Beyond GDP: COVID-19’s Effects on Macroeconomic Efficiency and Productivity Dynamics in OECD Countries

Abstract

The COVID-19 pandemic triggered unprecedented economic disruptions, raising critical questions about the resilience and adaptability of macroeconomic productivity across countries. This study examines the impact of COVID-19 on macroeconomic efficiency and productivity dynamics in 37 OECD countries using quarterly data from 2018Q1 to 2024Q4. By employing a Slack-Based Measure Data Envelopment Analysis (SBM-DEA) and the Malmquist Productivity Index (MPI), we decompose total factor productivity (TFP) into efficiency change (EC) and technological change (TC) across three periods: pre-pandemic, during-pandemic, and post-pandemic. Our framework incorporates both desirable (GDP) and undesirable outputs (inflation, unemployment, housing price inflation, and interest rate distortions), offering a multidimensional view of macroeconomic efficiency. Results show broad but uneven productivity gains, with technological progress proving more resilient than efficiency during the pandemic. Post-COVID recovery trajectories diverged, reflecting differences in structural adaptability and innovation capacity. Regression analysis reveals that stringent lockdowns in 2020 were associated with lower productivity in 2023–2024, while more adaptive policies in 2021 supported long-term technological gains. These findings highlight the importance of aligning crisis response with forward-looking economic strategies and demonstrate the value of DEA-based methods for evaluating macroeconomic performance beyond GDP.

1. Introduction

The COVID-19 pandemic, caused by the SARS-CoV-2 virus, began in December 2019 and rapidly escalated into a global health and economic crisis. Declared a pandemic by the WHO in March 2020, it resulted in over 777 million infections and more than seven million confirmed deaths by April 2025 (WHO, 2025). Distinct from conventional recessions rooted in financial or cyclical factors, COVID-19 presented as an acute exogenous health shock. This compelled governments worldwide to implement unprecedented containment measures—including widespread lockdowns, border closures, and mass vaccination campaigns. While vital for public health, these interventions inevitably triggered severe and multifaceted disruptions across the global economy, fracturing supply chains, creating unprecedented labor market volatility (including significant job losses and a rapid shift to remote work), and abruptly altering consumption patterns.

These profound disruptions and ensuing sectoral shocks exposed deep structural vulnerabilities within national economies and raised urgent questions about their resilience. Understanding the nuanced impact on macroeconomic efficiency, beyond headline GDP figures, is therefore crucial. To address this, our study introduces a novel application of the Data Envelopment Analysis (DEA)-based Malmquist Productivity Index (MPI). We utilize this framework to evaluate macroeconomic efficiency across 37 OECD countries, comprehensively assessing the COVID-19 crisis by incorporating not only desirable outputs (GDP) but also key undesirable macroeconomic outcomes—such as inflation, unemployment, housing price inflation, and interest rate distortions—across distinct pre-, during-, and post-pandemic phases. To the best of our knowledge, this represents the first application of DEA-MPI in this manner for a comparative analysis of OECD countries during this period.

The COVID-19 pandemic triggered an economic downturn that the World Bank (2020) described as one of the worst in over a century, surpassed only by the World Wars and the Great Depression. In 2020, global GDP contracted by 3.4%, while OECD countries experienced an even sharper average decline of 4.1% (OECD, 2021). The United Nations estimated that the global economy lost $8.5 trillion in output over the following two years.

The collapse was driven by severe supply chain disruptions, logistical delays, and escalating production costs (Ivanov & Dolgui, 2021; Pujawan & Bah, 2022). Labor markets were upended by job losses, rising unemployment, and a sudden shift to remote work—compounded by illness, caregiving demands, and safety concerns (OECD, 2021).

Impacts varied by sector. Tourism and other contact-intensive industries were among the hardest hit (Gössling et al., 2020), while digital services and technology sectors proved more resilient. Demand shocks caused shortages in key goods, such as semiconductors (del Rio-Chanona et al., 2020). Importantly, the crisis exacerbated existing inequalities, disproportionately harming women (Alon et al., 2020), low-income workers (Adams-Prassl et al., 2020) and widening the gap between developed and developing economies (World Bank, 2020).

In response, governments in OECD countries deployed massive fiscal and monetary stimulus measures. These included wage subsidies, direct transfers, tax relief, interest rate cuts, and large-scale quantitative easing (IMF, 2021). While these interventions provided crucial short-term relief and cushioned the immediate blow, they also contributed to rising public debt and posed longer-term challenges for economic recovery amidst increasing uncertainty regarding productivity developments, later compounded by energy crises and geopolitical tensions.

Beyond the immediate and visible impacts on Gross Domestic Product (GDP) and employment, the COVID-19 shock raises critical questions about its deeper effects on the macroeconomic efficiency of advanced economies. Understanding these effects is crucial, particularly given pre-existing concerns about a slowdown in productivity growth across many OECD countries prior to the pandemic. This paper argues that a comprehensive assessment requires moving beyond headline GDP figures to evaluate macroeconomic efficiency more holistically. This involves considering not only the desirable output (economic growth) but also undesirable macroeconomic outcomes—such as inflation, unemployment, housing market instability, and potentially distortionary interest rate environments—that significantly impact societal welfare. The concept aligns with broader calls to develop measures of economic performance that look “beyond GDP” to encompass environmental and social dimensions (Bloom et al., 2025).

Existing literature has begun to explore various facets of the pandemic’s economic impact. Studies have examined general productivity trends and the heightened uncertainty faced by firms. 1 Research using near-real-time data has investigated labor reallocation dynamics, finding evidence that while overall labor turnover fell, reallocation often remained productivity-enhancing, favoring higher-productivity and tech-savvy firms, although the strength of this link varied depending on policy design like job retention schemes (Dorville et al., 2025). Several analyses focused on Total Factor Productivity (TFP), often reveal complex dynamics: significant reductions in ‘within-firm’ productivity due to disruptions and higher costs were sometimes offset by positive ‘between-firm’ reallocation effects, as activity contracted more sharply in less productive sectors and firms (Andrews et al., 2021). Furthermore, a growing body of work has assessed the economic consequences of government response measures, particularly the impact of lockdowns or stringency measures on economic activity, often finding negative correlations with GDP growth but also highlighting the role of context, such as institutional quality or income inequality (Bloom et al., 2025).

Despite these valuable contributions, a significant research gap remains. There is a need for a comprehensive, comparative assessment of macroeconomic efficiency, explicitly incorporating undesirable macroeconomic outcomes, across the OECD countries. Furthermore, applying frontier analysis methods like Data Envelopment Analysis (DEA) and the Malmquist Productivity Index (MPI) allows for the decomposition of performance changes into efficiency catch-up and technological frontier shifts across distinct phases of the pandemic (pre-, during, and post-). To the best of our knowledge, this study is the first to employ a DEA-MPI methodology to comprehensively assess national macroeconomic efficiency across OECD countries during the COVID-19 crisis, explicitly accounting for undesirable outputs. Furthermore, it provides a novel examination of the contentious role of government stringency in shaping macroeconomic performance.

This study provides a systematic assessment of macroeconomic efficiency and productivity dynamics in 37 OECD countries across three distinct phases: pre-pandemic (2018Q1–2020Q1), during-pandemic (2020Q1–2022Q1), and post-pandemic (2022Q1–2024Q4). By employing a Slack-Based Measure Data Envelopment Analysis (SBM-DEA) and the Malmquist Productivity Index (MPI), we decompose total factor productivity (TFP) into efficiency change (EC) and technological change (TC). Our analysis is guided by three research questions:

• How did macroeconomic efficiency and productivity evolve before, during, and after the pandemic?
• What role did technological change versus efficiency change play in shaping productivity dynamics?
• How did government stringency measures influence post-pandemic productivity outcomes?

By addressing these questions, we aim to contribute to ongoing debates about the resilience of productivity and the policy measures that can support sustainable recovery in the post-pandemic era.

This paper makes several contributions to literature. Firstly, it provides a novel, comparative assessment of macroeconomic efficiency for 37 OECD countries navigating the unique economic shock of the COVID-19 pandemic. Secondly, it employs a sophisticated SBM-DEA model that explicitly incorporates key undesirable macroeconomic indicators (inflation, unemployment, housing prices, interest rates), offering a broader and arguably more policy-relevant perspective on national performance than traditional productivity studies focusing solely on GDP. The choice of SBM-DEA allows for a nuanced definition of macroeconomic performance, reflecting the trade-offs governments faced between stimulating growth and maintaining stability. Thirdly, it utilizes the MPI to disentangle the sources of TFP change (efficiency vs. technology) across distinct pre-, during-, and post-COVID periods, revealing the underlying dynamics of adaptation and recovery. Understanding whether productivity changes stem from catching up to best practices or from shifts in the frontier itself has distinct policy implications. Finally, it contributes empirical evidence to the ongoing debate on the relationship between the stringency of government pandemic responses and macroeconomic performance, attempting to quantify this link within the OECD context. The analysis acknowledges the complexity of this relationship, potentially influenced by factors like institutional quality and the non-linear effects of interventions.

The remainder of this paper is organized as follows: Section 2 presents a review of the relevant literature, focusing on the evolution of DEA-MPI applications in macroeconomic efficiency, environmental performance, and crisis response, including recent contributions related to the COVID-19 pandemic. Section 3 details the methodological framework, outlining the Slack-Based Measure Data Envelopment Analysis (SBM-DEA), the Malmquist Productivity Index (MPI), and the strategy for analyzing the influence of government stringency. Section 4 describes the data sources, variable definitions, sample countries, and time periods used in the analysis. Section 5 presents empirical results, including efficiency scores, MPI decomposition, and the second-stage regression findings. Section 6 concludes with a summary of key insights, discusses their implications for research and policy, acknowledges limitations, and proposes directions for future work.

2. Literature Review

The application of Data Envelopment Analysis (DEA) and the Malmquist Productivity Index (MPI) to assess macroeconomic efficiency and productivity has developed into a robust and interdisciplinary field of research. Initially used to evaluate productivity based on conventional inputs and outputs—such as labor, capital, and GDP—DEA-MPI frameworks have since evolved to incorporate more nuanced economic, environmental, and institutional considerations. Over time, the methodological scope has expanded from simple radial models to more sophisticated slack-based and directional distance functions, allowing researchers to better account for inefficiencies, technological changes, and undesirable outputs. This evolution has enabled the application of DEA-MPI across diverse domains, including energy, sustainability, and health systems, and more recently, to evaluate national responses to economic shocks such as the COVID-19 pandemic. In what follows, we review the progression of this literature, highlighting key empirical contributions and methodological advances that inform the present study’s approach to evaluating macroeconomic performance during and after the pandemic.

The application of DEA-MPI to assess macroeconomic performance and productivity has a rich history, evolving significantly over the past few decades. Foundational studies in the 1990s and early 2000s, such as those by Färe et al. (1994), Brockett et al. (1999), Kaüger et al. (2000), Forstner and Isaksson (2002), and Deliktas and Balcilar (2005), established the utility of these non-parametric techniques for evaluating total factor productivity (TFP) and overall macroeconomic efficiency across various groups of countries, including many OECD nations. These initial analyses typically centered on traditional inputs like capital and labor to explain GDP output, laying the groundwork for subsequent, more nuanced investigations.

A notable evolution in the literature, beginning around the mid-2000s, was the systematic incorporation of energy and environmental considerations into efficiency assessments. Researchers began to include energy consumption as a critical input and, importantly, recognized the need to account for undesirable outputs, with CO₂ emissions being a primary focus. Works by Hu and Wang (2006), Hu and Kao (2007), Chien and Hu (2007), and Zhou and Ang (2008) were instrumental in developing measures of energy efficiency. This line of inquiry rapidly expanded to encompass broader concepts of “eco-efficiency” and “environmental efficiency”. For instance, Zhou et al. (2010) focused on total-factor emission efficiency, Chang and Hu (2010) examined energy productivity, and Yeh et al. (2010) broadened the scope by including multiple pollutants like SO₂ alongside CO₂. This trend underscores a growing recognition that sustainable economic performance requires a holistic view that integrates environmental impacts.

