Nexus Between Institutions, Technological Efficiency and Labor Productivity: A Framework of Augmented Solow Model

Abstract

This paper adopts the Augmented Solow model as the core empirical framework. It aims to examine the influence of institutions as moderators in the relationship between technological efficiency and labor productivity. The existing literature has not sufficiently integrated technological efficiency within the Augmented Solow framework, and they also tend to examine these dimensions in isolation rather than capturing their interactive (moderating) effects on labor productivity. To achieve the objective of this study, countries were classified into High, Upper-Middle and Lower-Middle income countries. The model examines the vast gap between the countries and finds that, according to their income, the gap will widen due to destructive-applicable institutions. This study applies a panel analysis using the data for 175 countries during the period 1990–2019. The results provide recommendations for addressing these challenges to enhance output and promote long-term economic growth. The empirical results show that the quality of the institutional context determines its impact. In particular, the interaction term shows that while efficiency benefits are dampened under worse institutional settings, they are greatly magnified in contexts with above-average institutional quality. As a result, this study emphasizes that policies that only aim to increase productivity or encourage the adoption of new technologies are inadequate without a robust and well-functioning institutional environment.

1. Introduction

The enormous labor and capital mobility worldwide shifted the attention to examine the variations in economic performance and living standards. Average labor productivity, a measure of economic performance, highlights the nature of the labor market, human capital and their role in promoting living standards and economic growth (Van Ark & McGuckin, 1999). However, economies tend to invest in new techniques in production processes, adopting technological efficiency to achieve sufficient utilization of resources. Lessening waste in technological resources, such as wasting hardware, software, expertise, and associated financial resources, can be achieved under the umbrella of good institutions. According to Acemuglo and Johnson, institutions do explain the variations in economies’ procedures and performances (Acemoglu et al., 2005). Recently, institutions were referred to as the key solutions to the vulnerable issues in the literature concerning how income per capita varies across countries and how to improve economic performance. Consequently, analyzing the institutional framework within which the economy operates has become an indispensable subject of research. Globally, the cause of income inequality is explained based on technological disparities. The use of efficient technologies paves the way to producing lower-cost and higher-quality goods, accumulating capital, and boosting competitiveness. Thus, technological efficiency was addressed as an influencing variable affecting labor productivity in a moderating institutional environment that either enhances growth, applying efficiency principles and improving process management, or deteriorates productivity and increases costs. Therefore, considering institutions improves the quality of research and contributes to addressing the cultural and political development of societies (Todorova, 2022).

More recent strands of the literature emphasize the role of technological efficiency, the ability to optimally utilize existing technologies, as a key determinant of output performance, especially in developing and emerging economies. However, a growing body of research highlights the importance of institutions in shaping economic outcomes. Seminal contributions by Acemoglu et al. (2001) argue that institutional quality, encompassing governance effectiveness, regulatory quality, and rule of law, constitutes a fundamental determinant of economic performance. Despite this theoretical consensus, the empirical findings remain mixed. While some studies document a strong positive effect of institutions on productivity and growth (Rodrik et al., 2004), others find that institutional variables exert weak or context-dependent effects, particularly when interacting with technological and structural factors.

This study adopts the Augmented Solow model as an extension for the Neoclassical growth approach to analyze variations in income and growth rates between different countries. The Augmented Solow model, developed by (Mankiw et al., 1992), was extended to provide an explanation for cross-country differences in income per capita by incorporating technology into the model. The focal point has been on the impact of institutions on overall economic performance (such as growth and development as an output of economic performance) and not on the determinants of growth as inputs of economic performance. Institutions have large effects on certain aspects of economic performance; (Hall & Jones, 1999) recognized the critical role of institutions as elements of growth and enhanced components of traditional growth determinants (labor, capital and technology). Thus, the study suggests that the biggest contributor in technological inefficiency in developing countries is the degree of their institutional quality. Institutions control educational systems, employment, wages and policies that enforce corporations to train their employees. (Tebaldi & Mohan, 2009) concluded that differences in the quality of institutions according to the Augmented Solow model can determine the output per worker growth rate and level.

According to the World Bank’s classifications using the Gross National Income (GNI) per capita, countries are classified into four main categories: Low Income, Middle-Low Income, Middle-High Income, and High Income. Countries below the High-income level are considered “developing”. Development is an income-dependent factor. In 1981, the term “emerging” economies, introduced by Antonie W. Van Agtmael, defined these economies as developing countries with low- to middle-per capita income. “Emerging” and “developing” economies vary according to their implemented reforms, rates of economic growth, and their engagement in the global economy (Campbell & Ahmed, 2012). Tracking the growth of economic structures over time challenges developing countries to enhance their productivity by addressing issues including the inability to meet the needs of the labor market and the insufficient allocation of resources to meet the needs of the consumption market. On the contrary, developed countries have been able to achieve higher levels of productivity through investments in education, infrastructure, and innovation. They also benefit from more efficient markets and good governance. Consequently, understanding the causes and consequences of the productivity gap is crucial for policymakers and development experts seeking to improve economic growth rates and reduce global inequality.

