceuticals	Biotechnology (35201010), Pharmaceuticals (35202010)
Others	Health Care Distributors (35102010), Health Care Technology (35103010), Life Sciences Tools & Services (35203010)

Source: www.spglobal.com/spdji/en/ (accessed on 13 September 2024) “GICS Structure Effective in 2023”.

; it should be emphasized that the “Others” group mainly refers to auxiliary industries within the three major sectors of Medical Manufacturing, Health Care Services, and Bio-pharmaceuticals. This includes industries such as Health Care Distributors, which provide distribution services for medical products; Health Care Technology, which supports the information operations for Health Care Services; and Life Sciences Tools & Services, which provides R&D support for the biotechnology and pharmaceutical sector.

The regression results are presented in

Table 7. Regression results of the impact of technological progress on employment in the healthcare sector (by industry group).

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Medical Manufacturing	Health Care Services	Bio-Pharmaceuticals	Others	Health Care Equipment & Services (3510)	Pharmaceuticals, Biotechnology & Life Sciences (3520)
lntfp	0.165 **	0.093 ***	0.099 **	0.219 ***	0.127 ***	0.098 **
	(0.073)	(0.026)	(0.043)	(0.0644)	(0.039)	(0.041)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	539	352	682	330	1001	891
Adj. R2	0.739	0.761	0.773	0.791	0.734	0.779

Note: The values in parentheses represent cluster-robust standard errors; ***, ** indicate significance at the 1%, 5% levels, respectively.

. As shown in Table 7, among the four sub-industries, the industry with the highest coefficient of lntfp is “Others” (0.219), indicating that technological progress in this sub-industry has the most significant positive effect on employment. Specifically, a 1% increase in the level of technological progress leads to an approximately 0.219% increase in employment. The second-highest is Medical Manufacturing, with an lntfp coefficient of 0.165, meaning that a 1% increase in technological progress in this sector leads to about a 0.165% increase in employment. In contrast, the impact of technological progress on employment in the Health Care Services and Bio-pharmaceuticals is relatively smaller, with both having lntfp coefficients close to 0.1%.

This difference may stem from the varying labor demand characteristics across the sub-industries. “Others” primarily includes businesses involved in distribution and operations, which tend to have a higher demand for labor and lower requirements for labor quality. As a result, technological progress in this sub-industry is more easily translated into job growth. The Medical Manufacturing industry, being both capital- and labor-intensive, has a high demand for technical and skilled personnel, such as skilled operators for medical equipment production lines, which has also contributed to higher employment growth. In contrast, Health Care Services and Bio-pharmaceuticals place higher demands on labor quality. Here, technological progress is more likely to manifest as an increased demand for high-skilled labor, rather than a significant expansion of employment.

To further test whether this difference is robust, this study conducts pairwise comparisons of the differences between the four industries, with the results presented in

Table 8. Empirical p-values for inter-industry group differences.

Inter-Group Difference β0 − β1		β1
		Medical Manufacturing	Health Care Services	Bio-Pharmaceuticals	Other Industries
β0	Medical Manufacturing		0.072 (0.197)	0.066 (0.238)	−0.054 (0.312)
	Health Care Services	−0.072 (0.197)		−0.006 (0.476)	−0.126 (0.097)
	Bio-pharmaceuticals	−0.066 (0.238)	0.006 (0.476)		−0.12 (0.102)
	Others	0.054 (0.312)	0.126 (0.097)	0.12 (0.102)

Note: The values in parentheses represent the p-values for the inter-group difference test, calculated using Fisher’s permutation test (with 1000 samples).

.

Table 8 presents the results of the inter-group difference test. As shown in the table, with the exception of the “Health Care Services” and “Others”, which exhibit a statistically significant difference, the impact of technological progress on employment does not show significant statistical differences between the other sub-sectors within the healthcare sector. Specifically, the empirical p-values for the Health Care Services and Bio-pharmaceuticals are the largest, indicating that their statistical differences are the least pronounced. However, the non-significant difference in coefficients does not imply that inter-industry differences are entirely absent. The observed convergence in the impact of technological progress on employment across industries may stem from several reasons: first, spillover effects of technology and the free movement of labor across industries may have diminished the differences between sectors; second, the data sample used in this study is relatively limited, and especially after grouping, the sample size within each subgroup is further reduced. Even if there are differences, they may not be statistically significant due to the limited sample size.

From another perspective, the non-significant inter-industry differences also suggest, to some extent, that the positive impact of technological progress on employment within the healthcare sector is highly stable. This finding further supports the main conclusion of this study—that technological progress in the healthcare sector has a robust positive effect on employment.

