The Theory of Human Capital explains the relationship between investment and performance. It began with the works of Mincer, Friedman, Shultz and Rosen at the University of Chicago ( G. Becker, 1994), in response to observed income inequality among workers. Labour income distribution in the model proposed by Mincer ( Mincer, 1970) took length of schooling into account, because after an initial investment it would provide a time-delayed return on labour income by increasing worker productivity ( Portela, 2007).

In the concept of human capital as the result of the accumulated knowledge, skills and attitudes in a human being, who is the sole possessor of these attributes ( Huffman, 2009), several investors (including family and country) must intervene in the formation of this capital. Human capital thus embodies the knowledge, talent and experience of individual employees’ ( Bontis & Fitz‐enz, 2002), describes the aggregate of individuals within the firm ( Wright et al., 2014) and builds a unit-level resource from the emergence of individuals’ knowledge, skills, abilities and other characteristics ( Ployhart & Moliterno, 2011).

Valuing people as an asset is at the heart of the concept of human capital. It argues that each person has a different combination of current and potential human capital, which through management can be transformed into wealth ( Boon et al., 2017; Mayo, 2001) and generate a return like any other investment ( G. Becker, 1994; Schultz, 1993, 1989). Behaviour thus has significant value because unlike physical resources, it improves with use ( Pike et al., 2005) and allows material resources and financial capital to be transformed into added value: something that never occurs before human behaviour is applied to them ( Fitz-enz, 2009). Human resource management (HRM) involves systems deployed to recruit, select, remunerate and engage employees ( Wright et al., 2014), along with other initiatives to drive attitudes and desired behaviours ( Boon et al., 2017). It enhances the value of human capital, making it both inimitable and irreplaceable in generating competitive advantages for the company ( Barney et al., 2011; Shaw et al., 2013). Accordingly, scholars tend to consider all staff expenses as investment in human capital ( Ofurum & Aliyu, 2018; Onyekwelu et al., 2017; Rompho, 2017). They provide empirical evidence that HRM expenditures increase productivity ( Bassey & Tapang, 2012; Jackson et al., 2014) and improve financial results ( Huselid et al., 1997; Jackson et al., 2014).

However, the acquisition of human capital does not imply unconditional ownership of the asset, as employees are free to leave the organization and cancel out the competitive advantage that their human capital has provided ( Huffman, 2009). So, the question inevitably arises: should companies invest in human capital?

Human Capital Theory has been blamed for reducing corporate investment in human capital on the grounds that the employee, as owner and beneficiary of that capital, has the obligation to develop it ( Fleming, 2017). This lower investment is supported by unconventional forms of work that make workers similar to entrepreneurs: they pay for all their instruments and tools and accrue no rights to vacation or sick pay, leave, work-hour restrictions, or participation in retirement plans. Contracts such as these and the huge number of candidates who now expect to get work hours ‘on-demand’ have contributed to lower wages for conventional contracts. In addition, the excessive focus on results, regardless of the labour effort required, puts employees in positions of having to work extra hours without remuneration ( Fleming, 2017). Not surprisingly, labour markets in the neoliberal or mixed regimes of southern Europe feature high unemployment and low wages ( Schotter et al., 2021).

While this restructuring of capitalist production relations seems economically rational, it can actually contribute to lower organizational performance, which would make it economically irrational ( Fleming, 2017). It also favours the emergence of extreme right-wing movements rooted in the unhappiness caused by the deterioration of wages, benefits, and working conditions ( Cumming et al., 2020). This could provide fuel for recurrent political crises and decrease corporate productivity ( Cumming et al., 2020).

One of the most prominent theories of management is the Resource Based View (RBV). This theory argues that for an organization to outpace its competitors, it must possess unique and inimitable resources that give it an advantage in its specific market. Human capital (HC) is one such resource ( Jiang et al., 2013).

In the literature, there is broad agreement that investment in training has a positive impact on organizational performance. However, financial performance – an outcome influenced by many variables ( Aragon & Valle, 2013) and subject to appropriation of profits by human capital holders or their managers ( Crook et al., 2011) – is not always consistent with this assumption.

Increasing human capital influences a company’s financial performance ( Aragon & Valle, 2013; Crook et al., 2011; Huselid, 1995; Newbert, 2008), while divestment of human resource development (HRD) undermines organizational sustainability ( McCarthy & Sheehan, 2014).

In the relationship between human capital and performance, RBV separates the constructs of human capital and behaviour ( Ployhart, 2021), based on the idea that human capital cannot contribute to performance if it is not used. In fact, human capital that does align with resources usable by the company and transformable into economic performance is not human capital. Ployhart (2021) adds to this line of argument by incorporating behaviour – a combination of psychology, economics, strategic management and human resource management – to the relationship between human capital and performance. Other works have also helped unite multi-level, multi-disciplinary research around RBV theory constructs and terminology ( Call & Ployhart, 2021; Delery & Roumpi, 2017; Ployhart & Moliterno, 2011).