This divergence has given rise to an important debate in the empirical literature: are institutions a direct driver of productivity, or do they primarily operate as a moderating factor? Recent studies increasingly suggest the latter, emphasizing institutional complementarities, whereby the effectiveness of technological inputs depends on the surrounding institutional environment (Aghion & Howitt, 2009). Nevertheless, the empirical investigations explicitly modeling this interaction mechanism, particularly using standardized institutional measures and cross-country panel data, remain relatively limited. Moreover, insufficient attention has been paid to the interpretation of interaction effects, especially in the presence of transformed variables, which often lead to ambiguity in policy conclusions.

The current study seeks to address these gaps by examining the moderating role of institutional quality in the relationship between technological efficiency and labor productivity. Specifically, this study aims to (i) assess the direct impact of technological efficiency on the output per worker; (ii) evaluate the extent to which the institutional quality conditions this relationship; and (iii) provide a rigorous interpretation of interaction effects using standardized variables to enhance comparability and econometric robustness.

The contribution of this research is threefold. First, it advances the empirical literature by explicitly modeling the interaction between efficiency and institutions, thereby moving beyond additive specifications that may obscure conditional relationships. Second, by employing standardized institutional indicators, this study improves the interpretability of coefficients and addresses multicollinearity concerns commonly associated with interaction terms. Third, the analysis offers policy-relevant insights by distinguishing between short-term efficiency-enhancing measures and long-term structural reforms, highlighting the moderating role of institutional quality as a transmission channel through which technological gains are converted into productivity improvements.

2. Literature Review

2.1. Theoretical Foundations of Productivity and Technological Efficiency

The relationship between technological progress and productivity has long been central to economic growth theory. Technological efficiency is defined as the ability of an economy or firm to produce maximum output using a given set of inputs. Efficiency improvements reflect movements toward the production frontier rather than shifts in the frontier itself. However, contemporary theoretical developments suggest that such efficiency gains are not automatic and depend on complementary structural conditions, particularly institutional quality. A large body of the empirical literature confirms a positive association between technological advancement and labor productivity. Early evidence from growth accounting frameworks shows that productivity gains are strongly linked to technological progress and efficiency improvements (Solow, 1956; Romer, 1990). However, more recent empirical studies highlight significant heterogeneity in this relationship across countries, sectors, and institutional environments.

Institutional economics, pioneered by North (1990), conceptualizes institutions as the formal and informal “rules of the game” that shape economic behavior and reduce transaction costs. Building on this perspective, Acemoglu et al. (2001) argue that institutional quality is a fundamental determinant of long-run economic performance, influencing incentives, investment decisions, and resource allocation efficiency. More recent theoretical contributions emphasize the concept of institutional complementarities, whereby technological efficiency and institutional quality jointly determine productivity outcomes. In this view, institutions not only affect economic performance directly but also condition the effectiveness of technological and efficiency-enhancing processes (Aghion & Howitt, 2009).

2.2. Institutional Quality and Economic Performance

Institutional quality has been widely recognized as a key determinant of economic development and productivity. Cross-country empirical studies consistently show that governance indicators, such as rule of law, regulatory quality, and government effectiveness, are strongly associated with higher levels of output and productivity (Acemoglu et al., 2001; Rodrik et al., 2004).

North (1990) emphasizes that institutions reduce uncertainty and transaction costs, thereby facilitating efficient economic exchange. In modern empirical applications, institutional quality is often found to enhance investment efficiency, improve innovation outcomes, and strengthen productivity growth. Recent studies further extend this argument to digital and technological contexts. The evidence suggests that institutional quality plays a critical role in shaping the productivity effects of digital transformation, particularly by improving regulatory enforcement and facilitating efficient allocation of resources in the economy (Zhang & Zhang, 2025; Lu et al., 2025).

2.3. Interaction Between Technological Efficiency and Institutions

An emerging strand of the literature moves beyond direct effects to examine the interactive role of institutions in shaping technological outcomes. This perspective argues that technological efficiency alone is insufficient to generate sustained productivity gains unless supported by a strong institutional framework.

Empirical studies increasingly support this view. For instance, the research shows that the productivity effects of ICT adoption and digital technologies are significantly stronger in countries with high institutional quality, as better governance reduces inefficiencies and enhances the diffusion of technology (Fulgenzi et al., 2024). Similarly, studies on innovation systems suggest that institutional strength amplifies the economic returns of technological investment by improving coordination, reducing uncertainty, and ensuring effective implementation of economic policies.

At the macro level, cross-country analyses confirm that the relationship between technological development and productivity is nonlinear and strongly conditioned by institutional environments. Weak institutions can dampen or even neutralize the productivity gains from technological efficiency, while strong institutions significantly enhance these effects (Wu & Kikuchi, 2025).

Recent econometric evidence further confirms the presence of moderating effects, where institutional quality strengthens the impact of technological innovation and efficiency on economic growth and productivity (Nguyen & Be, 2025).

