How does the national human capital index influence the total factor productivity of the Romanian R&D firms?

Abstract

BACKGROUND:

Productivity is essential for economic development. However, the computation of the total factor productivity (TFP) for research and development (R&D) firms is largely ignored by the previous literature, in particular in the Eastern European countries.

OBJECTIVE:

The purpose of this paper is to analyze to what extent the national human capital index (NHCI) influences TFP for a set of 116 Romanian firms, acting in the R&D industry.

METHODS:

We compute the TFP level using different approaches, by analyzing firm-level data from the period of 2007 to 2016. At the same time, we resort to a common factor analysis (CFA) to derive the NHCI for Romania. Afterwards we apply a panel data investigation to see the impact of NHCI on TFP and we use several control variables as firm-level financial performances and firm’s size.

RESULTS:

Our estimation reveals that NHCI, a component of the national intellectual capital index, has a significant but marginal impact on TFP, whereas the financial performances play a more important role in enhancing firms’ productivity.

CONCLUSIONS:

The national human capital influences the firm productivity in the long run. At the same time, larger companies are financially more stable compared with their smaller counterparts and record a higher TFP.

Keywords

Human capital index total factor productivity Romanian R&D firms factor analysis panel analysis

Claudiu Albulescu is currently full Professor at the Management Department, Faculty of Management in Production and Transportations, Politehnica University of Timisoara. He is associated researcher at CRIEF, University of Poitiers, and associated professor at the Doctoral School of Economics and Business Administration within the West University of Timisoara. His research interests are financial macroeconomics, energy economics, banking and finance, corporate finance, entrepreneurship and innovation.

Serban MICLEA is Lecturer at the Management Department, Faculty of Management in Production and Transportations within the Politehnica University of Timisoara. His research interests are related to management and marketing, firm performances, and corporate finance.

1 Introduction

Productivity is essential for economic development [1]. However, the productivity of research and development (R&D) companies is almost neglected by the literature. Moreover, the computation of firms’ total factor productivity (TFP) in the Eastern Europe is practically unaccounted. Romania’s R&D sector has undergone profound changes during the last decades. However, the sector is still dominated by national research institutes. The productivity performance of the companies is questionable and largely influenced by their corporate governance [2]. However, some external factors might be also at work, brought for instance by the level of the national human capital. 1 Against this background, we analyze to what extent the national human capital index (NHCI) influences the TFP for a set of 116 Romanian firms, for the period 2007 to 2016.

The determinants of TFP are largely debated in the empirical literature (see, for example [4]). They usually refer to R&D investment, organizational capital, skills and technologies, but also to the human capital [5 –9], all these elements defining the firms’ managerial performance. Most of previous works analyze the impact of R&D investment and innovation capacity on firms’ productivity. Although most of these previous works report positive effects, [10] state that the impact of R&D expenditure on productivity is nonlinear, whereas [11] report only a small impact of R&D expenditure on firms’ efficiency. Further, [12] indicate a positive impact of R&D activities, but only in the presence of fiscal facilities. A distinctive set of studies put accent on the role of corporate governance in influencing the firms’ productivity (for a recent survey, please refer to [13]). Likewise, [14] and [15] show that board diversity enhances firms’ productivity, whereas [16] admit that board independence is important to increase the productivity level. On contrary, relying on the organization theory advanced by [17], other studies sustain that a higher independence ameliorates the decision process and therefore the firms’ efficiency. The ownership structure is equally investigated [18 –21]. Most of these works show that private companies perform better, while [22] report inconclusive findings.

Another category of elements is related to firms’ financial performances and refers to the role of the profitability, liquidity, or firms’ financial structure. In this line, several works (e.g. [23, 24]) show that leveraged companies innovate less and have smaller productivity performances. With a focus on firms’ size and profitability, [25] show the importance of both elements for the firms’ productivity.