Methodologically, this period saw the increasing adoption of more sophisticated DEA models capable of robustly handling undesirable outputs. The Slack-Based Measure (SBM) of efficiency, in particular, gained prominence. Studies by Choi et al. (2012), Apergis et al. (2015), Iftikhar et al. (2016), Park et al. (2018), Wang et al. (2020), Demiral and Sağlam (2021, 2023), and Wang et al. (2023) effectively utilized SBM or similar DEA variants, such as the Directional Distance Function (DDF). These approaches allowed for the explicit modeling of CO₂ emissions or other environmental “bads” as undesirable outputs, offering a more accurate and policy-relevant assessment of efficiency by appropriately accounting for environmentally detrimental activities alongside desirable economic outputs like GDP.

The geographical focus of these studies has varied, with many analyses concentrating on specific regions or economic blocks, which is valuable for targeted policy insights. A significant body of work, including several of the aforementioned studies, has specifically examined OECD countries, making this collective research highly pertinent for understanding efficiency dynamics within this group of advanced economies. Concurrently, other research has provided insights into regions such as China’s provinces, various European nations, APEC countries, and Latin American countries, reflecting diverse policy priorities and data landscapes.

The range of variables considered in these DEA and MPI studies has also expanded considerably. While capital and labor remain fundamental inputs, analyses now frequently incorporate different forms of energy consumption (e.g., total energy, specific fuel types, renewable energy). On the output side, the focus has broadened beyond solely GDP to include desirable outputs like renewable electricity generation and, crucially, a variety of undesirable outputs, most commonly CO₂ emissions but also other greenhouse gases. This evolution in variable selection reflects a more comprehensive understanding of the multifaceted nature of economic performance and sustainability. Consequently, the types of “efficiency” being measured have diversified, encompassing not only TFP and general productivity but also more specialized concepts such as technical efficiency, energy efficiency, emission efficiency, environmental efficiency, and overall eco-efficiency, highlighting the adaptability of DEA-based methods to address a wide array of research questions.

The COVID-19 pandemic constituted an unprecedented global disruption with far-reaching consequences for productivity. For example, Bloom et al. (2025) examined the potential long-term productivity impacts of remote work, while other studies highlight initial declines due to disruptions and sector-specific productivity shifts during lockdown periods. As the crisis unfolded, a growing number of studies applied Data Envelopment Analysis (DEA) to assess the efficiency of pandemic responses, especially across OECD nations. DEA has been used to evaluate overall management performance (Doğan et al., 2021), healthcare system efficiency (Ersoy & Aktaş, 2022; Selamzade et al., 2023), trade-offs between contagion control and treatment capacity (Martins & Sousa, 2024; Yu et al., 2024), and the efficacy of fiscal interventions (Husseiny & Badawy, 2022). Methodological innovations include network DEA for capturing process interdependencies (Pereira et al., 2022; Azadi et al., 2023), integration with multi-criteria decision-making (Selamzade et al., 2023), and machine learning techniques (Taherinezhad & Alinezhad, 2023). Other studies incorporate broader dimensions such as economic resilience (Klumpp et al., 2022) and cultural influences on efficiency (Min et al., 2022).

Recent studies have examined the long-term effects of the COVID-19 pandemic on productivity using diverse approaches. For instance, Baqaee and Farhi (2022) employ a disaggregated Keynesian framework to show how sectoral supply and demand shocks affected productivity. Bloom et al. (2021) find that U.S. patent filings for remote-work technologies surged during the pandemic, reflecting an accelerated shift toward digital infrastructure. Similarly, Andrews et al. (2021) analyze firm-level data for Australia, New Zealand, and the UK, finding that productivity-enhancing job reallocation persisted despite pandemic-induced disruptions. Most recently, Calligaris et al. (2023) highlight the role of job retention schemes in maintaining productive employment across 12 countries, while ensuring that structural reallocation continued during the crisis. These studies complement our DEA–MPI approach by offering alternative evidence of pandemic-driven technological and efficiency dynamics across countries.

Together, this body of work highlights the versatility of DEA in analyzing multi-faceted crisis responses. While focused on pandemic-specific outcomes, these studies provide essential context for understanding how national productivity and sustainability evolved during a period when policy effectiveness critically influenced both economic activity and resource use.

Building upon this evolving literature, the present study seeks to make distinct contributions. While previous research has significantly advanced the use of DEA-MPI for environmental and specific crisis-response efficiencies, a comprehensive assessment of macroeconomic efficiency across OECD countries—explicitly incorporating a range of undesirable macroeconomic indicators like inflation, unemployment, housing price inflation, and interest rate distortions within an SBM-DEA-MPI framework—through the distinct pre-, during-, and post-COVID phases, remains a developing area. By segmenting the analysis across these periods and subsequently examining the role of government pandemic stringency, this research aims to provide novel insights into national productivity dynamics during this unique global shock, leveraging the robust SBM methodology for a more nuanced understanding of the trade-offs involved in achieving macroeconomic stability and growth.

3. Methodology

This section outlines the quantitative methods employed to assess macroeconomic efficiency and productivity dynamics across the 37 OECD countries during the COVID-19 pandemic. We utilize a two-stage approach: first, applying non-oriented Slack-Based Measure Data Envelopment Analysis (NO-SBM-DEA) with undesirable outputs to measure efficiency, followed by the Malmquist Productivity Index (MPI) to analyze productivity changes and their components. Finally, a second-stage regression analysis is proposed to investigate the influence of government stringency measures.

3.1. Measuring Macroeconomic Efficiency: NO-SBM-DEA with Undesirable Outputs

Farrell (1957) first introduced the concept of measuring productive efficiency, which was later operationalized by Charnes et al. (1978) through the development of the foundational Data Envelopment Analysis (DEA) framework. DEA is a non-parametric, linear programming method used to evaluate the relative efficiency of decision-making units (DMUs) based on observed input and output data, without requiring assumptions about the functional form of the production function. Building on these foundational concepts, DEA has been extensively studied and applied over the last four decades. Recently, Camanho et al. (2024) reviewed the evolution of Data Envelopment Analysis (DEA) in assessing economic efficiency from 1978 to 2020, highlighting key methodological developments, empirical applications across sectors, and trends.

Traditional DEA models, such as CCR (Charnes et al., 1978) and BCC (Banker et al., 1984), primarily focus on radial efficiency and face limitations when dealing with non-zero slacks (input excesses or output shortfalls) and, crucially, undesirable outputs like pollution or negative macroeconomic indicators. To overcome these limitations, this study employs the Slack-Based Measure (SBM) of efficiency, developed by Tone (2001, 2010). The SBM model is non-radial and directly incorporates slacks into the efficiency measurement, providing a more comprehensive and accurate assessment of inefficiency. Its key advantage lies in its ability to handle inputs, desirable outputs, and undesirable outputs simultaneously within a unified framework. This makes NO-SBM-DEA particularly well-suited for evaluating macroeconomic efficiency, where the goal is not simply to maximize GDP (desirable output) but to do so while minimizing necessary inputs (e.g., Population, GFCF, Employment) and minimizing undesirable macroeconomic outcomes (e.g., CPI inflation, Unemployment Rate, excessive House Price Index changes, potentially unstable Short-Term Interest Rates). This approach avoids the need for arbitrary transformations or re-categorization of undesirable outputs as inputs, which can distort the underlying production technology and lead to misleading results.

Following Tone (2001, 2010) and the formulation adapted from studies like, we consider $n$ DMUs (OECD countries in this study). Each DMU $k (k=1,...,n$) uses $m$ inputs $x_{ik} (i=1,...,m)$ to produce $q_1$ desirable outputs $y_{r_{1}k}^g (r_1=1,...,q_1)$ and $q_2$ undesirable outputs $y_{r_{2}k}^g$ ($r_2=1,...,q_2$). The production possibility set under Variable Returns to Scale (VRS) is defined by the observed data. The SBM efficiency score $\left(\rho \right)$ for a specific DMU $k$ (where $k\in \{1,...,n\}$) is obtained by solving the following linear programming problem.

$$\begin{array}{l}{\rho }_k=Min. {\displaystyle \frac{1-\frac{1}{m}{\sum }_{i=1}^m\frac{s_i^{-}}{x_{ik}}}{1+\frac{1}{q_1+q_2}\left(\sum _{r_1}^{q_1}\frac{s_{r_1}^g}{y_{r_{1}k}^g}+\sum _{r_2}^{q_2}\frac{s_{r_2}^b}{y_{r_{2}k}^b}\right)}}\\ {s.t. x}_{ik}={\displaystyle \sum _{j=1}^n}{x}_{ij}{\lambda }_j+s_i^{-} i=1,2,\dots ,m;\\ y_{r_{1}k}^g={\displaystyle \sum _{j=1}^n}{y}_{r_{1}j}^g{\lambda }_j-s_{r_1}^g r_1=1,2,\dots ,q_1;\\ y_{r_{2}k}^b={\displaystyle \sum _{j=1}^n}{y}_{r_{2}j}^b{\lambda }_j+s_{r_2}^b r_2=1,2,\dots ,q_2;\\ {\sum }_{j=1}^n{\lambda }_j=1 j=1,2,\dots ,n;\\ {\lambda }_j,{s_i^{-}, s}_{r_1}^g,{ s}_{r_2}^b\ge 0; j\in R^n, i\in R^m, r_1\in R^{q_1}, r_2\in R^{q_2}\end{array}$$

(1)

where:

• ${\rho }_k$ is the SBM efficiency score for DMU $k$, ranging from 0 to 1. A score of ${\rho }_k=1$ indicates that DMU $k$ is fully efficient on the frontier, implying all slack variables (${s_i^{-}, s}_{r_1}^g,{ s}_{r_2}^b$) are zero.
• $x_{ik}, y_{r_{1}k}^g,$ and $y_{r_{2}k}^b$ are the observed inputs, desirable outputs, and undesirable outputs of DMU $k$.
• The constraint $\sum {\lambda }_j=1$ imposes the VRS assumption, appropriate for comparing national economies of potentially different scales.
• ${s_i^{-}, s}_{r_1}^g,{ s}_{r_2}^b$ are the slack variables representing input excess, desirable output shortfall, and undesirable output excess, respectively.
• ${\lambda }_j$ are the intensity vectors representing the contribution of DMU j to the frontier.

This formulation captures the objective of achieving high desirable output while minimizing both inputs and undesirable outputs. The non-oriented nature of the SBM model allows for simultaneous adjustments in inputs and outputs to reach the efficiency frontier.

3.2. Measuring Productivity Dynamics: Malmquist Productivity Index (MPI)

To evaluate changes in productivity over time, particularly shifts in Total Factor Productivity (TFP), this study utilizes the Malmquist Productivity Index (MPI). Originally conceptualized by Caves et al. (1982) building on Malmquist’s (1953) distance function concept and later adapted for DEA by Färe et al. (1994), the MPI assesses the productivity variation of a DMU between two periods ($t$ and $t+1$). This is achieved by computing the geometric mean of two distance function ratios, which are evaluated against the technological frontiers of both periods.

In this study, the Malmquist Productivity Index (MPI) is constructed using efficiency scores derived from a Slack-Based Measure (SBM) DEA model. Let $D^t\left(x_k,y_k^g,y_k^b\right)$ denote the efficiency score of ${DMU}_k$ when evaluated against the production frontier in period $t$, based on its inputs $x_k$, desirable outputs $y_k^g$, and undesirable outputs $y_k^b$. Similarly, $D^{t+1}\left(x_k,y_k^g,y_k^b\right)$ represents the efficiency of the same DMU relative to the frontier in period $t+1$.

To calculate MPI, cross-period (or “mixed”) efficiency scores are also required. Specifically, $D^t(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})$ evaluates period $t+1$ data against the period $t$ frontier, while $D^{t+1}(x_k^t,y_k^{g,t},y_k^{b,t})$ assesses period $t$ data against the period $t+1$ frontier. These four distance functions are then used to compute the MPI and its two components: EC, which captures movement relative to the existing frontier, and TC, which reflects shifts in the frontier over time.

Therefore, the $MPI$ between period $t$ and $t+1$ for ${DMU}_k$ is defined as follows:

$${MPI}_k^{t,t+1}{=\left[{\displaystyle \frac{D^t(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})}{D^t(x_k^t,y_k^{g,t},y_k^{b,t})}}\times {\displaystyle \frac{D^{t+1}(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})}{D^{t+1}(x_k^t,y_k^{g,t},y_k^{b,t})}}\right]}^{1/2}$$

(2)

An $MPI$ value greater than 1 indicates productivity improvement between periods t and t + 1, while a value less than 1 reflects a decline.