In addition, this study also performs heterogeneity analysis by directly classifying the healthcare sector into two major sub-sectors according to the GICS classification: Health Care Equipment & Services (3510) and Pharmaceuticals, Biotechnology & Life Sciences (3520). The regression results are presented in Columns (5) and (6) of Table 7. The empirical p-value for Fisher’s inter-group difference test is 0.336, indicating that the difference in the impact of technological progress on employment between the two major sub-sectors is not statistically significant, for the reader’s reference.

5.4.2. Heterogeneity Analysis Based on Firm Size

Considering that firm size may influence the impact of technological progress on employment, this study further classifies the listed firms in the healthcare sector based on firm size to examine whether there is heterogeneity in the effect of technological progress on employment across firms of different sizes. Specifically, using the average total assets in 2018 as the threshold, firms with total assets greater than or equal to the average in 2018 are classified as large firms (l_size), while those with total assets below the average are classified as small firms (s_size). The regression results are presented in Columns (1) and (2) of

Table 9. Regression results of the impact of technological progress on employment (by firm size and fixed asset proportion).

Variables	(1)	(2)	(3)	(4)
	l_size	s_size	h_far	l_far
lntfp	0.247 ***	0.090 ***	0.233 ***	0.095 ***
	(0.025)	(0.080)	(0.081)	(0.028)
Controls	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes
Observations	935	957	968	924
Adj. R2	0.816	0.729	0.795	0.744
Inter-group difference	0.157 ***		0.138 ***

Note: The values in parentheses represent cluster-robust standard errors; *** indicate significance at the 1% level. The inter-group coefficient differences are calculated using Fisher’s permutation test (with 1000 samples).

.

The results show that, for both large and small firms, technological progress has a significant positive effect on employment. However, the coefficient of lntfp for large firms is significantly higher than that for small firms, indicating that technological progress has a stronger impact on employment in large firms. Moreover, the empirical p-value for Fisher’s inter-group difference test shows that the difference in the promotion effect between the two groups is statistically significant. This finding may be related to large firms’ stronger resource integration capabilities, higher technological absorption efficiency, and broader market influence, which enable their technological progress to be more effectively translated into employment growth.

5.4.3. Heterogeneity Analysis Based on the Proportion of Fixed Assets

The proportion of fixed assets reflects the asset structure of a firm, and there may be heterogeneity between firms with different proportions of fixed assets. To examine this heterogeneity, this study classifies the listed firms in the healthcare sector based on their fixed asset ratio (the ratio of fixed assets k to total assets as) into capital-intensive firms (h_far) and asset-light firms (l_far). The grouping method follows the same approach as before, using the average fixed asset ratio in 2018 as the threshold. Firms with a fixed asset ratio above the average are classified as capital-intensive firms, while those with a ratio below the average are classified as asset-light firms. The regression results are presented in Columns (3) and (4) of Table 9.

The results show that the impact of technological progress on employment in capital-intensive firms is significantly higher than in asset-light firms within the healthcare sector. The empirical p-value for Fisher’s inter-group difference test further indicates that this difference is statistically significant. This finding may be related to the capital-intensive nature of these firms: asset-heavy firms typically require more labor to operate and maintain fixed assets (such as medical equipment production lines, etc.), making it easier for technological progress to be directly translated into employment growth. In contrast, technological progress in asset-light firms is more likely to manifest as efficiency improvements or process optimizations, which have a relatively weaker effect on employment.

In summary, this section systematically examines the heterogeneity of the impact of technological progress on employment by classifying sample firms in the healthcare sector based on different industry types and firm characteristics. This study finds that technological progress has a significant positive effect on employment across all industries within the healthcare sector; however, the differences in this effect between industries are not statistically significant. Furthermore, the employment-promoting effect of technological progress is more pronounced in large firms and capital-intensive firms. This may be attributed to large firms’ stronger resource integration capabilities and the higher demand for labor in capital-intensive firms. The above analysis provides a clear understanding of the varying impacts of technological progress on employment within the healthcare sector. It not only offers new insights into the relationship between technological progress and employment but also provides more targeted reference points for policymakers.