2.4. Research Gaps in Literature

Despite the growing body of research, several gaps remain. First, many empirical studies rely on additive models that fail to capture the conditional nature of the technology–institution relationship. This limits the understanding of how institutional environments shape the effectiveness of technological efficiency. Second, relatively few studies explicitly distinguish between technological efficiency and technological innovation in explaining productivity differences, despite their distinct theoretical implications. Third, there is limited attention to methodological issues related to standardized institutional variables, particularly in interaction models, where the interpretation of coefficients requires careful econometric treatment. Finally, much of the existing literature remains fragmented across firm-level and macro-level studies, with insufficient integration of cross-country evidence on institutional moderation effects.

3. Methodology

The research methodology consists of a combination of two types: quantitative research, which focuses on collecting and analyzing numerical data using statistical methods to measure phenomena and generalize results, and qualitative research, which aims to gain a deeper understanding of phenomena by collecting and analyzing previous studies to understand experiences and perspectives. This model aligns with the Augmented Solow model framework by considering technological efficiency (A) as a measurable variable using Data Envelopment Analysis (DEA). DEA is used to estimate the technical efficiency and overall productivity of factors, rather than treating them as unobserved residuals. Labor productivity is interpreted as a direct result of the efficiency level, while institutional factors play a crucial role in determining an economy’s position relative to the efficiency threshold, either by including them as non-optional inputs or through a two-stage analysis to explain efficiency differences between countries. This model focuses on how technological efficiency in the presence of institutions helps to drive labor productivity in the framework of the Augmented Solow model. It points out that the output per worker is determined by institutions. Institutions, in turn, can either stimulate or destruct productivity in the economy.

To provide a clear positioning of the proposed framework within the growth literature, a comparative table of growth models was constructed.

Table 1. Empirical Studies of growth theory.

Growth Theory	Authors	Variables of the Study	Conclusions
Classical Theory	MacDougall (1956)	Compared labor productivity and trade patterns in the US.	Concluded that countries tend to export goods that are of a comparative advantage to them, supporting aspects of David Ricardo’s classical economic theory.
Neo-classical theory	Sala-i-Martin and Barro (1995)	Employed cross-country regression analysis, convergence analysis, and panel data analysis.	Highlighted that the neo-classical growth theory explains the convergence of poorer countries to the income levels of richer countries.
	Mankiw et al. (1992)	Employed cross-country regression analysis of the relationship between growth and human capital, technology advancement, and physical capital.	Concluded that incorporating human capital into the growth model would improve income differences.
Endogenous Theory	Hall and Jones (1999)	Employed a regression model to the relationship between output per worker and human capital and physical capital, which helps to measure the Total Factor Productivity (TFP) of output growth not explained by capital and labor inputs.	Concluded that the cross-country variation in productivity could be identified by differences in human capital and institutions that support knowledge creation.
	Mankiw et al. (1992)	Employed the Augmented Solow model to measure the relationship between human capital, technological progress, and economic growth.	Concluded that human capital has a significant role in driving economic growth.

Source: (Chukwuemeka, 2024).

presents a classification of the literature based on growth theory.

This table highlights the key differences in the assumptions regarding technological progress, efficiency, and institutional roles. Unlike traditional models that treat Total Factor Productivity as an unobserved residual, the current study integrates Data Envelopment Analysis (DEA) to explicitly measure efficiency and technological performance. This approach allows for a more precise empirical representation of the Augmented Solow framework and strengthens the methodological contribution of this study.

This study utilizes a panel dataset for 175 countries spanning from 1990 to 2019. The starting year of 1990 was chosen to enable the inclusion of more sample countries. The following

Table 2. Variables of this study.

Variable	Description	Data Source
Labor productivity (LP)	Indexed by output per worker.	World Bank
Technical efficiency (TE)	Measures economic output per unit of labor input and affects the way of production.	Calculated using DEA
Institutions (INS)	Proxied by overall score of governance.	World Bank

Source: made by the researcher.

defines the variables used in the study, including the data sources used, which were obtained from reliable international databases to ensure the accuracy and comparability of the results.

This study is classified into two procedures. Firstly, measuring the efficiency variable using DEA. Secondly, measuring the moderating effect of institutions on labor productivity (dependent variable) and technological efficiency (independent variable) through categorizing countries into three sectors (High income, Upper-Middle income and Lower-Middle income). In this section, the first procedure is explained as follows:

Stage (1): estimate the technological efficiency variable using the DEA model among the sample countries to calculate the technological efficiency score, which is considered an independent variable.

There are two approaches used in estimating technological efficiency: parametric and non-parametric methods. SFA is a basic example of the parametric approach that separates technological inefficiency and random error (Aigner et al., 1977; Battese & Coelli, 1992). In contrast, DEA is a mathematical programming non-parametric approach to construct production frontiers and measure efficiency relative to the constructed frontiers. DEA assists decision-makers by calculating efficiency rates, showing whether a firm or industry or country is efficient or has the potential for improvement by setting input and output targets (inputs needed to be reduced or output needed to be increased).