This index can be decomposed as: ${MPI}_k^{t,t+1}={EC}_k^{t,t+1}\times {TC}_k^{t,t+1}$ where:

$${EC}_k^{t,t+1}={\displaystyle \frac{D^{t+1}(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})}{D^t(x_k^t,y_k^{g,t},y_k^{b,t})}}$$

(3)

and

$${TC}_k^{t,t+1}={\left[{\displaystyle \frac{D^t(x_k^t,y_k^{g,t},y_k^{b,t})}{D^{t+1}(x_k^t,y_k^{g,t},y_k^{b,t})}}\times {\displaystyle \frac{D^t(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})}{D^{t+1}(x_k^{t+1},y_k^{g,t+1},y_k^{b,t+1})}}\right]}^{1/2}$$

(4)

Decomposing $MPI$ into $EC$ and $TC$ clarifies whether changes in total factor productivity (TFP) result from improved use of existing resources or from shifts in the production frontier. $EC$ measures how DMU’s technical efficiency evolves between periods t and t + 1. An $EC > 1$ indicates improved efficiency (closer to the frontier), $EC < 1$ reflects a decline, and $EC = 1$ implies no change. Similarly, $TC$ captures changes in the production frontier due to innovation or technological progress. A $TC > 1$ signals technological advancement, $TC < 1$ indicates regression, and $TC = 1$ denotes stability.

3.3. Analyzing the Impact of Government Stringency

To address the third research question regarding the influence of government pandemic responses on productivity, we propose a second-stage regression analysis. This approach is common in DEA literature, where efficiency or productivity scores from the first stage are regressed on environmental or contextual variables.

In this analysis, the dependent variables were the cumulative Malmquist Productivity Index (MPI), cumulative efficiency change (catch-up effect), and cumulative technological change (frontier shift), each calculated for the one-year period spanning 2023Q1 to 2024Q1 for each country (i). The primary independent variables of interest were measures of government stringency during the preceding crisis years (2020–2022), derived from the Oxford COVID-19 Government Response Tracker (OxCGRT) Stringency Index (Our World in Data, 2024). This index compiles information on various containment policies, such as school and workplace closures and travel restrictions, into a composite score reflecting the overall strictness of measures. We implemented two main regression models: one using annual average stringency scores for 2020, 2021, and 2022 as distinct independent variables, and another using quarterly average stringency scores for the same period as time dummies.

The general form of the regression model used is specified as:

$$\begin{gathered} {Productivity Change}_{i, 2023Q1-2024Q1 } \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}={\beta }_0+{\beta }_1{Stringency}_i,period+{\epsilon }_i \end{gathered}$$

(5)

where:

• ${Productivity Change}_{i, 2023Q1-2024Q1 }$ represents the cumulative MPI, cumulative Efficiency Change, or cumulative Technological Change for country $i$ over the 2023Q1–2024Q1 period (depending on the specific regression).
• ${Stringency}_{i, period }$ is the measure of government stringency for country $i$. In Model I, this term is replaced by separate independent variables representing the average annual stringency in 2020, 2021, and 2022. In Model II, this is represented by a series of dummy variables indicating the average quarterly stringency for each quarter from 2020Q1 to 2022Q4.
• ${\beta }_0$ is the intercept, ${\beta }_1$ represents the coefficients of interest, estimating the impact of government stringency during the specified period on the cumulative productivity change in 2023Q1–2024Q1, and ${\epsilon }_i$ is the error term.

A significant methodological consideration is the potential endogeneity of government stringency measures. Stricter policies were often enacted in response to more severe outbreaks, and the severity of the pandemic itself directly impacted economic activity and productivity. This potential for reverse causality could bias estimates of stringency’s effect. While the study design, focusing on the impact of stringency from 2020–2022 on productivity in a later period (2023–2024), helps to examine lagged effects, this limitation should be acknowledged when interpreting the results. Additionally, the analysis considered the possibility of non-linear relationships between stringency levels and productivity impacts.

Crucially, the primary source for the Stringency Index, OxCGRT, ceased providing real-time updates at the end of 2022. This data constraint necessitated our analytical approach, focusing on the impact of stringency data available up to 2022 on productivity outcomes in the subsequent 2023–2024 period. This design allowed us to investigate the delayed consequences of pandemic-era policies despite the data availability limitation for the later period.

4. Data

This section presents the dataset used to evaluate macroeconomic efficiency and productivity dynamics for 37 OECD countries over the period 2018Q1 to 2024Q4. The data are structured quarterly to enable consistent comparisons across three distinct phases of the COVID-19 pandemic: pre-COVID (2018Q1–2020Q1), during-COVID (2020Q1–2022Q1), and post-COVID (2022Q1–2024Q4). All variables are sourced from the OECD Data Explorer (OECD, 2025) to ensure standardization and cross-country comparability. The sample includes all OECD member countries except Türkiye, which is excluded due to missing quarterly data on population and short-term interest rates. Additionally, while all other countries have seasonally adjusted figures for gross fixed capital formation (GFCF), Türkiye’s GFCF data is available only in non-seasonally adjusted form. The subsequent subsections describe the input and output variables used in the analysis, along with their definitions and relevance to the efficiency modeling framework.

Table A1. Quarterly Averages of Macroeconomic Indicators (2018–2024).

Year	Quarter	Population1	GFCF	Employment	GDP	Consumer Price Index2	Unemployment	House Price Index3	Short-Term Interest Rate4
2018	Quarter 1	34,288.61	79,034.11	16,048.39	1,576,646.84	103.77	906.11	111.75	1.70
2018	Quarter 2	34,338.66	78,988.35	16,136.32	1,596,324.61	104.70	889.52	112.61	1.72
2018	Quarter 3	34,393.18	78,127.25	16,167.64	1,611,335.79	105.14	872.20	113.13	1.74
2018	Quarter 4	34,447.21	78,568.07	16,211.96	1,627,762.60	105.55	876.85	114.27	1.80
2019	Quarter 1	34,493.91	79,096.43	16,275.16	1,649,237.32	105.63	877.40	115.21	1.84
2019	Quarter 2	34,542.05	80,937.65	16,347.66	1,675,689.36	106.75	853.48	115.86	1.80
2019	Quarter 3	34,565.67	80,266.79	16,375.60	1,699,761.89	106.90	849.61	117.10	1.67
2019	Quarter 4	34,641.41	82,048.22	16,443.67	1,715,390.58	107.25	847.24	118.29	1.62
2020	Quarter 1	34,660.81	80,560.05	16,378.32	1,702,561.03	107.49	870.79	119.97	1.56
2020	Quarter 2	34,684.33	72,134.54	15,142.24	1,530,888.04	107.45	1367.71	120.41	1.27
2020	Quarter 3	34,707.91	79,612.57	15,663.77	1,684,013.95	107.80	1253.70	121.85	1.05
2020	Quarter 4	34,729.36	83,873.17	15,930.22	1,711,125.09	107.99	1131.76	124.44	0.96
2021	Quarter 1	34,738.08	85,855.36	15,904.02	1,740,155.79	108.80	1095.79	127.33	0.93
2021	Quarter 2	34,767.94	87,965.93	16,071.46	1,792,894.14	110.14	1060.84	129.80	0.93
2021	Quarter 3	34,801.18	88,028.18	16,236.31	1,836,878.64	111.31	965.18	132.42	1.01
2021	Quarter 4	34,842.38	89,361.99	16,354.18	1,900,407.88	113.23	895.78	134.55	1.28
2022	Quarter 1	34,898.09	90,946.69	16,510.10	1,951,090.11	115.93	849.50	135.22	1.68
2022	Quarter 2	34,972.31	90,418.41	16,630.60	2,009,591.63	120.03	820.60	134.93	2.29
2022	Quarter 3	35,031.82	89,414.68	16,655.19	2,044,101.15	123.06	813.09	133.57	3.48
2022	Quarter 4	35,092.59	90,131.96	16,704.46	2,059,488.18	125.26	812.99	130.56	4.66
2023	Quarter 1	35,154.26	94,421.64	16,833.70	2,080,094.16	127.17	807.14	128.44	5.31
2023	Quarter 2	35,208.98	95,952.73	16,890.10	2,090,217.15	128.86	807.64	128.00	5.76
2023	Quarter 3	35,270.71	96,493.16	16,930.03	2,111,493.60	129.72	816.07	127.66	5.93
2023	Quarter 4	35,334.05	97,042.64	16,950.64	2,124,346.85	130.21	827.33	128.25	5.86
2024	Quarter 1	35,384.06	98,076.29	16,989.04	2,149,108.21	131.17	838.64	128.90	5.67
2024	Quarter 2	35,438.94	97,598.43	17,018.02	2,173,948.11	132.51	842.49	129.84	5.49
2024	Quarter 3	35,488.58	100,057.74	17,053.15	2,195,558.98	133.11	846.03	130.88	5.24
2024	Quarter 4	35,529.84	99,291.72	17,062.51	2,218,422.12	130.90	840.23	132.21	4.79

This table presents quarterly average values for key macroeconomic indicators from 2018 Q1 to 2024 Q4. Variables include Population (in thousands), Gross Fixed Capital Formation (GFCF, in millions), Employment (in thousands), Gross Domestic Product (GDP, in millions), Consumer Price Index (CPI, base year unspecified), Unemployment (in thousands), House Price Index (base year unspecified), and Short-Term Interest Rate (percent). The dataset covers both pre- and post-COVID periods, allowing for longitudinal analysis of economic recovery and trends.

and

Table A2. Descriptive Statistics of Macroeconomic Indicators by Year (2018–2024).

Year	Variable	Population5	GFCF	Employment	GDP	Consumer Price Index6	Unemployment	House Price Index7	Short-Term Interest Rate8
2018	Mean	34,367	78,679	16,141	1,603,017	104.79	886	112.94	1.74
2018	Stan. Dev.	59,451	185,148	28,260	3,477,137	3.24	1279	9.43	2.02
2018	Minimum	354	1440	198	20,177	100.52	6	95.54	0.04
2018	Median	10,335	26,117	4755	478,026	104.15	301	112.77	0.53
2018	Maximum	328,795	1,099,753	155,763	20,656,516	115.79	6315	137.73	8.91
2019	Mean	34,561	80,587	16,361	1,685,020	106.63	857	116.62	1.73
2019	Stan. Dev.	59,740	192,693	28,605	3,623,402	4.11	1228	11.78	2.00
2019	Minimum	362	1285	201	21,822	101.17	7	94.59	0.08
2019	Median	10,354	27,191	4832	491,175	106.07	290	116.70	0.49
2019	Maximum	330,513	1,148,680	157,537	21,539,982	119.87	5999	150.96	9.12
2020	Mean	34,696	79,045	15,779	1,657,147	107.68	1156	121.67	1.21
2020	Stan. Dev.	59,976	193,121	27,103	3,586,995	5.19	2228	13.47	1.37
2020	Minimum	367	1149	195	20,451	99.91	11	96.14	0.15
2020	Median	10,385	25,280	4802	493,960	107.65	351	122.26	0.61
2020	Maximum	331,840	1,152,467	147,813	21,354,105	122.89	12,950	154.04	6.52
2021	Mean	34,787	87,803	16,141	1,817,584	110.87	1004	131.02	1.04
2021	Stan. Dev.	60,111	210,400	27,897	3,957,897	6.54	1598	17.73	1.12
2021	Minimum	373	1474	197	22,953	101.13	13	97.17	0.11
2021	Median	10,419	29,027	4783	565,004	110.91	344	130.23	0.62
2021	Maximum	332,505	1,259,944	152,584	23,681,171	129.53	8630	169.46	5.51
2022	Mean	34,999	90,228	16,625	2,016,068	121.07	824	133.57	3.03
2022	Stan. Dev.	60,401	229,334	28,799	4,339,218	10.02	1201	20.95	2.91
2022	Minimum	383	1740	210	28,058	103.93	8	94.51	0.67
2022	Median	10,527	28,624	4940	674,001	119.50	300	130.73	1.19
2022	Maximum	334,373	1,389,702	158,295	26,006,893	140.13	5993	180.72	10.82
2023	Mean	35,242	95,978	16,901	2,101,538	128.99	815	128.09	5.72
2023	Stan. Dev.	60,812	244,098	29,274	4,616,391	13.87	1188	20.34	2.99
2023	Minimum	395	1954	220	30,158	106.34	8	89.39	0.85
2023	Median	10,578	31,256	5028	644,550	126.09	294	127.73	4.28
2023	Maximum	337,141	1,482,408	161,041	27,720,710	159.22	6077	172.18	14.49
2024	Mean	35,460	98,756	17,031	2,184,259	131.92	842	130.46	5.30
2024	Stan. Dev.	61,258	258,362	29,362	4,850,266	16.73	1266	23.30	2.13
2024	Minimum	407	2208	230	31,045	85.26	8	84.46	1.04
2024	Median	10,631	26,428	5065	676,222	130.31	306	130.81	4.42
2024	Maximum	340,212	1,573,491	161,349	29,184,890	166.93	6761	181.05	12.12

This table reports yearly descriptive statistics—mean, standard deviation, minimum, median, and maximum—for key macroeconomic indicators across the 2018–2024 period. Variables include Population (in thousands), Gross Fixed Capital Formation (GFCF, in millions), Employment (in thousands), GDP (in millions), Consumer Price Index (CPI), Unemployment (in thousands), House Price Index, and Short-Term Interest Rate (percent). These statistics provide insight into the distribution and variability of macroeconomic conditions across years and serve as a foundation for trend and volatility analyses.

in the Appendix A provide a comprehensive view of macroeconomic conditions from 2018 to 2024. Table A1 presents quarterly trends in key indicators such as population, investment, employment, inflation, and interest rates, serving as the foundation for analyzing shifts in economic performance before, during, and after the COVID-19 pandemic. Table A2 complements this by summarizing yearly descriptive statistics, highlighting the distribution, central tendencies, and variability of these indicators, and enabling identification of structural changes and volatility patterns associated with major economic disruptions and recovery phases.