We add to this last category of works, but we build our analyzing around the role of human capital in enhancing the TFP. The importance of intellectual capital (a broader measure of the quality of human resources) for firms’ efficiency was intensively investigated in the literature (e.g. [26, 27]). However, as far as we know, none of the previous papers investigates the impact of the national human capital index (NHCI) on the TFP, using firm-level data from the scientific R&D industry.

Therefore, alongside the previous works we contribute to the exiting empirical literature along several dimensions. First, in line with [2] we compute the TFP for 116 Romanian R&D firms, relying on [28] and [29] approaches. The focus on Romanian R&D sector is particularly appealing given that Romania faced major challenges in terms of restructuring R&D and innovation activities during the last decades. Almost 50% of the companies unregistered under the scientific R&D industry are state-owned research institutes. Second, we compute the NHCI for Romania using a common factor analysis (CFA) approach. Like [30], we include in the analysis individual indicators related to the education system, labor markets or access to internet. Third, we run a multiple Ordinary Least Square (OLS) regression to test the impact of the NHCI on TFP, while we control for the role of firm profitability and firm size (our control variables). We discover that the positive and significant impact of NHCI on the firm productivity is reduced. However, both the profitability and firms’ size positively influence their productivity level. The rest of the paper presents the data and methodology, the results and the conclusions.

2 Data and methodology

Annual data are extracted from AMADEUS (Bureau van Dijk –BvB) database and covers the time span 2006 to 2016. Complete data are available for 116 firms under the NACE code “72 - Scientific research and development”. 2 Data for the NHCI computation are extracted from the World Bank Development Indicators. 3

2.1 The TFP computation

To calculate the TFP, we start from the classic Cobb-Douglas production function whose inputs are the capital and labor: $Y_{it} = A_{it} K_{it}^{α} L_{it}^{β},$ (1) where Y_it is the output (or the added value in our case - AV) of a firm i at time t, K_itis the capital (stock) and L_itrepresents the labor (number of employees).

The stock of capital (K_it) for each firm i at each period t is calculated using the Perpetual Inventory Method (PIM) as follows: $K_{it} = K_{it}^{BV} - K_{it - 1}^{BV} + DEP R^{BV},$ (2) where $K_{t}^{BV}$ and $K_{t - 1}^{BV}$ represent the current and, respectively, the lagged book value of tangible fixed assets, while DEPR^BVis the book value of depreciation and amortization.

Applying this method to compute the capital stock implies that we lose data for one year. Consequently, the final data sample covers the period 2007 to 2016. Based on Equation (1), TFP is defined as the output unexplained by the inputs: $\ln (A_{it}) = ln (Y_{it}) - α ln (K_{it}) - β ln (L_{it}),$ (3) where ln(A_it)is equivalent to the ln(TFP_it).

In order to avoid losing too many observations, the added value (AV) is internally computed as follows: $A V_{it} = L_COST S_{it} + EBITD A_{it},$ (4) where $L_COST S_{it}$ is the labor cost related to the employees and EBITDA_it(Earnings Before Interest Taxes Depreciation and Amortization) is a measure of profits, or a part of income oriented to the capital investment.

Finally, for the TFP computation, we resort to [28] and [29] approaches. Building upon [28, 29] proposes a one-step General Method of Moments (GMM) procedure with consistent standard errors. Consequently, different instruments for different equations are specified while ω_it = f [h (m_it-1 + K_it-1)] + a_it. The TFP is computed based on:

$\begin{matrix} Y_{it} & = & c + α K_{it} + β L_{it} + f [h (m_{it - 1} + K_{it - 1})] \\ + a_{it} + δ_{it} . \end{matrix}$ (5)

In Table 1 we present the panel unit root tests for the two measures of TFP, relying on [28] (TFP – LP) and [29] (TFP – W), respectively. We resort to [31] and [32] tests from the first generation. The results indicate that the productivity series are stationary, and the OLS regression can be used.