With recent developments, this RBV-based construct has become known as Human Capital Resource (HCR), defined as ‘a unit-level resource that is created from the emergence of individuals’ knowledge, skills, abilities, and other characteristics (KSAOs)’ ( Ployhart & Moliterno, 2011, p. 128). The definition was later changed to ‘individual or unit-level capacities based on individual KSAOs that are accessible for unit-relevant purposes’ ( Ployhart et al., 2014, p. 374). Although the second definition no longer includes the term, the construct is based on the emergence of a resource from KSAOs. This goes beyond a simple aggregation of individual KSAOs that are then amplified and transformed through intra-organizational dynamics. Instead, it provides a competitive advantage because it is valuable, rare and inimitable by the competition ( Ployhart & Moliterno, 2011). Emergence is thus understood to result from the interaction of various behaviours ( Moliterno & Nyberg, 2019). While HCR, as currently defined, impacts organizational performance and goes beyond individual KSAOs, the lack of clarity about aspects that emerge from them has led scholars to apply various measures of KSAOs, work activities and attitudes at the individual level, in an effort to capture the phenomenon. However, a recent meta-analysis led Zhang et al., (2023) to conclude that standardization is still a long way off. In that work, the authors considered personnel expenses to be measures of financial resources and not HCR, although they corresponded to amounts spent on hiring and training employees, HRM systems and workforce salaries. Assuming that the company will only pay a salary corresponding to human capital – the salary that compensates the contribution of relevant KSAOs to organizational performance (and not other KSAOs the employee may have) – and implements the HRM systems that it considers will enhance an individual’s KSAOs, personnel expenses seem to us an adequate proxy to capture HCI.

The relationship between IHC and OP can be influenced by the context in which employees behave. The context provides several variables that can determine or moderate organizational behaviour ( Clitheroe et al., 1998; Gomes et al., 2015), influence emotions and attitudes ( Valdivia & Mart, 2020), and impact research outcomes ( Johns, 2006).

This article explores the life cycle stage of the company as a context variable that can be calculated from the SCIE database.

Life Cycle Theory (LCT) starts with the premise that organizations are similar to living beings that pass through different stages of development from birth to death ( Liu & Chou, 2016). Generally, five stages are identified: Introduction, Growth, Mature, Decline and Shake-out ( Dickinson, 2011; Habib & Hasan, 2017; Jackson & Schuler, 1995; Yan & Zhao, 2010). Lately, they have been assessed on the basis of a descriptive composition of four items (annual dividends scaled by income, percentage of sales growth, capital expenditure as a proportion of a firm’s value, and the age of the firm) ( Liu & Chou, 2016), or the net cash-flow pattern, which was used in this study ( Dickinson, 2011; Habib & Hasan, 2017).

In the organizational environment in which the behaviours of employees and managers occur, profitability tends to be higher in Stage 2, 3 and 5 (especially Stage 3), and lower in Stages 1 and 4. Investment is high in Stages 1 and 2, while in Stage 5 companies sell assets to pay debts. Debt is also higher in Stages 1 and 2, following investment, and decreases in Stage 3, when companies distribute more shareholder dividends or invest in unprofitable projects to decrease profits and lower taxes ( Dickinson, 2011).

Stage 3 has the lowest risk of insolvency (Domingo Terreno et al., 2017) and CEO remuneration has higher fixed values in Stages 3 and 5, while pay-for-performance predominantly occurs in Stages 1 and 2 ( Liu & Chou, 2016).

The database included official balance sheet data from all companies operating in the private sector inPortugal, comprising a vast set of annual information for statistical, fiscal and accounting purposes.

The companies selected had 100 or more employees, based on the conventional cut-off point betweenlarge and small firms in labour economics research ( Centeno & Novo, 2012; Winter-Ebmer & Cardoso, 2010) and the assumption that the phenomenon would be identifiable at that level. Onlycompanies that remained active throughout the study period (2010–2016) were included. From thisselective process we obtained a sample of 1566 companies, distributed according to their relevantStage, Size and Sector variables, as summarized in Table 1:

Table 1
Variables

Source: Author’s own computations using SCIE data.Note: The size variable was obtained by dividing the sample into 5 quintiles.

Linear regression was applied to the sample, to model the dependent variable of organizationalperformance (OP) as a function of the independent (IHC, Stage) and control (Sector, Size) variables.

The procedure was repeated for each of the indicators chosen to measure OP (NS, GVA, EBITDA,ROA, ROE and ROS).

Two estimation models were used for the regression: Ordinary Least Squares (OLS) and Fixed Effects(FE).

Although OLS rendered more expressive results, only FE results are presented because this morerigorous estimation model captured the specific effect of the variables of interest.

The results presented here were obtained from the following equation:

where i is the number of observations of the explanatory variable at time t, which corresponds to thereference year; OP_it corresponds to the response variable of observation i at time t, tested here for NS, GVA, EBITDA, ROA, ROE and ROS; μ _i is the firm’s unobserved heterogeneity; β0, β1, β2, β3 are the parameters of the model estimated to describe the effect of the explanatory variable on OP when allother explanatory variables remain constant; and ε _it represents errors assumed not to be correlated withthe explanatory variables included in the model, and not serially correlated.