4.1. Input Variables

To capture the macroeconomic production capacity of each country, we include three key input variables: population, gross fixed capital formation (GFCF), and employment. These variables reflect demographic scale, capital investment, and labor utilization—core components of economic output in macro-level production functions. All data are reported on a quarterly basis and sourced from the OECD Data Explorer. Where available, seasonally adjusted values are used to ensure comparability across time and countries.

Population ($x_1$): The population dataset covers quarterly figures from 2018-Q1 to 2024-Q4 for 37 OECD countries (excluding Türkiye due to data limitations), capturing demographic shifts over the pre-, during-, and post-COVID periods. Overall, the data reveal steady and moderate population growth across most countries, despite the disruptions caused by the pandemic. For example, the United States saw its population rise from 328.1 million in 2018-Q1 to approximately 341.2 million by 2024-Q4, reflecting an average annual growth rate of about 0.6%. European countries such as Germany, France, and the United Kingdom also experienced incremental increases, with Germany growing from 82.8 million to 84.7 million, and the UK from 66.2 million to 69.5 million over the same period. Smaller economies like Luxembourg and Iceland exhibited more rapid relative growth—Luxembourg’s population rose from 604,000 to over 682,000, averaging more than 2% annually—likely due to higher migration inflows. These trends are particularly relevant in the context of macroeconomic efficiency analysis, as population serves not only as a demographic indicator but also as a proxy for labor supply and scale. The variations in growth rates and density trends across countries are essential for interpreting differences in input utilization and contextualizing national performance under the SBM-DEA framework.

GFCF ($x_2$): In general, most OECD countries experienced a rebound in capital formation following the initial pandemic shock in 2020. Countries such as Ireland, Lithuania, and Slovenia recorded some of the highest GFCF growth rates over the seven-year period, with Ireland notably increasing from approximately 16.2 billion USD in 2018-Q1 to over 43 billion USD by 2024-Q4—nearly tripling its capital formation. This surge reflects Ireland’s sustained investment-driven growth, possibly fueled by multinational corporate activity and robust tech and pharmaceutical sectors.

On the other hand, countries like Japan, Mexico, and Norway showed relatively modest or stagnant growth in GFCF. For instance, Japan’s capital formation remained relatively flat, moving from about 250 billion USD in 2018-Q1 to just above 255 billion USD in 2024-Q4, indicating subdued investment dynamics possibly linked to long-term structural stagnation and an aging population.

Several countries—Greece, Portugal, and Italy—showed steady but moderate growth in GFCF, likely driven by gradual economic recovery, EU-funded infrastructure initiatives, and renewed business confidence post-pandemic. Meanwhile, the United States saw robust GFCF levels throughout the period, peaking post-COVID, consistent with its aggressive fiscal stimulus and large-scale public and private investment in infrastructure and technology.

Overall, the average GFCF levels increased over the period for nearly all countries, signaling a broader recovery in investment activity after the contraction in 2020. However, the rate and extent of recovery were uneven across countries, reflecting differences in fiscal policy responses, structural economic characteristics, and exposure to global supply chain disruptions. The disparities in GFCF trajectories also underscore the importance of policy frameworks that support resilient, long-term investment, particularly in the face of future economic shocks.

Employment ($x_3$): The employment data reveals consistent growth in total employment, though the pace and resilience of recovery vary by country, especially through the COVID-19 shock and the subsequent recovery period.

Across the OECD, most countries experienced a temporary dip or stagnation in employment during 2020, corresponding with the peak of the pandemic and widespread lockdown measures. For example, countries like Canada, the United Kingdom, and the United States saw modest declines or plateaus in employment during 2020, but subsequently showed strong rebounds, with employment surpassing pre-pandemic levels by 2022.

Smaller and service-oriented economies—such as Ireland, Iceland, and Luxembourg—demonstrated more volatile employment trends, likely reflecting their sensitivity to international trade and travel restrictions. On the other hand, countries with larger domestic markets and diversified labor structures, such as Germany, France, and Japan, exhibited steadier, if slower, employment growth throughout the pandemic period.

Eastern European members (e.g., Poland, Czechia, and Slovakia) also showed positive employment trajectories, although their growth was more gradual. Other major emerging OECD economies such as Mexico and Colombia also recorded moderate employment gains over the entire period.

Overall, average employment growth across countries ranged from 2% to 10%, reflecting a combination of demographic trends, economic policy responses, and labor market resilience. The post-2021 recovery phase is marked by increasing labor demand in most countries, with employment levels generally exceeding their 2018–2019 baselines by the end of 2024. This trend suggests a robust, albeit uneven, labor market recovery within the OECD.

4.2. Output Variables

The output variables selected for this analysis aim to provide a holistic view of macroeconomic performance, capturing both economic growth and critical sources of instability. In our DEA model, output variables are categorized into desirable outputs, which reflect positive economic performance, and undesirable outputs, which capture economic distortions or inefficiencies. While Gross Domestic Product (GDP) is included as the primary desirable output, the study also incorporates a set of undesirable outputs—namely, the Consumer Price Index (CPI), Unemployment Rate, House Price Index (HPI), and Short-Term Interest Rate Volatility. These indicators reflect inflationary pressures, labor market health, housing market volatility, and monetary policy constraints, respectively. Together, this combination allows for a more nuanced evaluation of macroeconomic efficiency, accounting for the trade-offs policymakers face between growth and economic stability.

The inclusion of short-term interest rate volatility as an undesirable output is motivated by its disruptive effects on macroeconomic efficiency. Unlike long-term policy-driven interest rates, which can be used strategically to stimulate economic activity, short-term interest rates that exhibit excessive volatility or abrupt changes can lead to market uncertainty, credit tightening, and reduced investment. Our approach does not assume that “lower interest rates are always better”, but rather that stable and predictable interest rate environments are more conducive to sustained productivity growth. Therefore, interest rate volatility is treated as a factor that detracts from overall macroeconomic efficiency, in line with the treatment of other undesirable indicators like inflation or unemployment.

The other undesirable outputs—CPI inflation, unemployment rate, and house price inflation—are standard indicators of macroeconomic imbalances that can reduce efficiency. Their inclusion ensures that the analysis goes beyond GDP growth, providing a more comprehensive assessment of economic health during the pandemic and post-pandemic recovery.

The SBM-DEA framework used in this study is a non-parametric method that does not impose predefined weights on inputs or outputs. Instead, the model endogenously determines the most favorable set of weights for each decision-making unit (DMU) to maximize its efficiency score while satisfying data constraints. This approach ensures that no single undesirable indicator is arbitrarily emphasized or downweighted, avoiding subjective bias in assessing macroeconomic performance.

It is important to note that equal treatment of undesirable outputs does not imply that they have identical economic impact. Each indicator represents a distinct dimension of inefficiency—e.g., inflation reflects price instability, unemployment captures underutilized labor resources, house price inflation signals asset market imbalances, and interest rate volatility measures financial instability. By leaving weights flexible, the DEA model allows country-specific data to determine the relative importance of each indicator based on its contribution to observed inefficiency. This non-parametric weighting approach is widely used in macro-level DEA studies to ensure an objective and data-driven efficiency evaluation across countries.

GDP ($y_1$): The GDP dataset comprises quarterly real GDP figures for 37 OECD countries spanning from Q1 2018 through Q4 2024, expressed in millions of national currency units and adjusted for seasonality. This high-resolution temporal coverage enables a detailed analysis of macroeconomic performance across three distinct phases: pre-COVID (2018–2019), during-COVID (2020–2021), and post-COVID (2022–2024). Most countries in the dataset display a general upward trend in output, reflecting ongoing economic expansion. However, this trend is interrupted sharply in early 2020 with the onset of the COVID-19 pandemic, where significant contractions are observed across nearly all countries, particularly in Q2 2020. For example, the United Kingdom, Italy, and Spain faced some of the steepest declines during this period. In contrast, the latter part of the series shows signs of recovery, although at varying rates. Some economies, such as the United States and Korea, demonstrate strong rebounds, while others, including some European nations, show slower or uneven recovery trajectories. These quarterly GDP patterns offer a critical lens through which to assess not just levels of economic output, but also the resilience and adaptability of national economies to large-scale shocks.

CPI ($y_2$): During the pre-COVID period (2018Q1–2020Q1), most OECD countries experienced stable and moderate inflation. For example, Germany and France maintained average quarterly CPI growth rates between 0.4% and 0.6%, reflecting well-managed price dynamics under relatively strong economic fundamentals. The U.S. also saw stable inflation, averaging around 0.5% per quarter during this period.

However, with the onset of the pandemic in early 2020, inflation pressures eased significantly. In second quarter of 2020, CPI growth in countries like Italy, Spain, and Japan dipped close to zero or turned slightly negative, reflecting sharp contractions in consumption and energy demand. Japan, in particular, recorded near-zero inflation throughout 2020, consistent with its historically low inflationary environment and weaker domestic demand.

Starting in 2021, inflation began to accelerate markedly across most OECD economies. The U.S. and the United Kingdom saw some of the steepest CPI increases: U.S. inflation peaked at over 2% quarterly growth in the second quarter of 2022, while the UK experienced sustained high inflation from mid-2021 through late 2023. Similarly, Eastern European countries such as Poland and the Czech Republic recorded significant price hikes, with quarterly CPI growth exceeding 2.5% in 2022, partly due to energy price shocks and currency depreciation effects.

Nordic countries like Sweden and Norway also experienced elevated inflation, though to a lesser extent, with peaks around 1.5% per quarter in 2022, driven by food and electricity prices. Meanwhile, Japan and Switzerland remained relative outliers, maintaining lower CPI increases throughout, rarely exceeding 1% per quarter, underscoring structural differences in inflationary sensitivity and policy environments.

By 2023–2024, CPI growth began to decline in most countries, with notable slowdowns in the U.S., Germany, and Canada, reflecting the effects of central bank rate hikes and improving global supply chains. Still, by the end of 2024, price levels remained significantly above their 2019 baseline, with cumulative inflation exceeding 15–20% in many countries compared to pre-pandemic levels, emphasizing inflation’s long-lasting impact on macroeconomic stability.

Unemployment ($y_3$): Prior to the COVID-19 pandemic, most countries maintained relatively stable unemployment levels. For instance, Germany and the Netherlands reported consistent unemployment around 1.3 to 1.6 million and under 400,000, respectively. However, a notable spike occurred in 2020Q2, aligning with the onset of the pandemic and the imposition of lockdown measures. The United States experienced one of the most dramatic increases, with unemployment surging from 6.4 million in 2020Q1 to over 20 million in 2020Q2. Canada, similarly, saw unemployment rise from around 1 million to nearly 2.8 million during the same period.