Table 1

Panel unit root tests for TFP series

	Inverse chi-squared (P)	Inverse normal (Z)	Inverse logit t (L*)	Modified inverse chi-squared (Pm)
TFP-LP	743.4***	– 8.946***	– 14.53***	24.13***
TFP-W	738.5***	– 10.32***	– 15.22***	23.90***

Notes: (i) ***p < 0.01, **p < 0.05, *p < 0.1; (ii) TFP-LP is the Levinsohn and Petrin’s (2003) measure of TFP whereas TFP-W is the Wooldridge (2009) measure of TFP.

2.2 The NHCI computation

The variables used in the NHCI computation are indicated in [30] and presented in Table 2. Nevertheless, different from [30], we look to five categories of indicators (education, employment, internet users, research activity and wealth distribution) and we resort to a dynamic analysis of the index (the NCHI is computed for each year considering different factor loadings). Further, we include among our variables the GINI index, showing the dispersion in wealth distribution.

Table 2
Individual indicators used in the NHCI computation

Category Indicators

Education Government expenditure on education, total (% of GDP)

Educational attainment, at least completed primary, population 25+ years, total (%) (cumulative)

Adolescents out of school (% of lower secondary school age)

School enrollment, tertiary (% gross)

Employment Employment to population ratio, 15+, total (%) (national estimate)

Informal employment (% of total non-agricultural employment)

Unemployment, total (% of total labor force) (national estimate)

Internet users Individuals using the Internet (% of population)

Research activity Research and development expenditure (% of GDP)

Technicians in R&D (per million people)

Wealth Distribution GINI index (World Bank estimate)

Category	Indicators
Education	Government expenditure on education, total (% of GDP)
	Educational attainment, at least completed primary, population 25+ years, total (%) (cumulative)
	Adolescents out of school (% of lower secondary school age)
	School enrollment, tertiary (% gross)
Employment	Employment to population ratio, 15+, total (%) (national estimate)
	Informal employment (% of total non-agricultural employment)
	Unemployment, total (% of total labor force) (national estimate)
Internet users	Individuals using the Internet (% of population)
Research activity	Research and development expenditure (% of GDP)
	Technicians in R&D (per million people)
Wealth Distribution	GINI index (World Bank estimate)

Source: World Bank Development Indicators.

To compute the index, we thus resort to a Common Factor Analysis (CFA). The purpose of CFA is to describe a set of Q variables (x₁, x₂, …, x_Q)with a smaller number of factors m. The model is: $\begin{matrix} x_{1} = α_{11} F_{1} + \dots + α_{1 m} F_{m} + e_{1} \\ x_{2} = α_{21} F_{2} + \dots + α_{2 m} F_{m} + e_{2} \\ \dots \end{matrix}$ (6) $x_{Q} = α_{Q 1} F_{Q} + \dots + α_{Qm} F_{m} + e_{Q},$ where x are the original, individual variables, α_i1, …, α_im are the factor loadings, F₁, …, F_Q are common (uncorrelated) factors, and e₁, …, e_q are specific (i.i.d) factors.

After the identification of eigenvalues, we proceed to the selection of factors retained in the analysis. Therefore, the system of Equations (6) can be rewritten as: $Y = α_{1} F_{1} + α_{2} F_{2} + \dots + α_{m} F_{m},$ (7) where F_m are the factors and α _m are the factor loadings.

Finally, to see if the sample is recommended for the CFA computation, we resort to the Kaiser-Meyer-Olkin (KMO) criterion [33]: $KMO = \frac{\sum \sum_j \neq k r_{jk}^{2}}{\sum \sum_j \neq k r_{jk}^{2} + \sum \sum_j \neq k p_{jk}^{2}},$ (8) where r_jk is the correlation level between variables, whereas the p_jk is the partial correlation.

The results indicate a KMO equal to 0.7, which recommends the use of CFA.

2.3 Panel data approach

We perform a multiple pooled OLS regression. Whereas the financial performance is captured by the return on assets (ROA), the firm size is estimated as the natural log of the firm total assets (size). For robustness purpose we use the profit margin (pm) as a proxy for firm’s financial performance.