The DEA approach was used to test the country’s Decision-Making Unit (DMU), in which it converted multiple inputs and outputs into a scalar measure of efficiency (from 0 to 1) to identify the efficient (good) performers creating a benchmark frontier and the inefficiency of others creating their own frontier in comparison to these good performers; this is extensively used in economics and operational research (Seiford & Zhu, 1998). The distance between the benchmark frontier and the other one can be determined as inefficiency. However, technological efficiency can be used as a measure of aggregate performance, but it does not indicate the source of the inefficiency, which could be one of the following:

-

Resources might be misallocated.

-

Resources might be operating on a non-sufficient scale.

According to this theory, technological efficiency refers to the DMU’s ability to apply the frontier, and that deviations from these boundaries represent inefficiencies. This procedure was developed by (Farrell, 1957), who estimated the boundaries of the countries’ output or production and was further explored by (Charnes et al., 1981) and (Battese & Coelli, 1992), who analyzed technological efficiency using the DEA method. The method, therefore, assumes that there are N countries that produced Y output through N inputs to understand the exact approach of multiple inputs and outputs, within the standard non-parametric approach (Coelli et al., 2005).

Benchmark economies can be used as case studies for their performance to enhance the performance of other economies. Emerging countries may exhibit increasing returns to scale, indicating potential benefits from scaling up operations. Meanwhile, developed countries are more likely to operate at constant returns to scale, meaning they are already optimized for their scale of operations using inputs (physical capital and human capital).

DEA models can assume Constant Returns to Scale (CRS) or Variable Returns to Scale (VRS). CRS assumes that outputs will change proportionally with inputs, while VRS allows for increasing or decreasing returns to scale (Charnes et al., 1981). The VRS model is more reliable for several reasons:

• It calculates a variety of input and output variables.
• It does not assume a functional relationship between input and output variables.
• It accommodates different measurement sets for input and output variables.

Efficiency under VRS involves increasing, constant, and decreasing returns to scale, unlike the CRS assumption in a traditional DEA model. The VRS assumption is more realistic for evaluating a country’s scale efficiency as it considers convex combinations where inputs and outputs are benchmarked against similar-sized DMUs (Sulaiman & Ismail, 2021). VRS is appropriate when all of the countries do not operate at an optimal scale (Coelli et al., 2005). DEA models can assume Constant Returns to Scale (CRS) or Variable Returns to Scale (VRS). CRS assumes that the output will change proportionally with inputs (Charnes et al., 1981), while VRS allows for increasing or decreasing returns to scale.

This study applied DEA using the Cobb Douglas production function in a panel dataset of 175 countries during the period 1990–2019. The data were obtained from the World Bank database, which provides more reliable evidence on performance because the variance in time enables the researcher to track their performance, as shown in the following equation:

$$Y=a{K}^{b}H{L}^{1-b}$$

(1)

The worker’s share of real GDP (Y): expressed in output per worker in PPP terms (at constant prices).

Physical capital stock per worker (K): the Perpetual Inventory Method (PIM) is used to calculate a capital stock chain for each country from the gross fixed capital formation index (at constant prices).

The human capital stock per worker (H): the total labor force variable, weighted by the average years of study.

EViews 12 statistical package was utilized to estimate the results of this study. Technological efficiency using the DEA model measures the fraction of the potential output produced by the country and highlights the relative value of the country’s efficiency score in the range of 0 to 1, which indicates that the value of 1 is efficient and that a value less than 1 is inefficient. A country is expected to maximize the output produced based on the available technology. The data can be summarized in the following

Table 3. Efficiency scores’ frequencies in developing and developed countries.

Frequency	0	0.1–0.49	0.5–0.59	0.6–0.69	0.7–0.79	0.8–0.89	0.9–0.99	1.00
Developing	653	1494	433	277	300	270	145	12
Developed	0	292	87	152	252	350	95	2

Source: made by the researcher.

as efficiency frequency in a sample of 4800 during the period 1990–2019.

The efficiency scores range from 0 to 1, where 1 indicates that the DMU is efficient and a benchmark. Lower efficiency scores than the benchmark might be more common for most developing countries, and they also identify gaps and potential areas for improvement due to limited resources, less advanced technology, and lower levels of human capital. Meanwhile, higher efficiency scores for most developed countries act as reference points for improving efficiency due to better access to resources, advanced technology, and higher levels of human capital. Table 3 presents the technological efficiency frequency by country from 1990 to 2019.

As shown in Table 3, the data shows a clear distinction between the developed and the developing countries in terms of efficiency. The developed countries tend to have higher efficiency scores, with the majority in the range 0.8–0.89, while the developing countries have a wider distribution across all ranges but a significant number with a score of 0. The developed countries generally show higher efficiency scores compared to the developing countries, but note that the level of technology used in the developed countries is more advanced than that which is used in the developing countries.

As shown in

Table 4. Percentage of efficiency scores in developing and developed countries.

Percentage	0	0.5–0.59	0.6–0.69	0.7–0.79	0.8–0.89	0.9–0.99	1.00
Developing	31.24	20.72	13.25	14.35	12.92	6.94	0.57
Developed	0	9.28	16.2	26.87	37.31	10.13	0.21

Source: made by the researcher.