Spain consistently displayed the highest unemployment across the period, exceeding 3 million even before the pandemic and remaining elevated thereafter, highlighting structural labor market issues. In contrast, countries like Japan, Switzerland, and the Czech Republic exhibited relatively low and stable unemployment levels, typically below 500,000, even during the pandemic shock. From 2021 onward, most countries began to show signs of recovery, with gradual reductions in unemployment figures, though the pace varied. Some countries, such as the US and Canada, demonstrated relatively faster rebounds, while others, including Italy and Greece, experienced more persistent unemployment levels. This dataset illustrates the heterogeneous labor market responses and recovery paths across OECD economies in the wake of COVID-19.

HPI ($y_4$): The Housing Price Index (HPI) data highlights pronounced shifts in residential real estate markets across OECD countries, particularly shaped by the COVID-19 pandemic and its aftermath. In the early pre-pandemic period (2018–2019), most countries experienced modest, steady housing price increases. However, beginning in mid-2020, several countries witnessed accelerated growth, driven by historically low interest rates, increased household savings, and shifts in housing preferences due to remote work.

For example, the United States saw consistent housing price increases from 2020Q2 onward, with the index rising sharply into 2022 before gradually plateauing. Similarly, Canada experienced rapid appreciation during the pandemic, especially in suburban and semi-urban areas, as urban flight and investment demand surged. New Zealand exhibited one of the steepest post-2020 climbs among OECD countries, with HPI values peaking in late 2021 before slightly correcting. On the other hand, Germany and Japan demonstrated more tempered growth, indicating more stable or regulated housing markets. In Sweden and the United Kingdom, sharp gains during the pandemic years gave way to mild price corrections in 2023–2024, possibly reflecting interest rate normalization and affordability pressures.

Overall, while housing prices broadly rose across the OECD during the pandemic, the rate and duration of these increases varied by country depending on housing supply constraints, monetary policy, and broader economic resilience.

Short-Term Interest Rate$\left(y_5\right)$: The short-term interest rate data across OECD countries from 2018 to 2024 reveals substantial fluctuations driven by global macroeconomic dynamics, particularly during and after the COVID-19 pandemic. Prior to the pandemic, rates remained relatively stable and low, especially in the Eurozone (e.g., Germany at ~0.5%) and Japan (close to 1%), reflecting accommodative monetary policy. The United States and Canada maintained moderately higher rates, averaging around 2.5–3.5% through 2019.

As the pandemic unfolded, central banks implemented aggressive rate cuts to mitigate economic disruptions, leading to a sharp decline in interest rates across most economies by mid-2020. For example, the U.S. rate dropped from 3.39% in 2018-Q4 to around 0.33% in 2021-Q2. Similarly, the UK reduced its rates significantly, and many countries maintained near-zero or even negative rates through 2021.

Post-2021, a pronounced upward trend began, particularly in advanced economies responding to persistent inflation. The U.S. Federal Reserve led a sharp hiking cycle from 2022 onward, with rates reaching above 5% by 2024. The UK, Canada, and Australia followed similar paths, though at varied paces. In contrast, countries like Japan and Switzerland continued to maintain ultra-low rates.

This divergence reflects varying inflationary pressures and central bank policy orientations. Overall, the short-term interest rate landscape between 2018 and 2024 illustrates a full monetary policy cycle—starting from pre-pandemic normalization, moving through crisis-induced easing, and culminating in post-crisis tightening.

Figure 1 displays quarterly percentage changes in major macroeconomic indicators between 2018 and 2024, highlighting the disruptive effects of the COVID-19 pandemic and the uneven recovery that followed. A dramatic spike in unemployment and a sharp contraction in GFCF during 2020 Q2 reflect the initial economic shock. In contrast, short-term interest rates remained low until mid-2022, after which they rose steeply in response to inflationary pressures before gradually declining. Other indicators such as GDP, employment, and the CPI show more moderate fluctuations, indicating a slower but steady post-pandemic stabilization. This figure offers a dynamic view of macroeconomic volatility and policy response across the observed period.

5. Results and Discussion

This section presents a comprehensive analysis of productivity dynamics across OECD countries during three distinct periods—pre-COVID (2018Q1–2020Q1), during-COVID (2020Q1–2022Q1), and post-COVID (2022Q2–2024Q4)—using the Malmquist Productivity Index (MPI) and its two decomposed components: efficiency change (catch-up effect) and technological change (frontier shift). The results reveal substantial cross-country variation in the level, stability, and drivers of productivity over time. While the pre-pandemic period was characterized by broad-based but uneven productivity gains, the pandemic introduced both disruptions and adaptive transformations. In the post-COVID recovery phase, some economies rebounded strongly, while others experienced lingering stagnation or volatility. To capture these shifts, we first examine aggregate trends in MPI (Section 5.1), followed by efficiency dynamics (Section 5.2) and technological progress (Section 5.3), before turning to regression-based analysis of policy effects (Section 5.4) and a detailed discussion of robustness and endogeneity (Section 5.5).

5.1. MPI Results

In the Appendix A,

Table A3. Country-Level Malmquist Productivity Index (MPI) Statistics Across Periods.

	Pre-COVID (2018Q1–2020Q1)						During-COVID (2020Q1–2022Q1)						POST-COVID (2022Q1–2024Q4)						TOTAL (2018Q1–2024Q4)
Code	AVG.	CUM.	MIN.	MED.	MAX.	S.D.	AVG.	CUM.	MIN.	MED.	MAX.	S.D.	AVG.	CUM.	MIN.	MED.	MAX.	S.D.	AVG.	CUM.	MIN.	Q1	Q2	Q3	MAX.	S.D.
AUS	1.019	1.162	1.010	1.019	1.028	0.006	1.016	1.136	0.965	1.019	1.060	0.027	1.007	1.084	0.979	1.003	1.040	0.018	1.014	1.430	0.965	1.003	1.018	1.025	1.060	0.018
AUT	1.018	1.151	0.997	1.019	1.030	0.011	1.006	1.035	0.886	1.024	1.076	0.061	1.003	1.032	0.964	0.999	1.073	0.027	1.008	1.229	0.886	0.992	1.011	1.029	1.076	0.036
BEL	1.020	1.170	0.996	1.024	1.028	0.010	1.011	1.082	0.940	1.018	1.060	0.042	1.008	1.090	0.964	0.998	1.076	0.032	1.012	1.380	0.940	0.992	1.018	1.028	1.076	0.030
CAN	1.005	1.039	0.987	1.006	1.019	0.012	1.014	1.120	0.987	1.012	1.058	0.021	1.011	1.128	0.979	1.011	1.040	0.016	1.010	1.311	0.979	1.000	1.010	1.019	1.058	0.016
CHL	1.014	1.116	0.983	1.012	1.064	0.024	1.006	1.049	0.961	1.007	1.055	0.032	1.016	1.182	0.958	1.024	1.059	0.030	1.012	1.384	0.958	0.992	1.014	1.030	1.064	0.028
COL	1.130	2.384	0.998	1.040	1.616	0.219	1.010	1.065	0.924	0.999	1.118	0.072	1.038	1.461	0.956	1.024	1.251	0.082	1.057	3.710	0.924	0.998	1.017	1.076	1.616	0.137
CRI	1.092	1.803	0.751	1.074	1.362	0.192	0.980	0.796	0.740	1.031	1.136	0.130	0.982	0.766	0.776	0.966	1.157	0.114	1.014	1.099	0.740	0.904	1.027	1.090	1.362	0.146
CZE	1.018	1.157	1.005	1.016	1.042	0.012	0.986	0.883	0.935	0.976	1.042	0.047	1.014	1.159	0.928	1.019	1.072	0.046	1.007	1.185	0.928	0.989	1.013	1.041	1.072	0.040
DNK	1.019	1.163	0.980	1.019	1.076	0.029	1.010	1.081	0.961	1.013	1.059	0.035	1.003	1.026	0.915	1.014	1.046	0.038	1.010	1.289	0.915	0.992	1.014	1.036	1.076	0.033
EST	1.005	0.991	0.819	1.006	1.130	0.115	1.066	1.413	0.764	1.069	1.549	0.239	1.003	0.821	0.672	1.015	1.315	0.210	1.022	1.149	0.672	0.874	1.055	1.130	1.549	0.188
FIN	1.017	1.142	1.012	1.015	1.030	0.006	1.007	1.057	0.952	1.011	1.046	0.028	1.014	1.155	0.975	1.005	1.053	0.028	1.013	1.395	0.952	1.000	1.014	1.030	1.053	0.023
FRA	1.021	1.177	0.976	1.028	1.034	0.020	1.003	1.011	0.918	1.004	1.070	0.054	1.009	1.107	0.979	1.009	1.044	0.018	1.011	1.317	0.918	0.997	1.014	1.031	1.070	0.032
DEU	1.011	1.095	1.001	1.014	1.021	0.007	1.008	1.058	0.934	1.017	1.059	0.037	1.004	1.043	0.972	1.007	1.029	0.017	1.007	1.208	0.934	0.997	1.012	1.019	1.059	0.022
GRC	1.012	1.097	0.984	1.014	1.051	0.024	0.999	0.978	0.886	1.002	1.071	0.054	1.004	1.038	0.942	1.003	1.091	0.038	1.005	1.113	0.886	0.985	1.003	1.029	1.091	0.038
HUN	1.021	1.175	0.987	1.018	1.070	0.025	1.002	0.999	0.864	1.000	1.113	0.075	1.015	1.170	0.905	1.033	1.062	0.045	1.013	1.374	0.864	0.988	1.016	1.045	1.113	0.049
ISL	1.004	1.034	0.963	1.006	1.034	0.023	0.991	0.928	0.966	0.992	1.013	0.016	0.990	0.897	0.957	0.994	1.004	0.013	0.995	0.861	0.957	0.982	0.996	1.004	1.034	0.018
IRL	1.042	1.312	0.916	0.995	1.257	0.137	1.068	1.547	0.839	1.029	1.376	0.174	1.008	0.998	0.788	0.979	1.247	0.136	1.036	2.025	0.788	0.929	1.011	1.092	1.376	0.142
ISR	1.006	1.048	0.997	1.006	1.014	0.006	1.011	1.089	0.947	1.023	1.069	0.039	1.009	1.098	0.984	1.005	1.044	0.021	1.009	1.254	0.947	0.997	1.006	1.020	1.069	0.024
ITA	1.015	1.130	0.996	1.017	1.025	0.009	1.000	1.000	0.933	1.003	1.036	0.032	1.006	1.061	0.960	1.001	1.046	0.024	1.007	1.199	0.933	0.997	1.013	1.020	1.046	0.023
JPN	1.003	1.022	0.990	1.001	1.016	0.009	1.010	1.082	0.938	1.012	1.061	0.035	1.066	1.946	0.919	1.038	1.276	0.094	1.031	2.153	0.919	0.999	1.016	1.038	1.276	0.067
KOR	1.013	1.107	0.996	1.009	1.033	0.013	1.011	1.088	0.992	1.015	1.024	0.013	1.013	1.148	0.970	1.016	1.040	0.020	1.012	1.382	0.970	0.999	1.015	1.023	1.040	0.016
LVA	1.035	1.285	0.917	1.047	1.133	0.079	0.999	0.985	0.952	0.992	1.059	0.045	1.006	1.016	0.734	1.030	1.086	0.097	1.013	1.287	0.734	0.971	1.026	1.052	1.133	0.076
LTU	1.030	1.260	0.947	1.030	1.091	0.048	1.001	1.000	0.931	1.004	1.097	0.050	0.988	0.855	0.819	0.993	1.047	0.060	1.004	1.077	0.819	0.981	1.004	1.030	1.097	0.054
LUX	1.006	1.051	0.988	1.008	1.022	0.011	1.014	1.109	0.976	1.000	1.094	0.040	1.005	1.048	0.971	1.005	1.061	0.024	1.008	1.223	0.971	0.996	1.005	1.015	1.094	0.026
MEX	1.018	1.151	0.978	1.011	1.061	0.029	1.000	0.999	0.971	0.992	1.076	0.035	1.010	0.787	0.627	0.995	1.624	0.276	1.010	0.905	0.627	0.971	1.002	1.053	1.624	0.170
NLD	1.018	1.152	0.994	1.011	1.061	0.021	1.021	1.171	0.948	1.027	1.088	0.046	1.012	1.133	0.953	1.005	1.098	0.042	1.016	1.527	0.948	0.998	1.011	1.039	1.098	0.037
NZL	1.018	1.151	1.007	1.015	1.030	0.008	0.989	0.916	0.957	0.983	1.034	0.026	1.014	1.157	0.964	1.010	1.069	0.025	1.008	1.220	0.957	0.998	1.011	1.021	1.069	0.024
NOR	1.005	1.041	0.976	0.998	1.051	0.026	1.061	1.577	0.911	1.067	1.163	0.077	1.001	0.947	0.767	1.024	1.220	0.110	1.020	1.554	0.767	0.976	1.026	1.066	1.220	0.083
POL	1.021	1.181	0.999	1.023	1.033	0.011	1.020	1.170	0.972	1.024	1.048	0.024	1.001	1.007	0.958	1.005	1.040	0.027	1.013	1.392	0.958	0.999	1.020	1.029	1.048	0.024
PRT	1.015	1.123	0.972	1.021	1.029	0.018	1.007	1.039	0.866	1.030	1.088	0.074	1.016	1.180	0.978	1.003	1.084	0.033	1.013	1.376	0.866	0.995	1.021	1.031	1.088	0.044
SVK	1.028	1.244	0.990	1.038	1.056	0.023	1.006	1.042	0.965	1.005	1.057	0.034	1.009	1.083	0.908	1.007	1.102	0.054	1.013	1.403	0.908	0.979	1.021	1.042	1.102	0.040
SVN	1.033	1.292	0.996	1.035	1.058	0.019	0.988	0.907	0.922	1.003	1.024	0.038	1.015	1.124	0.821	1.019	1.115	0.094	1.012	1.317	0.821	0.963	1.020	1.049	1.115	0.063
ESP	1.007	1.057	0.954	1.013	1.029	0.024	1.015	1.118	0.892	1.030	1.059	0.052	1.011	1.130	0.979	1.006	1.067	0.024	1.011	1.336	0.892	0.998	1.014	1.029	1.067	0.033
SWE	1.026	1.225	0.982	1.029	1.047	0.019	1.009	1.069	0.930	1.015	1.050	0.034	1.012	1.133	0.990	1.007	1.034	0.014	1.015	1.485	0.930	1.005	1.018	1.029	1.050	0.023
CHE	1.013	1.109	0.982	1.009	1.039	0.018	1.024	1.191	0.884	1.043	1.065	0.060	1.008	1.084	0.966	1.007	1.068	0.026	1.014	1.431	0.884	1.005	1.009	1.033	1.068	0.036
GBR	1.018	1.156	0.998	1.019	1.038	0.012	1.002	0.988	0.880	1.024	1.121	0.084	1.006	1.055	0.946	1.002	1.093	0.042	1.008	1.205	0.880	0.976	1.014	1.038	1.121	0.050
USA	1.005	1.041	0.991	1.006	1.011	0.006	1.017	1.145	0.966	1.021	1.067	0.028	1.010	1.113	1.004	1.011	1.016	0.005	1.011	1.327	0.966	1.005	1.009	1.015	1.067	0.016
AVG.	1.022	1.189	0.973	1.018	1.075	0.035	1.011	1.079	0.923	1.014	1.090	0.054	1.009	1.088	0.914	1.008	1.105	0.055	1.013	1.392	0.889	0.984	1.015	1.039	1.143	0.052
S.D.	0.024	0.239	0.050	0.015	0.112	0.049	0.019	0.154	0.055	0.019	0.096	0.043	0.013	0.190	0.094	0.014	0.114	0.056	0.010	0.452	0.086	0.027	0.009	0.024	0.156	0.045