Consequently, the two general equations we test are: $\begin{matrix} TFP - L P_{it} & = & c + β_{1} NHC I_{it} + β_{2} RO A_{it} \\ + β_{2} siz e_{it} + ɛ_{i}, \end{matrix}$ (9) $\begin{matrix} TFP - W_{it} & = & c + β_{1} NHC I_{it} + β_{2} RO A_{it} \\ + β_{2} siz e_{it} + ɛ_{i}, \end{matrix}$ (10) where c is the intercept, β are the regression coefficients, ɛ are the errors, I = 1 ... N (N = 116 firms).

3 Results

Table 3 presents the main findings of the empirical analysis. We first show that NHCI has a positive but marginal impact on the TFP for the companies included in the scientific research and development industry from Romania. This result, which remains robust under the two specifications of TFP (that is TFP – LP and TFP – W) indicate that the quality of education, the labor market, the access to internet, the R&D investment and the wealth disparity index have a positive but limited effect on the total factor productivity. Two elements might explain this result. Firstly, the NHCI does not record important dynamics during the analyzed period, whereas the productivity of the companies improved. Secondly, other elements related to managerial skills, institutional quality or the general business context might play a more important role in enhancing the TFP of companies. The NHCI is supposed to have a very long-run effect on firms’ efficiency.

Table 3
Main empirical findings – OLS regression

Levinsohn and Petrin Wooldridge

TFP – LP TFP – W

c 1.341* [0.831] 1.281* [0.644]

NHCI 0.002* [0.002] 0.003* [0.002]

ROA 0.785* [0.121] 0.812* [0.096]

size 0.233 [0.098] 0.263* [0.014]

R² 0.241 0.322

	Levinsohn and Petrin		Wooldridge
c	1.341***	[0.831]	1.281***	[0.644]
NHCI	0.002*	[0.002]	0.003*	[0.002]
ROA	0.785***	[0.121]	0.812***	[0.096]
size	0.233**	[0.098]	0.263***	[0.014]
R²	0.241		0.322

Notes: (i) ***p < 0.01, **p < 0.05, *p < 0.1; (ii) standard errors in square brackets; (iii) 1,055 observations.

Along these lines, it seems that the profitability of companies contributes to an increase in TFP. These findings are explained by the fact that higher profits can be used to sustain further investments, to reward the employees and to avoid external financing costs. Moreover, the size of the company has a positive and significant impact on its TFP. More precisely, an increase of 1% of the firm’s total assets (expressed in natural logarithm) leads to an increase of 0.23% of its TFP under [28] approach and to an increase of 0.26% when we resort to the [29] specification. Nevertheless, the explanatory power of the model (R²) is relatively reduced and indicates that other factors and driving forces are at work to explain the TFP dynamics.

Even if there is a large consensus between the results of the regression tested (Equations (9) and (10)), we still need to check the robustness of these estimations. To do so, we use an alternative indicator to assess firm’s financial performance. In our equations, we replace the profitability ratio (ROA) with the profit margins (pm). The new results, reported in Table 4, clearly demonstrate the robustness of the main findings. As in the previous case, profitability and firm’s size positively influences the TFP, in a very significant way. Moreover, the results are better if we look the coefficients of our interest variable (that is, the NHCI). Therefore, we may conclude that the investment of Government in education and R&D, the access to internet and a reduce level of wealth disparities might enhance the TFP of companies, but the effect is reduced. Indeed, these elements have a long-run impact on the firm productivity, but in the short run, their role cannot be neglected. At the same time, firms’ internal performances are essential for boosting their productivity.

Table 4

Robustness check results – profit margins

	Levinsohn and Petrin		Wooldridge
	TFP – LP		TFP – W
c	1.216***	[0.537]	0.988***	[0.450]
NHCI	0.017*	[0.008]	0.023**	[0.007]
ROA	0.677***	[0.138]	0.809***	[0.074]
size	0.254***	[0.019]	0.273***	[0.015]
R²	0.282		0.338

Notes: (i) ***p < 0.01, **p < 0.05, *p < 0.1; (ii) standard errors in square brackets; (iii) 1,055 observations.