, the largest percentage of efficiency scores in developing countries fall within the near range (0%). The developed countries tend to have higher efficiency scores, with the majority in the range (0.8–0.89%). The developed countries generally show higher efficiency percentage scores compared to the developing countries.

Table 5. Key observations from the frequency and percentage Table 3 and Table 4.

Developing Countries	Developed Countries
The majority (31.24%) have an efficiency score of 0	No developed countries have an efficiency score of 0
There is a decreasing trend in the frequency as the efficiency score increases	The highest concentration (37.31%) falls in the 0.8–0.89 efficiency range
Only 0.57% of developing countries achieve the highest efficiency score of 1	Only 0.21% achieved the highest efficiency score of 1

Source: made by the researcher.

presents the key observations as concluded from the previous Table 3 and Table 4.

As shown in Table 5, it can be observed from the frequency and percentage tables that there is a decreasing trend in the frequency as the efficiency score increases in the developing countries. The developed countries generally show higher efficiency scores compared to the developing countries, but note that no developed countries have a zero-efficiency score.

According to the previous three tables, it can be concluded that some countries are doing better than others. In the framework of the Augmented Solow model explanation, these variations are due to high investment rates in physical capital, spending a large fraction of time on education, having low population growth rates and having high levels of technology in countries with higher rates of outcomes. This model was selected as it declared that the source of growth is the accumulation of physical and human capital, in addition to technological efficiency, as it appears to be a stimulating factor for labor productivity. This suggests that the developed countries generally perform better in terms of efficiency. Meanwhile, the developing countries have a significant number of cases with very low efficiency (score of 0), while developed countries do not have any such cases. However, both groups have a relatively small percentage of countries achieving perfect efficiency (score of 1.00). Achieving perfect efficiency is rare for both developed and developing countries, suggesting that there is always room for improvement in efficiency practices and policies. After calculating the “efficiency variable” using DEA, the second procedure was applied.

Stage (2): examine the role of institutions as moderators in the relationship between technological efficiency and labor productivity.

Calculating the moderating role of institutions in the relationship between technological efficiency and labor productivity in the framework of the Augmented Solow model is a multidimensional stage which can be divided into three subsections.

1. Calculating the Coefficient of Variation (CV) for each sector.

The Coefficient of Variation (CV)1 is an important tool in data analysis in studies. It is used to determine the extent of dispersion of the data around the mean. A variable with a lower Coefficient of Variation is more homogeneous. The CV for the efficiency and output variables was used for the three-year panel data on countries categorized as Higher, Middle, and Lower income according to the World Bank’s classifications based on recent data. According to results in the Appendix C, CV3 (Lower–Middle income) refers to the most homogenous variable showing less dispersion from its mean, which indicates a strong model. The next step is calculating the Pearson Correlation coefficient.

2. Calculating the Pearson Correlation Coefficient (r)

The Pearson Correlation Coefficient (r) measures the strength and direction of the relationship between two variables (X and Y), and its value ranges among (+1, 0, −1). It can be calculated through the following steps:

• Calculating the mean for each variable (µ).
• Calculating the standard deviation (ε).
• Calculating the summation of standard deviation for each.
• Applying the rule of the Pearson Correlation Coefficient:

  $$r= {\displaystyle \frac{\sum \left(xi- \widehat{x}\right) \ast (yi- \widehat{y})}{\surd \sum \left(xi- \widehat{x}\right)^2 \ast (yi- \widehat{y})^2}}$$

The result from the previous equation is r = +1. This means that there is a strong correlation (direct relationship) between the variables. That is, as variable X increases, variable Y increases at a constant rate. That is, as technological efficiency increases, productivity increases.

4. Results

To measure the impact of institutions as moderators in the relationship between technological efficiency and labor productivity in the framework of the Augmented Solow model, this study conducts a moderated regression analysis. The analysis involves Pooled Least Squares (PLS), the fixed effect and the random effect model. This study conducts a hypothesis test under the panel regression analysis to evaluate the best estimator of the models. The hypothesis is carried out through the Hausman test to select the fixed-effects and the random effect model.

In this case, H₀: α_i = X_it, Z_i and H₁: α_i ≠ X_it, Z_i, where H₀ is a random effect and H₁ is a fixed-effect model. The null hypothesis is dismissed if the p-value is significant (p < 0.05), suggesting that the fixed effect is the better estimator model.

The general model of the panel regression is expressed in the form of Equation (2):

$$Y_{it}= {\alpha }_i+X_{it}\beta +{\mu }_{it} t=1,\dots ,N$$

(2)

Institutions are consistent with a moderating role that influences the strength or direction of the relationship between the independent variable (technological efficiency) and the dependent variable (labor productivity) under the framework of panel regression and GLM. In this case, the dataset has the following:

• An index of labor productivity (output per worker).
• Technical efficiency scores.
• Institutions.

For moderation analysis, an interaction term between the independent variable and the moderator is needed.