This table presents detailed statistics for the Malmquist Productivity Index (MPI) across 37 OECD countries over four periods: pre-COVID (2018Q1–2020Q1), during-COVID (2020Q1–2022Q1), post-COVID (2022Q1–2024Q4), and the total period (2018Q1–2024Q4). Metrics include average (AVG), cumulative (CUM), minimum (MIN), median (MED), maximum (MAX), and standard deviation (S.D.), enabling comparative assessments of productivity levels and volatility over time.

reports Malmquist Productivity Index (MPI) values, highlighting variation in total factor productivity levels and volatility across the pre-, during-, and post-COVID periods.

Pre-COVID period (2018Q1–2020Q1): Most OECD countries experienced positive productivity growth, as indicated by Malmquist Productivity Index (MPI) values consistently exceeding 1.00. This points to broad-based improvements in total factor productivity (TFP) across the region. According to the results Belgium, Austria, and Australia posted stable and sustained productivity gains, with MPI averages around 1.018–1.020 and minimal quarter-to-quarter variability (e.g., Australia, Israel and Finland: Standard Deviation (SD) = 0.006). These patterns suggest resilient labor markets, steady capital investment, and strong institutional environments. France and Poland also displayed robust and consistent MPI performance (average = 1.021), underscoring efficient investment cycles and balanced macroeconomic fundamentals.

In contrast, several emerging and structurally transitioning economies—including Colombia (mean = 1.130; SD = 0.219), Costa Rica (mean = 1.092; SD = 0.192), and Chile—recorded more volatile MPI values. These fluctuations likely reflect external commodity demand shocks, domestic policy reforms, or delayed effects of structural transitions. Ireland stands out for its pronounced variation (mean = 1.042; SD = 0.137), with occasional MPI surges exceeding 1.20, likely attributable to volatile FDI inflows and the dominant influence of multinationals in tech-intensive sectors.

Advanced economies such as Japan (1.003), Canada and the United States (1.005) exhibited marginal productivity gains with low variability, consistent with mature markets facing demographic headwinds and slower innovation spillovers. Meanwhile, Eastern European countries—including Latvia (1.035), Lithuania (1.030), Slovakia (1.028), and Slovenia (1.033)—recorded strong and steady MPI growth, reflecting ongoing EU-driven catch-up processes, improved governance, and structural fund absorption. Overall, while productivity gains were common across the OECD, the pace, stability, and drivers of TFP growth varied significantly by country, shaped by economic structure, investment dynamics, and institutional readiness.

During-COVID period (2020Q1–2022Q1): OECD countries exhibited a mixed picture in terms of productivity dynamics, with a majority maintaining MPI values above 1.00, indicating continued—but often uneven—growth in total factor productivity (TFP). Table A1 in the Appendix A presents summary statistics showing that advanced economies such as Australia (average MPI = 1.016), Canada (1.014), and the United States (1.017) retained stable productivity trajectories, supported by strong institutional frameworks, digital adaptability, and expansive fiscal policies. Meanwhile, Ireland (1.068) and Estonia (1.066) posted some of the highest average MPI values, though with considerable volatility (SD = 0.174 and 0.239 respectively), suggesting the influence of sectoral concentration—particularly in tech and pharma—and sensitivity to external capital flows or structural shifts.

A group of countries, including Switzerland (1.024), The Netherlands (1.021), and Poland (1.020), combined robust MPI values with relatively low standard deviations, indicating consistent productivity gains during the crisis—likely reflecting effective crisis management, targeted policy support, and resilient export sectors. In contrast, emerging economies such as Costa Rica (mean = 0.980) and Czechia (0.986) reported MPI values below 1.00, reflecting productivity declines that may stem from disruptions in tourism, supply chains, or weaker fiscal buffers. Countries like Italy (1.000) and Mexico (1.000) hovered around the neutral mark, showing stagnation likely linked to structural rigidities or prolonged lockdown effects.

Notably, some nations exhibited high productivity growth but with significant fluctuations—such as Norway (growth = 1.577, SD = 0.077)—possibly reflecting energy market volatility, and Ireland, whose productivity dynamics continued to reflect large swings tied to multinational activity. Meanwhile, countries like Japan (MPI = 1.010) and Korea (1.011) maintained moderate but stable growth, consistent with diversified industrial bases and effective pandemic containment.

Overall, the COVID-19 period revealed sharp divergences in TFP performance across the OECD. While some economies demonstrated resilience through adaptive capacity and digital transformation, others faced declines tied to structural fragility or reliance on disrupted sectors. The heterogeneous outcomes underscore the importance of institutional strength, sectoral diversification, and proactive policy in sustaining productivity during crisis periods.

Post-COVID period (2022Q1–2024Q4): The majority of OECD countries continued to experience productivity improvements, as evidenced by average Malmquist Productivity Index (MPI) values exceeding 1.00. This trend suggests a general recovery and adaptation following the economic disruptions caused by the pandemic. However, the nature of these gains varied significantly across countries in terms of magnitude and volatility.

Countries such as Japan (avg MPI = 1.066, growth = 1.946) and Colombia (avg MPI = 1.038, growth = 1.461) recorded some of the strongest total productivity growth. Japan’s surge, accompanied by a relatively high standard deviation (0.094), likely reflects delayed productivity catch-up in sectors undergoing digital transformation, while Colombia’s performance may stem from structural reforms and recovery in commodity-driven sectors. However, both countries also exhibited considerable volatility, with Japan’s MPI ranging from 0.919 to 1.276, and Colombia from 0.956 to 1.251.

Several European economies, including Czechia (avg MPI = 1.014, growth = 1.159), Finland (1.014, 1.155), and New Zealand (1.014, 1.157), demonstrated steady recovery with relatively narrow dispersion, suggesting balanced improvements in efficiency and technology post-COVID. Portugal (1.016, 1.180) and Chile (1.016, 1.182) also reported strong average productivity gains with moderate volatility, reflecting effective investment and demand recovery strategies.

On the other hand, Costa Rica (avg MPI = 0.982, growth = 0.766) and Lithuania (0.988, 0.855) underperformed relative to peers. These results may indicate persistent structural challenges or a slower pace of technological adaptation. Notably, Mexico (avg = 1.010, growth = 0.787) displayed wide fluctuations (SD = 0.276), suggesting instability in recovery momentum—possibly linked to external demand shocks and policy uncertainty.

The United States (avg MPI = 1.010, growth = 1.113) and Germany (1.004, 1.043) posted modest but stable gains, consistent with their status as mature economies. Meanwhile, Estonia (1.003, growth = 0.821) and Ireland (1.008, growth = 0.998)—which had shown strong performance pre- and during-COVID—saw relatively muted post-COVID growth, potentially indicating the fading effects of earlier catch-up phases or FDI volatility.

In summary, post-COVID productivity growth across the OECD reflects broad-based but uneven recovery. While many countries have returned to stable or accelerating productivity trajectories, persistent volatility in several emerging and smaller economies points to lingering structural and external vulnerabilities. These divergences underscore the importance of sustained investment in innovation, labor market adaptability, and institutional resilience in supporting long-term productivity growth.

Figure 2 presents hierarchical cluster dendrograms of OECD countries based on their Malmquist Productivity Index (MPI) values during the pre-COVID (2018Q1–2020Q1), during-COVID (2020Q1–2022Q1), and post-COVID (2022Q2–2024Q4) periods. Before the pandemic, most countries formed cohesive clusters, indicating synchronized productivity trends, particularly among advanced economies like Austria, Finland, Germany, and South Korea, likely due to strong institutions and industrial resilience. Outliers such as Ireland, Colombia, and Chile displayed distinct productivity paths, possibly reflecting structural or sectoral differences.

During the pandemic, clustering became more dispersed, signaling heterogeneous policy responses and varying economic resilience. Countries like the U.S., Greece, Czech Republic, and Slovakia clustered together, suggesting commonalities in adaptation strategies, while Estonia and Ireland stood apart, likely reflecting sharper disruptions or divergent sectoral impacts.

Post-COVID, divergence widened further. A high-performing cluster—U.S., Sweden, Canada, and Israel—emerged, possibly driven by strong digital sectors and effective recovery policies. Meanwhile, Mexico, Estonia, Norway, and Costa Rica remained isolated, suggesting persistent structural barriers to productivity recovery. These shifting cluster patterns underscore growing heterogeneity in productivity dynamics across OECD countries.

5.2. Efficiency Changes (Catch-Up Effect)

Table A4. Country-Level Efficiency Change (Catch-Up Effect) Statistics Across Periods.