4 Conclusions

The purpose of the paper was twofold. First, we have computed the TFP for the scientific R&D companies from Romania. Second, we have verified to what extent the national human capital, a component of the national intellectual capital index, influences the TFP.

Resorting to firm level data (116 private and public firms) for the period 2007 to 2016, we compute the TFP using two well-known approaches, namely [28] and [29]. Afterwards we perform a panel OLS estimation (pooled OLS) to see the impact if the NHCI on the TFP, while controlling for several elements as the firm profitability and firm size.

Our results, which are robust to different specifications of TFP and firm profitability, clearly indicate that the NHCI has a positive but reduced impact on firms’ productivity. At the same time, firms’ internal financial conditions are very important to boost their productivity. In addition, it seems that larger firms, which have in place complex management systems, are financially more stable and record a higher TFP.

Our analysis, even if it generates robust results, is characterized by several limits. Firstly, during the analyzed period a crisis episode affected the economy (with a pick recorded in 2009). Our linear specification cannot capture the effect of the crisis. The split of the sample in two sub-periods is not recommended for such short time span, whereas the use a dummy crisis variable is subjected to debates about the length of the crisis. Secondly, we do not verify the role of other elements that influence firms’ activity and the general business context. Therefore, the empirical analysis can be extended in at least two directions. It is possible to investigate the asymmetric effect the NHCI has on TFP. A panel quantile regression allows to see if the impact is higher or smaller for companies which record higher levels of TFP, compared to those with smaller productivity levels. At the same time, it will be useful to make the distinction between private firms and national research institutes active within this industry in Romania. This way, we can see if the private or public firms benefit more from investment in education, communication infrastructure and labor market’ reforms.

Footnotes

Acknowledgments

This work was supported by a Grant of the Romanian National Authority for Scientific Research and Innovation, CNCS–UEFISCDI, Project Number PN-III-P1-1.1-TE-2019-0436.

Author contributions

CONCEPTION: Claudiu Albulescu

INTERPRETATION OR ANALYSIS OF DATA: Serban Miclea and Claudiu Albulescu

PREPARATION OF THE MANUSCRIPT: Claudiu Albulescu

REVISION FOR IMPORTANT INTELLECTUAL CONTENT: Serban Miclea

SUPERVISION: Claudiu Albulescu

As [] mentions, human capital is becoming the most important factor in economic growth, supporting thus the economic development of a country.

Data were extracted in 2018.

We resort to the NHCI computation given that the Human Capital Index computed by the World Bank is available for Romania starting with 2017 only.

References

Jones

, Romer

. The New Kaldor facts: Ideas, Institutions, population, and human capital. American Economic Journal: Macroeconomics. 2010;2(1):224–45.

Albulescu

, Turcu

. Total factor productivity, financial performances and corporate governance: An analysis of the R&D sector in Romania. INFERWorking papers. 2018:09.

Xin

, Gang

, Qi

, Lei

, Jianqiu

. A study on customer knowledge management, inbound open innovation and firm performance. Human Systems Management. 2020;39(2):183–95.

Syverson

. What determines productivity? Journal of Economic Literature. 2011;49(2):326–65.

Janshanlo

, Nurgaliyeva

, Oralbaeva

, Dzhakisheva

, Aktureeva

. The content and role of human capital in the modern economy of Kazakhstan. Human Systems Management. 2020;39(1):81–92.

Corry

, Valero

, Van Reenen

. UK economic performance since 1997: Growth, productivity and jobs. Centre for Economic Performance, London School of Economics. 2011; Special Paper No. 24.

Ilmakunnas

, Miyakoshi

. What are the drivers of TFP in the Aging Economy? Aging labor and ICT capital. Journal of Comparative Economics. 2013;41:201–11.