Interaction Term = TE × Inst

(3)

Hence, the regression is conducted with the following model:

$${LP}_{it} = {\beta }_0 +{ \beta }_1\left({TE}_{it}\right) +{ \beta }_2\left({Inst}_{it}\right) +{ \beta }_3(Interaction Term) + \epsilon t$$

(4)

where

• β₀: the y-intercept;
• ${\beta }_1$: effect of technological efficiency on labor productivity;
• β₂: direct effect of institutions on labor productivity;
• β₃: moderating effect of institutions on the relationship between technological efficiency and labor productivity; if β3 (coefficient of the interaction term) is statistically significant, institutions moderate the relationship;
• εt: random error;
• LP_it: labor productivity represented in output per worker;
• TE_it: technological efficiency, as calculated using DEA;
• Inst_it: institutions, proxied by indicators including overall score of governance.

To examine the role of institutions as moderators in the relationship between technological efficiency and labor productivity, estimation of model (4) was carried out by disaggregating the sample countries’ quality of institutions into proxies for each country. To measure institutions, the literature has focused on several sets of variables (Glaeser et al., 2004), such as the set used by (Knack & Keefer, 1995) and the set used by (Kaufmann & Kraay, 2010). If technological efficiency increases with the increase in the quality of institutions, this will mean that institutions are boosted in the relationship between technological efficiency and labor productivity and vice versa.

Table 6. Descriptive statistics analysis of this study, 1990–2019 (n = 29).

	Obs.	Mean	Median	Std. Dev.	Min	Max	Normality Test
Dependent Variables: Output per worker	3685	11.12	11.01	2.151569	5.149194	16.83947	(17.12339) *
Independent Variables: Technological efficiency	3685	0.90	0.90	0.049944	0.665857	1.00000	(4.129106) *
Moderator Variable (Interaction term): Efficiency × Institutions	3685	3.24	−0.15	0.816612	−1.791681	1.799392	(148.8411) *
Control Variables: Inst_Overall score	3685	9.20	−0.17	0.902494	−2.015045	1.946802	(152.8616) *

Note: * indicates significance at 1%.

exhibits the descriptive statistics of the variables used in this analysis. The table shows the descriptive statistics including means, medians, standard deviations, and other relevant metrics as calculated by the Statistical Package EViews.

According to Table 6, the institutional quality variable is expressed in standardized (Overall-score) form, where the values represent deviations from the sample mean. Negative values, therefore, indicate below-average institutional performance rather than absolute negative levels. Similarly, the interaction term captures the joint effect of technological efficiency and institutional quality; negative values reflect efficiency operating under relatively weak institutional conditions. This transformation facilitates interpretation and mitigates multicollinearity in the models including interaction effects.

Table 7. Summary of the descriptive statistics analysis outcomes.

1. Dependent Variable: Output per Worker The average level of output per worker is 11.12, with a median of 11.01, indicating a relatively symmetric distribution around the central tendency. However, the relatively wide range (from 5.15 to 16.84) and a standard deviation of 2.15 suggest substantial cross-country heterogeneity in productivity levels. This dispersion is expected given the inclusion of multiple countries at different stages of development. The normality test is statistically significant at the 1% level, indicating departure from normal distribution, which may reflect structural differences across economies or the presence of outliers. This justifies the use of robust estimation techniques in subsequent analysis. 2. Independent Variable: Technological Efficiency Technological efficiency exhibits a high mean value of 0.90, with a very small standard deviation (0.0499), suggesting that most countries operate relatively close to the technological frontier. The narrow range (0.67–1.00) further confirms limited variability, implying that differences in efficiency across countries are relatively modest. Despite this concentration, the normality test is significant, indicating slight deviations from normality, possibly due to clustering near the upper bound (efficiency scores approaching 1). This limited dispersion may also imply that the direct effect of efficiency alone could be insufficient to explain large productivity differences, reinforcing the importance of interaction effects in the model. 3. Moderator Variable: Interaction Term (Efficiency × Institutions) This represents the moderating variable, which may moderate the relationship between the independent variables and the dependent variable. The interaction term displays notable distributional features. While the reported mean (3.24) appears inconsistent with the median (−0.15) and the range (−1.79—1.80), this likely reflects scaling or reporting differences, and the median provides a more reliable indication of central tendency in this case. Importantly, the presence of both positive and negative values confirms that the interaction term captures heterogeneous institutional environments in which efficiency operates. Negative values indicate contexts where technological efficiency is combined with below-average institutional quality, whereas positive values reflect more supportive institutional settings. The relatively higher standard deviation (0.82) compared to the efficiency variable suggests that institutional variation is a key source of heterogeneity in the interaction effect. The highly significant normality test statistic further indicates a non-normal distribution, likely driven by cross-country disparities in institutional quality. 4. Control Variable: Institutional Quality (Inst_Overall Score) The institutional variable exhibits a mean that appears inconsistent with its range and median (mean = 9.20, median = −0.17, range ≈ −2.02 to 1.95), strongly suggesting that the variable is standardized and that the mean may be affected by scaling or reporting conventions. The median close to zero confirms that the variable is centered around the sample average. The distribution spans both negative and positive values, reflecting variation in institutional performance relative to the sample mean. Negative values indicate below-average institutional quality, and positive values indicate above-average institutional quality. The relatively large standard deviation (0.90) indicates substantial cross-country institutional heterogeneity, which is crucial for identifying the moderating effect in the empirical model. The strongly significant normality test suggests non-normality, consistent with the uneven distribution of institutional development across countries.