	Pre-COVID (2018Q1–2020Q1)						During-COVID (2020Q1–2022Q1)						POST-COVID (2022Q1–2024Q4)						TOTAL (2018Q1–2024Q4)
Code	AVG.	CUM.	MIN.	MED.	MAX	S.D.	AVG.	CUM.	MIN.	MED.	MAX	S.D.	AVG.	CUM.	MIN.	MED.	MAX.	S.D.	AVG.	CUM.	MIN.	Q1	Q2	Q3	MAX	S.D.
AUS	1.021	1.173	0.960	1.025	1.074	0.041	0.990	0.901	0.883	0.985	1.109	0.078	1.002	1.010	0.934	1.004	1.061	0.045	1.004	1.067	0.883	0.961	1.004	1.056	1.109	0.054
AUT	1.033	1.247	0.861	1.037	1.174	0.112	0.972	0.657	0.674	1.028	1.258	0.219	1.019	1.051	0.740	1.024	1.389	0.180	1.009	0.861	0.674	0.901	1.025	1.120	1.389	0.168
BEL	1.031	1.239	0.895	1.032	1.153	0.091	0.973	0.713	0.754	0.996	1.266	0.177	1.014	1.054	0.810	1.015	1.251	0.146	1.007	0.931	0.754	0.895	1.015	1.100	1.266	0.137
CAN	1.003	1.018	0.953	1.018	1.041	0.033	0.995	0.951	0.930	0.980	1.056	0.046	1.008	1.076	0.928	1.000	1.101	0.046	1.002	1.042	0.928	0.974	1.000	1.034	1.101	0.040
CHL	0.998	0.983	0.969	0.996	1.048	0.025	0.986	0.874	0.869	1.003	1.085	0.083	1.059	1.194	0.515	1.009	1.904	0.329	1.019	1.026	0.515	0.969	1.001	1.065	1.904	0.208
COL	1.076	1.606	0.976	1.007	1.612	0.217	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.023	1.606	0.976	1.000	1.000	1.000	1.612	0.116
CRI	1.053	1.341	0.742	1.025	1.347	0.196	1.008	1.000	0.779	1.000	1.283	0.135	1.023	0.777	0.546	1.000	1.898	0.344	1.027	1.042	0.546	0.965	1.000	1.095	1.898	0.243
CZE	1.002	1.016	0.975	1.006	1.022	0.016	0.965	0.662	0.723	0.965	1.275	0.183	1.050	1.206	0.613	1.017	1.744	0.288	1.011	0.812	0.613	0.911	1.002	1.058	1.744	0.202
DNK	1.042	1.301	0.824	1.053	1.233	0.141	0.975	0.707	0.717	1.013	1.264	0.197	1.015	1.030	0.772	1.004	1.366	0.169	1.011	0.947	0.717	0.886	1.009	1.148	1.366	0.162
EST	0.977	0.727	0.690	1.040	1.125	0.180	1.074	1.514	0.697	1.094	1.394	0.219	1.005	0.790	0.710	1.024	1.458	0.242	1.017	0.870	0.690	0.847	1.040	1.132	1.458	0.210
FIN	1.037	1.274	0.864	1.026	1.202	0.123	0.972	0.693	0.712	1.008	1.194	0.188	1.027	1.155	0.778	1.019	1.439	0.181	1.014	1.019	0.712	0.880	1.019	1.138	1.439	0.161
FRA	1.015	1.125	0.999	1.017	1.027	0.011	0.988	0.902	0.934	0.978	1.033	0.036	1.005	1.051	0.964	1.007	1.033	0.018	1.003	1.066	0.934	0.990	1.007	1.021	1.033	0.025
DEU	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
GRC	1.000	1.000	1.000	1.000	1.000	0.000	0.987	0.898	0.926	1.000	1.003	0.027	1.046	0.983	0.496	1.000	2.016	0.360	1.015	0.883	0.496	1.000	1.000	1.000	2.016	0.221
HUN	1.004	1.027	0.972	1.003	1.053	0.025	0.983	0.832	0.878	0.943	1.162	0.117	1.052	1.158	0.555	1.039	1.894	0.320	1.017	0.989	0.555	0.945	0.998	1.048	1.894	0.206
ISL	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
IRL	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
ISR	1.027	1.177	0.844	1.032	1.181	0.123	0.977	0.712	0.703	1.008	1.234	0.199	1.025	1.112	0.738	1.032	1.412	0.191	1.012	0.932	0.703	0.868	1.032	1.141	1.412	0.168
ITA	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
JPN	0.999	0.988	0.977	1.000	1.015	0.014	0.998	0.983	0.962	1.002	1.031	0.026	1.036	1.444	0.993	1.000	1.264	0.080	1.014	1.402	0.962	0.994	1.000	1.019	1.264	0.054
KOR	1.010	1.080	0.966	1.013	1.042	0.025	0.993	0.941	0.926	0.995	1.041	0.034	1.008	1.090	0.961	1.008	1.077	0.035	1.004	1.108	0.926	0.983	1.001	1.025	1.077	0.031
LVA	0.981	0.814	0.814	1.000	1.101	0.114	1.027	1.229	0.975	1.013	1.112	0.045	1.003	1.000	0.827	1.000	1.209	0.086	1.004	1.000	0.814	1.000	1.000	1.027	1.209	0.083
LTU	0.991	0.915	0.882	1.000	1.089	0.065	1.000	0.988	0.918	1.009	1.093	0.055	0.989	0.751	0.705	0.989	1.418	0.184	0.993	0.678	0.705	0.931	1.000	1.025	1.418	0.120
LUX	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
MEX	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.021	1.000	0.633	1.000	1.616	0.226	1.008	1.000	0.633	1.000	1.000	1.000	1.616	0.138
NLD	1.012	1.087	0.932	1.012	1.063	0.051	0.988	0.876	0.828	1.012	1.131	0.100	1.011	1.079	0.858	1.018	1.217	0.094	1.004	1.027	0.828	0.945	1.017	1.061	1.217	0.082
NZL	1.030	1.228	0.870	1.029	1.175	0.102	0.962	0.669	0.724	0.979	1.154	0.151	1.026	1.150	0.735	1.038	1.430	0.179	1.008	0.944	0.724	0.921	1.023	1.108	1.430	0.146
NOR	1.036	1.208	0.780	1.049	1.215	0.166	1.023	0.985	0.716	1.052	1.394	0.238	1.006	0.957	0.813	0.991	1.272	0.154	1.020	1.138	0.716	0.856	1.032	1.187	1.394	0.175
POL	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
PRT	0.998	0.981	0.954	1.001	1.029	0.023	0.985	0.845	0.823	0.986	1.134	0.112	1.052	1.161	0.525	0.999	1.866	0.313	1.016	0.963	0.525	0.967	0.995	1.061	1.866	0.201
SVK	1.010	1.083	0.984	1.004	1.038	0.021	0.994	0.938	0.904	0.990	1.122	0.063	1.033	1.040	0.598	0.995	1.776	0.278	1.015	1.056	0.598	0.979	0.998	1.026	1.776	0.174
SVN	1.003	0.999	0.830	1.010	1.085	0.083	0.995	0.944	0.903	1.024	1.075	0.067	0.991	0.778	0.768	0.970	1.384	0.177	0.996	0.734	0.768	0.920	1.009	1.075	1.384	0.121
ESP	0.992	0.934	0.966	0.993	1.016	0.015	1.005	1.038	0.944	1.007	1.039	0.031	1.007	1.076	0.983	1.000	1.043	0.018	1.002	1.043	0.944	0.990	1.000	1.016	1.043	0.022
SWE	1.040	1.326	0.896	1.044	1.174	0.096	0.963	0.642	0.728	0.978	1.258	0.196	1.022	1.133	0.786	1.011	1.328	0.156	1.010	0.964	0.728	0.917	1.011	1.128	1.328	0.150
CHE	1.050	1.323	0.792	1.021	1.296	0.182	0.967	0.671	0.702	1.014	1.197	0.181	1.009	1.043	0.863	1.008	1.172	0.108	1.009	0.926	0.702	0.887	1.013	1.153	1.296	0.149
GBR	1.005	1.044	0.981	1.006	1.026	0.014	0.999	0.978	0.900	1.009	1.089	0.058	0.999	0.989	0.948	1.000	1.060	0.025	1.001	1.010	0.900	0.991	1.000	1.018	1.089	0.034
USA	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	0.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.000
AVG.	1.013	1.088	0.923	1.013	1.099	0.062	0.993	0.912	0.868	1.002	1.121	0.088	1.015	1.036	0.813	1.006	1.327	0.134	1.008	1.002	0.788	0.953	1.007	1.056	1.353	0.108
S.D.	0.022	0.167	0.085	0.017	0.126	0.066	0.020	0.173	0.116	0.024	0.117	0.080	0.019	0.129	0.168	0.013	0.317	0.117	0.008	0.155	0.166	0.049	0.011	0.056	0.312	0.079

This table reports technical efficiency change statistics—reflecting catch-up to the production frontier—for OECD countries across pre-COVID, during-COVID, post-COVID, and total periods. Measures include average, cumulative, minimum, median, maximum, and standard deviation, supporting cross-country comparisons of efficiency dynamics and recovery paths.

presents detailed metrics on efficiency change (catch-up effect), capturing how effectively countries improved their technical efficiency relative to the production frontier across the pre-, during-, and post-COVID periods.

Pre-COVID period (2018Q1–2020Q1): During the pre-COVID period (2018Q1–2020Q1), countries showed significant variation in efficiency change, reflecting their ability to catch up to the production frontier. Colombia led with the highest efficiency growth (1.606), driven by a sharp improvement in the third quarter of 2019 (EC = 1.612). Strong performers also included Sweden (1.326), Switzerland (1.323), Denmark (1.301), Finland (1.274), Austria (1.247), and Belgium (1.239), each showing substantial catch-up effects through consistent quarterly gains or strong rebounds.

Others, like Israel, Norway, and New Zealand, exhibited notable growth with some volatility, while France and Netherlands posted steady, moderate efficiency improvements. In contrast, Chile (0.983), Spain (0.934), and Estonia (0.727) showed weak or declining efficiency, with multiple quarters below 1.0. Latvia and Lithuania also struggled with inconsistent and generally low scores.

Several countries—including Germany, Greece, Italy, Mexico, Ireland, Iceland, and Luxembourg—maintained flat efficiency (growth = 1.000), suggesting no change in relative technical efficiency. These patterns reflect differences in how effectively countries improved internal operations and resource use before the COVID-19 shock.

During-COVID period (2020Q1–2022Q1): During the COVID-19 period, most countries experienced instability and declines in technical efficiency, reflecting operational disruptions from lockdowns, supply shocks, and policy volatility. While some countries achieved temporary gains, sustained catch-up was rare.

Estonia stood out with the highest efficiency growth (1.514), driven by strong gains in early 2021 and a remarkable spike in the third quarter 2021 (EC = 1.394). Latvia (1.229) and Czechia (1.275) also showed meaningful efficiency improvements, benefiting from relatively consistent scores above 1.0. Other strong performers included Norway, Spain, and Costa Rica, which managed to recover efficiency despite early setbacks.

In contrast, several countries experienced sharp declines in efficiency. Sweden (growth = 0.642), Czechia (0.662), Switzerland (0.671), Austria (0.657), and Finland (0.693) all displayed early gains followed by steep declines, indicating uneven adaptation to pandemic conditions. Belgium, Israel, Denmark, and Portugal also had growth scores below 0.75, with highly volatile quarterly performance. Some countries maintained a flat trajectory with no measurable change in efficiency—Germany, United States, Italy, Mexico, Ireland, Luxembourg, and Poland all recorded a constant growth index of 1.000, indicating either resilience or lack of adjustment.

Overall, efficiency dynamics during the pandemic highlight the difficulties many countries faced in maintaining or improving technical performance. While a few managed sustained catch-up, most experienced regressions or volatility, with pandemic-era uncertainty limiting operational optimization.

Post-COVID period (2022Q1–2024Q4): The post-COVID period (2022Q1–2024Q4) marked a phase of recalibration and uneven recovery across OECD economies. Efficiency change, as captured through quarterly catch-up effects, reveals substantial heterogeneity in how countries navigated the aftermath of the pandemic. Japan reported the highest efficiency growth (1.444), reflecting a consistent improvement in later quarters—particularly in 2023Q2–Q3 (EC = 1.264)—suggesting successful adaptation and optimization post-pandemic. Other notable performers include Czechia (1.206), Chile (1.194), Portugal (1.161), Finland (1.155), and New Zealand (1.150), all of which showed strong quarterly gains, especially in 2024, following mid-period dips.

In contrast, some countries faced renewed inefficiencies. Estonia (growth = 0.790), Slovenia (0.778), and Lithuania (0.751) experienced notable fluctuations with periods of significant decline. For example, Estonia’s efficiency change value fell below 0.75 in early 2022 before recovering. Similarly, countries like Norway and Belgium displayed inconsistent patterns with late recoveries that barely offset earlier setbacks. A number of countries, including Germany, Italy, Ireland, Poland, Mexico, and Luxembourg, maintained a flat efficiency trajectory (growth = 1.000), indicating neither improvement nor decline, while others like the United States and United Kingdom hovered around baseline with minimal change.