Van Ark

. The measurement of productivity: what do the numbersmean? In: Gelauff G, Raes KS, Roelandt, T. (Eds.), “Fostering Productivity”. Elsevier, Amsterdam, 2004: pp. 29-61.

Venturini

. The modern drivers of productivity. Research Policy. 2015;44:357–69.

10.

Kancs

, Siliverstovs

. R&D and non-linear productivity growth. Research Policy. 2016;45:634–46.

11.

Ugur

, Trushin

, Solomon

, Guidi

. R&D and productivity in OECD firms and industries: A hierarchical meta-regression analysis. Research Policy. 2016;45:2069–86.

12.

Minniti

, Venturini

. R&D policy, productivity growth and distance to frontier. Economics Letters. 2017;156:92–4.

13.

Kong

, Kong

. Corporate governance, human capital, and productivity: Evidence from Chinese non-listed firms. Applied Economics. 2017;49(27):2655–68.

14.

Chiang

M-H

, Lin

J-H

. The Relationship between corporate governance and firm productivity: Evidence from Taiwan’s manufacturing firms. Corporate Governance. International Review. 2007;15(5):768–79.

15.

Schoar

. Effects of corporate diversification on productivity. The Journal of Finance. 2002;57(6):2379–403.

16.

Jiraporn

, Lee

, Park

, Song

. How do independent directors influence innovation productivity? A quasi-natural experiment. Applied Economics Letters. 2018;25(7):435–41.

17.

Donaldson

, Davis

. Stewardship theory or agency theory: CEO governance and shareholder returns. Australian Journal of Management. 1991;16:49–64.

18.

Bastos

, Monteiro

, Straume

. The impact of private vs. public ownership on the level and structure of employment. Economics of Transition. 2014;22(2):247–80.

19.

Boeing

, Mueller

, Sandner

. China’s R&D explosion—Analyzing productivity effects across ownership types and over time. Research Policy. 2016;45:159–76.

20.

González-Páramo

, Hernández De Cos

. The impact of public ownership and competition on productivity. Kyklos. 2005;58(4):495–517.

21.

Jefferson

, Rawski

, Wang

, Zheng

. Ownership, productivity change, and financial performance in Chinese industry. Journal of Comparative Economics. 2000;28:786–813.

22.

Dimelis

, Papaioannou

. Entry regulation, public ownership and TFP growth: Industry-level evidence from south European countries. The Manchester School. 2016;84(6):749–70.

23.

Chen

, Guariglia

. Internal financial constraints and firm productivity in China: Do liquidity and export behavior make a difference? Journal of Comparative Economics. 2013;41(4):1123–40.

24.

Ferrando

, Ruggieri

. Financial constraints and productivity: Evidence from euro area companies. International Journal of Finance & Economics. 2018;23(3):257–82.

25.

, Dosi

, Grazzi

, Lei

. Inside the virtuous circle between productivity, profitability, investment and corporate growth: An anatomy of Chinese industrialization. Research Policy. 2017;46:1020–38.

26.

Andriessen

, Sense of intellectual capital. Designing a method for the valuation of intangibles. Elsevier Butterworth-Heinemann, London; 2004.

27.

Bueno

, Salmador

, Rodriguez

. The role of social capital in today’s economy. Empirical evidence and proposal of a new model of intellectual capital. Journal of Intellectual Capital. 2004;5(4):556–74.

28.

Levinsohn

, Petrin

. Estimating production functions using inputs to control for unobservables. Review of Economic Studies. 2003;70(2):317–41.

29.

Wooldridge

. On estimating firm-level production functions using proxy variables to control for unobservables. Economics Letters. 2009;104(3):112–4.

30.

Weziak

. Measurement of national intellectual capital – application to EU countries. IRRIS Workin papers 2007-13.

31.

Maddala

, Wu

. A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and Statistics. 1999;61(Supplement 1):631–52.

32.

Choi

. Unit root tests for panel data. Journal of International Money and Finance. 2001;20:249–72.

33.

Kaiser

. An index of factorial simplicity. Psychometrika. 1974;39:31–6.