Source: made by the researcher.

presents a summary of the explanations and interpretations for the outcomes as previously mentioned in Table 6.

The next step is to check the correlation using the correlation matrix between the variables, as shown in

Table 8. Correlation matrix between study variables.

	OUTPUT_LN_Y	EFFICIENCY	INST_OVERALL_SCORE	INTERACTION
OUTPUT_LN_Y	1.000000	0.536238	0.158809	0.168050
EFFICIENCY	0.536238	1.000000	−0.001121	−0.010623
INST_OVERALL_SCORE	0.158809	−0.001121	1.000000	0.998438
INTERACTION	0.168050	−0.010623	0.998438	1.000000

Source: made by the researcher using Statistical Package EViews 12.

.

Table 8 shows the correlation coefficients between the variables of this study, and these coefficients were interpreted as shown in

Table 9. Interpretation of correlation coefficients.

Pearson Correlation Coefficient (r)	Correlation Strength
0.00 < r ≤ 0.19	Very weak correlation
0.20 ≤ r ≤ 0.39	Weak correlation
0.40 ≤ r ≤ 0.59	Intermediate correlation
0.60 ≤ r ≤ 0.79	Strong correlation
0.80 ≤ r ≤ 1.00	Very strong correlation

Source: Papageorgiou (2022).

.

As previously shown in Table 8, Table 9 and

Table 10. Interpretation of Pearson correlation coefficients.

Correlation Strength	Pearson Correlation Coefficient (r)
0.10 ≤r ≤ 0.29	Poor correlation
0.30 ≤r ≤ 0.49	Intermediate correlation
0.50 ≤r ≤ 1.00	Strong correlation

Source: Papageorgiou (2022).

, the correlation matrix was used to check for a correlation between the variables of this study. From Table 8, the correlation coefficients between variables shows that efficiency is closely related to labor productivity indexed by Output_Ln_Y, as shown in the economic theory that efficiency increases productivity. The Output_Ln_Y and efficiency coefficients have a strong correlation at a value of (0.536). The moderating variable—the interaction term—is positively correlated with Inst_overall_score (0.99), suggesting that it captures aspects of institutions that can influence economic outcomes. In conclusion, the correlation matrix suggests that efficiency is the most important driver of output, followed by the moderating variable—the interaction term. The relatively counterintuitive relationships between Inst_Overall_Score and Output_Ln_Y highlight the complex nature of these variables in the dataset.

Then, the stationarity of the variables was checked. The Panel unit root test was conducted to confirm the stationarity of the variables. The results of the unit root tests are provided in

Table 11. Unit root test.

Variable	Stationary	Panel Unit Root Test	Prob. **	Results
Output_LN_Y	Level	−3.87894	0.0001	I(0)
EFFICIENCY	Level	−7.59202	0.0000	I(0)
INST_Overall_Score	Level	−8.99295	0.0000	I(0)
Interaction Term	Level	−9.43461	0.0000	I(0)

** refers to level significance at 1%. Source: made by the researcher using Statistical Package EViews 12.

.

As the p-values are less than 0.05, we reject the null hypothesis of the Panel unit root test of stationary of the variables. Table 11 presents the results of a Panel unit root test, specifically the Panel unit root test, applied to several variables at level. The interpretation of the Panel unit root test can be summarized as the following:

Stationary variables (I(0)): LN_Y (logarithm of output) and efficiency are stationary at their levels.

Stationary variables (I(0)): INST_Overall_Score and the interaction term are stationary at level.

After confirming the correlation matrix, it was utilized to check for the multicollinearity of the variables and their stationarity. The Generalized Linear Model is often considered the best model for dealing with autocorrelation in panel data because it effectively handles both endogeneity and autocorrelation. The method is consistent and efficient, making it a robust choice for panel data analysis.

Table 12. Panel Generalized Linear Model.

Variable	Coefficient	Std. Error	Prob.
Efficiency	24.96617	0.568473	0.0000
Overall_Score	−11.04028	0.561925	0.0000
Interaction	12.64241	0.621218	0.0000
C	−11.37476	0.512510	0.0000
Root MSE	1.689059	Mean dependent var	11.09703
Sum squared resid	10,393.19	S.D. dependent var	2.146631

Source: made by the researcher using Statistical Package EViews.

represents the GLM outcomes as follows:

The interpretation of the table can be summarized as the following:

Coefficients and significance efficiency (24.96617): This variable has a statistically significant and positive coefficient (p-value = 0.0000). This implies that an increase in efficiency leads to an increase in the log of the output, suggesting that efficiency plays a crucial role in driving productivity or output growth in the panel.