6. Conclusions and Recommendations

6.1. Research Conclusions

In the context of continuous technological advancements and widespread applications, the healthcare sector has entered a new phase of development. Technological progress in the healthcare sector not only drives industry development but also has a significant impact on employment. To address this issue, this paper empirically analyzes the impact of technological progress on employment in the healthcare sector using panel data from publicly listed U.S. healthcare companies between 2013 and 2023. Using Total Factor Productivity (TFP) as a measure of technological progress and employing a two-way fixed effects model, this study fills the gap in micro-level research on the relationship between technological progress and employment, enriching the empirical research on the healthcare sector. The main conclusions of this study are as follows:

Firstly, the empirical baseline results show that technological progress in the U.S. healthcare sector has a significant positive effect on employment. A 1% increase in technological progress leads to a 0.116% increase in employment, with the main conclusion remaining robust after conducting tests for robustness and addressing endogeneity.

Secondly, to further explore the impact mechanisms of this promotion effect, this study integrates existing theories and research to verify both the output mechanism and the capital deepening mechanism. The results indicate that the technological progress in the healthcare sector primarily promotes employment by increasing output. The capital deepening brought by technological progress does not result in a substitution effect for employment, and although its indirect effect is small, it still demonstrates a positive role in promoting employment.

Thirdly, considering the potential heterogeneity among different sub-industries, enterprise sizes, and asset structures within the healthcare sector, this study further conducts a heterogeneity analysis. The results indicate that technological progress has a relatively larger employment-promoting effect in the medical manufacturing and auxiliary industries represented by operations and distribution, whereas its effect on the employment of medical services and bio-pharmaceutical industries is relatively smaller. However, this difference is only significant in certain industries. Moreover, large enterprises with scale advantages and those with a higher proportion of fixed assets show a more significant employment-promoting effect of technological progress.

6.2. Policy Recommendations

To fully harness the employment-enhancing effects of technological progress in the healthcare sector, foster the sustainable development of both the healthcare sector and society, and ensure stable employment growth, this paper offers the following recommendations based on the conclusions above.

Strengthen support for research and development (R&D) and technology adoption in the healthcare sector to drive technological progress. For instance, a dedicated technology innovation fund and subsidies for the healthcare sector could be established, providing tax incentives to enterprises engaged in technological research and the adoption of new technologies. And, healthcare sector parks and innovation cooperation platforms should be created to facilitate collaboration and exchange between businesses, universities, research institutions, and other relevant organizations within the industry. Furthermore, healthcare enterprises should enhance independent innovation and technology adoption, increase R&D investments, and form specialized research teams to drive technological advancements.

Promote the integrated development and market expansion of the healthcare sector, thereby enhancing revenue generation and encouraging increased capital investment. Technological progress not only promotes employment by increasing output and revenue but also by deepening capital. Therefore, on one hand, policies should encourage the integration of the healthcare sector with sectors such as the Internet, elderly care, and tourism. On the other hand, policies should guide financial institutions to develop diversified financial products tailored to the characteristics of the healthcare sector, such as intellectual property pledge loans and special loans for the transformation of technological achievements. This would address the financial challenges faced by health enterprises in their investment processes, promote capital deepening, and ultimately drive employment growth.

Provide targeted support based on specific needs and characteristics. For high-tech medical manufacturing industries, more training programs can be offered, while for the healthcare services industries, investments in information technology and smart services can be increased. Meanwhile, it is also necessary to support the development of small and light-asset enterprises: while large enterprises and heavy-asset companies have a more significant impact on employment from technological progress, small and light-asset businesses may require more support. Policies can help small and medium-sized enterprises (SMEs) by reducing financing barriers, encouraging technological adoption, and fostering innovation, which would enable them to leverage technological advances and create more job opportunities.

By implementing these policies, the government can ensure that technological advancements in the healthcare sector drive sustainable development, while creating broad employment opportunities at all levels, ultimately achieving social stability and sustainable development.

6.3. Limitations and Perspective

While this study provides valuable insights into the relationship between technological progress and employment in the healthcare sector, it is not without limitations.

First, due to the use of a single-factor production function, the results of this study only reflect the preliminary relationship between a single factor and output. Future research could further explore the interrelationships among capital, labor, and technology by adopting a multi-factor production function, which would more accurately capture the employment effects of technological progress in the healthcare sector. Second, this research focuses on data from publicly listed healthcare companies in the United States from 2013 to 2023, which may not fully represent the entire global healthcare sector, and the relatively short sample period also imposes certain limitations on the findings. Future studies could expand the dataset to include companies from other countries and regions; conduct cross industry, national, and regional comparisons to improve the generalizability; and track these changes over extended periods to provide deeper insights into the sustainability of these trends. Lastly, this study primarily focuses on the impact of technological progress on employment, but, due to limitations in data availability, it does not address other potentially important outcomes, such as changes in wage distribution or job quality. Future studies could explore the broader labor market impacts of technological progress in the healthcare sector, including how these changes affect different skill levels, job types, and income inequality.