The model was tested to capture the effect of variations in company behaviour over time. Accordingly, Δ _lnIHC _it were added to the model and corresponded to the difference between IHC for year t and IHC for year t-1. The entire model is expressed in the following equation:

Since investment in human capital (IHC) on company balance sheets relates to the course of a yearbut might not produce immediate effects, the possibility that the operational performance of year t isdue to the investment in years t and t-1 was also tested using the following equation:

The inverse regression was also calculated: IHC was modelled as a function of organizationalperformance (OP), using equations 4 and 5 for years t and t-1, respectively:

The data show very disperse ROE, IHC and Size variables resulting from the considerable size differences of the companies that comprise the database ( Table 2).

Table 2
Means and Standard Deviations

Source: Author’s own computations using SCIE data.

The results presented in Table 3 show a positive and significant relationship between companyperformance and IHC when measured with the economic variables NS, GVA and EBITDA, but notwith the financial variables ROA, ROE and ROS.

Table 3
Regression Results

Source: Author’s own computations using SCIE data.Note: The headers show equations 1, 2 and 3, modelled as a function of OP. The results show the effects on NS, GVA, EBITDA, ROA, ROE and ROS. Significance levels*** 1 %** 5 %* 10 %

The overall adjustment of the model tested was good for NS and GVA variables, with R ²>0,3 but weaker for EBITDA, with R ²=0,17, for equations 1 (OPt = f(IHCt)) and 2 (OPt = f(IHCt + ΔIHCt-1)). The model fit was lower for Equation 3 (OPt = f(IHCt-1)), probably indicating that variables notconsidered in the study gained strength as time passed.

Since the model contains categorical variables for Size, Sector, Stage, and a large number ofobservations, R ² is comparable to findings in the existing literature.

Results for the effect of IHC on OP indicated that a 1% increment in IHC in one year produced anincrease of 0,49% on NS, 0,63% on GVA and 0,37% on EBITDA the same year. If the rise isconsistent over two years NS could rise by 0,51%, GVA by 0,65% and EBITDA would remain at 0,37%.

With Equation 3 (OPt = f(IHCt-1)) the effect faded (0,15% for NS, 0,13% for GVA and 0,12% for EBITDA). This means that investing in HC was effective for leveraging performance immediately and not later, as we had thought.

Table 4 shows the results of the inverse regression calculations, where IHC was modelled as afunction of OP (see equations 4 and 5). This was done to explore the possibility that the results obtained in the earlier regressions might simply be due to the abundance of financial resources.

The findings verified that organizational performance modelling based on IHC produces more expressive results than its opposite. This evidence suggests that a causal relationship between IHC and OP is more likely than between OP and IHC, as was the case for GVA and EBITDA but not NS, which had the same effect in both models. Mean while GVA and EBITDA showed a great difference, suggesting that these two OP measures better capture what we seek to understand: the effect of investment on HC.

Table 4
Reverse Regression Results

Source: Author’s own computations using SCIE dataNote: The header shows equations 1, 2 and 3, modelled as a function of IHC or OP. The results show the effect on NS, GVA and EBITDA. Significance levels *** 1 %** 5 %* 10 %.

Regressions with the same variables for each Stage ( Table 5) demonstrated a very poor overall fit for EBITDA (ranging between 0.05 and 0.14 for R ² in Stages 1 and 2). The results were not significant for Stages 3 and 4.

Table 5
Results by Life Cycle State

Source: Author’s own computations using SCIE data.Note: The headers show equations 1 and 2 modelled as a function of OP. The results show the effects on GVA and EBITDA in each Stage. Significance levels: *** 1 %** 5 %* 10 %

In summary, the variables NS, GVA, EBITDA, ROA, ROE and ROS were tested to measure OP and verify that:

1. 1. ROA, ROE and ROS produced non-significant results in our model.
2. 2. NS produced the same effect when modelled as a function of IHC and when used as the response variable, preventing us from reaching any conclusions about causality between OP and IHC.
3. 3. EBITDA results were significant in Stage 1, marginally significant in Stage 2 and not significant in Stage 3 or 4. The explanatory capacity of the model was low and the error high in Stage 4.
4. 4. GVA was the only response variable that remained consistent in all tests: the results differed when modelled as the response variable vs the explanatory variable; the results were positive and significant, behaviour varied in Stage modelling samples and overall fit was acceptable.

This study included only companies with more than 100 employees operating in Portugal, so the conclusions cannot be directly applied to smaller companies or other countries where culture, labour and tax laws or other constraints may generate different employer or employee behaviour.

The period considered only includes years following the 2008 economic crisis, so the results cannot be directly generalized to previous moments without studying the behaviours revealed by the data.

However, research suggests the performance advantages of investing in human capital during periods of economic turbulence ( Maley, 2019).

Pathways in the field of attitudinal variables should be explored, since our results point to a greater effect on GVA when company behaviour is consistent over two years in any life cycle stage, and immediate effects in Stage 4 – Decline.

^aligia.portovedo@gmail.com^balveloso@psi.uminho.pt^cmiguel.portela@eeg.uminho.pt