Overall_Inst_Score (−11.04028): This variable has a negative and significant coefficient (p-value = 0.0000). The negative relationship suggests that an increased Overall_Inst_Score might be associated with a reduction in output growth in this specific model, which could reflect the impact of governance reforms or efficiency in different contexts.

Interaction_term (Efficiency*Overall_Score) (12.64241): This variable is significant (p-value = 0.0000) and positively related to output. The result suggests that improving the Interaction_term fosters output growth, indicating that a stable political environment positively impacts economic performance.

5. Discussion

The empirical findings provide robust evidence for the conditional role of technological efficiency in enhancing labor productivity within the examined economies. The relatively low Root Mean Square Error (RMSE = 1.689059) indicates a satisfactory model fit, suggesting that the estimated relationships closely reflect the observed data patterns and support the reliability of the empirical specification. A key result of this study is the statistically significant and positive moderating effect of institutional quality on the relationship between technological efficiency and labor productivity. The large and highly significant coefficient of the interaction term (p-value = 0.0000) implies that institutions play a pivotal moderating role, strengthening the extent to which technological advancements translate into productivity gains. In economic terms, this finding suggests that technological efficiency alone is insufficient; rather, its effectiveness is substantially amplified in environments characterized by strong institutional frameworks.

This result aligns with the recent empirical literature, which emphasizes that the productivity-enhancing effects of technology are contingent upon institutional quality, including regulatory quality, governance effectiveness, and rule of law. Conversely, it helps explain why some prior studies report weaker or inconsistent effects of technology in developing and emerging contexts, where institutional deficiencies may constrain the diffusion and efficient use of innovation (Egbeleo & Sodokin, 2025). From a comparative perspective, these findings are consistent with cross-country evidence showing that economies with stronger institutional structures tend to achieve higher returns from technological investments, while those with weaker governance systems experience limited productivity spillovers. This reinforces the view that structural and institutional heterogeneity is a key determinant of technology-driven growth outcomes.

Theoretically, the results contribute to both the Augmented Solow growth model and endogenous growth theory. Within the Augmented Solow framework, institutions can be interpreted as a critical factor that conditions the productivity of capital and technology, thereby influencing steady-state outcomes. From an endogenous growth perspective, the findings support the argument that technology-driven growth is internally sustained but dependent on the economy’s absorptive capacity, which is largely shaped by institutional quality. In conclusion, this article argues that the vast differences in income per capita go back to how societies and markets are organized. Institutions, aligned with Douglass North, are all about having proper benchmarks and standards for empowering productivity and technology. For instance, the rule of law can secure property rights, stable policy, and low cost of trade, which are critical for sustainable growth. Therefore, improving technological efficiency should be a key policy tool for improving aggregate performance. This can be done through investments in human capital (education and training), improvements in institutional quality, and the adoption of more efficient technologies and management practices.

6. Conclusions

This study examines the relationship between technological efficiency and labor productivity, with particular emphasis on the moderating role of institutional quality. The empirical analysis yields the following conclusions. The results indicate that technological efficiency, while positively associated with output per worker, does not exert a consistent effect across countries. Instead, its impact is conditional upon the quality of the institutional environment. The estimated interaction term between efficiency and institutional quality provides robust evidence of a moderating effect, as efficiency gains are significantly enhanced in countries with stronger institutional frameworks and constrained in those with weaker governance structures. In particular, the presence of negative interaction values highlights cases where technological efficiency operates under suboptimal institutional conditions, limiting its contribution to productivity.

Despite its contributions, this study is subject to several limitations that open avenues for future research. First, the use of aggregated institutional indicators may obscure the distinct effects of specific institutional dimensions (e.g., control of corruption, rule of law, government effectiveness), suggesting the need for more disaggregated analysis. Second, potential issues related to endogeneity, such as reverse causality between productivity and institutional quality, are not fully addressed and warrant the application of more advanced estimation techniques (e.g., instrumental variables or dynamic panel models).

Overall, this study reinforces the conclusion that technological efficiency alone is insufficient to drive sustainable productivity growth. Rather, its effectiveness depends critically on the institutional context, underscoring the need for policies that strengthen the regulatory and governance environment as a foundation for long-term economic performance.

7. Recommendations

The empirical results demonstrate that the impact of technological efficiency is significantly moderated by institutional quality. As a result, policy priorities should concentrate on bolstering regulatory frameworks, especially by renovating investment laws, improving legal safeguards, and enforcing quality standards. These steps guarantee that increases in efficiency are successfully converted into increases in productivity. Policy efforts aimed at enhancing education and skills, improving governance quality, and fostering the adoption of advanced technologies are essential to achieving sustained productivity growth. This integrated approach ensures that technological progress is effectively translated into broader economic performance. Similarly, by increasing the economy’s capacity and efficiency, expenditures in infrastructure and technological education support long-term productivity growth. Complementary policies such as technology education and SME support are still crucial, but their success ultimately relies on the existence of a strong and efficient institutional framework.