Overall, the post-COVID recovery period reflects a re-sorting of efficiency across countries: while some economies capitalized on the reopening phase to enhance technical performance, others remained stagnant or volatile, highlighting differences in institutional resilience and adaptability to post-pandemic economic realities.

Figure 3 presents hierarchical cluster dendrograms of OECD countries based on Efficiency Change (EC) values across the pre-COVID (2018Q1–2020Q1), during-COVID (2020Q1–2022Q1), and post-COVID (2022Q2–2024Q4) periods. In the pre-pandemic period, countries such as Netherlands, France, Poland, and Sweden clustered closely, reflecting broadly similar efficiency gains supported by stable institutions and effective input use. In contrast, Estonia, Mexico, and Colombia formed outlier clusters, likely due to structural inefficiencies or reliance on less adaptable sectors.

During the pandemic, clustering became more fragmented. France and Poland maintained alignment, suggesting effective policy responses, while Estonia and Sweden shifted into distinct clusters—possibly reflecting contrasting containment strategies and their impact on input reallocation. Mexico and Ireland also diverged, indicating challenges in adjusting production processes or absorbing shocks.

Post-COVID clusters reveal continued divergence. Poland, Slovakia, and Czech Republic emerged as a leading cluster, potentially driven by manufacturing recovery and efficient reallocation of inputs. Meanwhile, Sweden, Estonia, and Mexico remained isolated, suggesting persistent sectoral rigidities or delayed recovery. These shifting and increasingly dispersed cluster structures reflect growing asymmetry in countries’ ability to adapt and recover in terms of technical efficiency. The patterns highlight the role of institutional flexibility, policy responsiveness, and sectoral composition in shaping the speed and success of post-pandemic efficiency recoveries.

5.3. Technological Changes (Frontier-Shift)

In the Appendix,

Table A5. Country-Level Technological Change (Frontier Shift) Statistics Across Periods.

	Pre-COVID (2018Q1–2020Q1)						During-COVID (2020Q1–2022Q1)						POST-COVID (2022Q1–2024Q4)						TOTAL (2018Q1–2024Q4)
Code	AVG.	CUM.	MIN.	MED.	MAX	S.D.	AVG.	CUM.	MIN.	MED.	MAX	S.D.	AVG.	CUM.	MIN.	MED.	MAX.	S.D.	AVG.	CUM.	MIN.	Q1	Q2	Q3	MAX	S.D.
AUS	1.000	0.991	0.954	0.997	1.060	0.039	1.031	1.260	0.938	1.028	1.139	0.065	1.007	1.073	0.938	1.000	1.073	0.046	1.012	1.340	0.938	0.966	1.002	1.053	1.139	0.050
AUT	0.995	0.923	0.864	0.989	1.174	0.113	1.079	1.576	0.822	1.018	1.451	0.226	1.013	0.982	0.715	1.000	1.350	0.179	1.027	1.428	0.715	0.896	1.000	1.143	1.451	0.172
BEL	0.996	0.944	0.888	0.991	1.139	0.089	1.068	1.518	0.837	1.036	1.355	0.192	1.012	1.034	0.806	0.980	1.231	0.142	1.024	1.482	0.806	0.920	0.997	1.139	1.355	0.141
CAN	1.003	1.021	0.973	1.001	1.041	0.024	1.021	1.177	0.956	1.023	1.099	0.044	1.005	1.048	0.914	1.003	1.090	0.046	1.009	1.258	0.914	0.982	1.003	1.035	1.099	0.039
CHL	1.016	1.135	0.999	1.015	1.031	0.009	1.024	1.200	0.962	1.004	1.113	0.057	1.039	0.990	0.534	0.997	1.888	0.323	1.028	1.348	0.534	0.988	1.014	1.032	1.888	0.199
COL	1.054	1.484	1.003	1.023	1.293	0.097	1.010	1.065	0.924	0.999	1.118	0.072	1.038	1.461	0.956	1.024	1.251	0.082	1.034	2.310	0.924	0.998	1.020	1.030	1.293	0.081
CRI	1.040	1.345	0.963	1.011	1.194	0.079	0.975	0.796	0.876	0.981	1.067	0.081	1.034	0.985	0.565	1.024	1.768	0.296	1.018	1.055	0.565	0.949	1.012	1.056	1.768	0.191
CZE	1.016	1.138	0.997	1.019	1.040	0.014	1.049	1.333	0.817	1.001	1.333	0.172	1.028	0.961	0.571	0.983	1.621	0.269	1.031	1.459	0.571	0.960	1.016	1.091	1.621	0.187
DNK	0.993	0.894	0.840	0.986	1.204	0.129	1.074	1.529	0.825	1.019	1.425	0.220	1.011	0.996	0.742	0.995	1.314	0.161	1.024	1.361	0.742	0.887	0.995	1.154	1.425	0.166
EST	1.048	1.362	0.880	1.001	1.319	0.147	1.003	0.934	0.782	0.985	1.295	0.166	1.007	1.039	0.831	1.004	1.174	0.083	1.018	1.321	0.782	0.939	1.004	1.097	1.319	0.125
FIN	0.993	0.896	0.848	0.988	1.192	0.125	1.070	1.526	0.846	1.020	1.387	0.200	1.014	1.001	0.699	1.003	1.354	0.174	1.024	1.369	0.699	0.891	0.994	1.154	1.387	0.163
FRA	1.006	1.047	0.976	1.005	1.027	0.015	1.015	1.120	0.954	1.027	1.048	0.034	1.005	1.053	0.965	1.002	1.048	0.026	1.008	1.235	0.954	0.990	1.006	1.027	1.048	0.025
DEU	1.011	1.095	1.001	1.014	1.021	0.007	1.008	1.058	0.934	1.017	1.059	0.037	1.004	1.043	0.972	1.007	1.029	0.017	1.007	1.208	0.934	0.997	1.012	1.019	1.059	0.022
GRC	1.012	1.097	0.984	1.014	1.051	0.024	1.013	1.089	0.886	1.015	1.106	0.068	1.054	1.056	0.488	1.007	2.023	0.363	1.029	1.261	0.488	0.982	1.008	1.040	2.023	0.225
HUN	1.017	1.145	0.993	1.017	1.039	0.013	1.026	1.201	0.913	0.994	1.165	0.088	1.040	1.010	0.529	1.011	1.861	0.315	1.029	1.389	0.529	0.976	1.015	1.071	1.861	0.197
ISL	1.004	1.034	0.963	1.006	1.034	0.023	0.991	0.928	0.966	0.992	1.013	0.016	0.990	0.897	0.957	0.994	1.004	0.013	0.995	0.861	0.957	0.982	0.996	1.004	1.034	0.018
IRL	1.042	1.312	0.916	0.995	1.257	0.137	1.068	1.547	0.839	1.029	1.376	0.174	1.008	0.998	0.788	0.979	1.247	0.136	1.036	2.025	0.788	0.929	1.011	1.092	1.376	0.142
ISR	0.992	0.891	0.845	0.982	1.196	0.126	1.073	1.529	0.832	1.017	1.419	0.216	1.013	0.987	0.703	0.998	1.371	0.180	1.025	1.345	0.703	0.898	0.997	1.153	1.419	0.171
ITA	1.015	1.130	0.996	1.017	1.025	0.009	1.000	1.000	0.933	1.003	1.036	0.032	1.006	1.061	0.960	1.001	1.046	0.024	1.007	1.199	0.933	0.997	1.013	1.020	1.046	0.023
JPN	1.004	1.035	0.995	1.004	1.015	0.007	1.012	1.101	0.975	1.017	1.041	0.024	1.029	1.348	0.919	1.015	1.124	0.065	1.017	1.536	0.919	0.995	1.009	1.028	1.124	0.043
KOR	1.003	1.025	0.979	1.002	1.031	0.018	1.019	1.155	0.967	1.022	1.077	0.035	1.005	1.053	0.924	1.005	1.076	0.040	1.009	1.247	0.924	0.986	1.003	1.031	1.077	0.033
LVA	1.061	1.579	0.924	1.050	1.133	0.068	0.974	0.802	0.877	0.966	1.059	0.061	1.003	1.016	0.888	1.025	1.081	0.062	1.012	1.287	0.877	0.955	1.030	1.052	1.133	0.069
LTU	1.042	1.378	0.986	1.037	1.138	0.049	1.003	1.012	0.913	0.992	1.115	0.067	1.028	1.139	0.700	1.009	1.457	0.195	1.025	1.587	0.700	0.968	1.009	1.072	1.457	0.127
LUX	1.006	1.051	0.988	1.008	1.022	0.011	1.014	1.109	0.976	1.000	1.094	0.040	1.005	1.048	0.971	1.005	1.061	0.024	1.008	1.223	0.971	0.996	1.005	1.015	1.094	0.026
MEX	1.018	1.151	0.978	1.011	1.061	0.029	1.000	0.999	0.971	0.992	1.076	0.035	0.990	0.787	0.785	1.002	1.295	0.156	1.001	0.905	0.785	0.978	1.002	1.035	1.295	0.098
NLD	1.008	1.060	0.953	1.002	1.082	0.045	1.041	1.337	0.913	1.027	1.206	0.103	1.009	1.049	0.825	0.991	1.206	0.105	1.018	1.487	0.825	0.968	1.006	1.064	1.206	0.087
NZL	0.996	0.937	0.869	0.989	1.162	0.102	1.052	1.369	0.871	1.022	1.337	0.175	1.014	1.007	0.703	1.006	1.368	0.173	1.020	1.292	0.703	0.901	0.996	1.118	1.368	0.149
NOR	0.993	0.862	0.820	0.990	1.256	0.165	1.080	1.600	0.814	1.020	1.423	0.220	1.010	0.989	0.749	0.989	1.275	0.158	1.026	1.365	0.749	0.844	0.994	1.206	1.423	0.174
POL	1.021	1.181	0.999	1.023	1.033	0.011	1.020	1.170	0.972	1.024	1.048	0.024	1.001	1.007	0.958	1.005	1.040	0.027	1.013	1.392	0.958	0.999	1.020	1.029	1.048	0.024
PRT	1.017	1.144	0.992	1.018	1.039	0.013	1.029	1.229	0.938	0.990	1.155	0.085	1.043	1.017	0.524	1.006	1.912	0.328	1.031	1.430	0.524	0.975	1.016	1.051	1.912	0.205
SVK	1.018	1.149	1.000	1.014	1.047	0.017	1.015	1.110	0.942	1.006	1.129	0.061	1.033	1.041	0.581	1.005	1.732	0.270	1.023	1.328	0.581	0.977	1.012	1.047	1.732	0.168
SVN	1.035	1.294	0.954	1.030	1.200	0.080	0.997	0.960	0.921	0.979	1.124	0.073	1.055	1.445	0.736	1.024	1.384	0.223	1.032	1.795	0.736	0.954	1.003	1.088	1.384	0.149
ESP	1.016	1.132	0.987	1.021	1.025	0.013	1.010	1.077	0.945	1.022	1.042	0.033	1.005	1.050	0.951	1.006	1.049	0.028	1.010	1.281	0.945	0.996	1.013	1.025	1.049	0.025
SWE	0.995	0.924	0.877	0.992	1.158	0.102	1.082	1.666	0.835	1.039	1.395	0.203	1.011	1.000	0.757	0.986	1.300	0.159	1.027	1.540	0.757	0.904	0.997	1.125	1.395	0.155
CHE	0.991	0.838	0.799	0.989	1.273	0.172	1.088	1.775	0.859	1.048	1.336	0.181	1.009	1.039	0.859	0.976	1.169	0.115	1.027	1.545	0.799	0.879	1.001	1.155	1.336	0.150
GBR	1.013	1.107	0.991	1.016	1.019	0.009	1.002	1.010	0.924	1.023	1.034	0.041	1.006	1.067	0.946	1.002	1.067	0.036	1.007	1.193	0.924	0.991	1.016	1.028	1.067	0.031
USA	1.005	1.041	0.991	1.006	1.011	0.006	1.017	1.145	0.966	1.021	1.067	0.028	1.010	1.113	1.004	1.011	1.016	0.005	1.011	1.327	0.966	1.005	1.009	1.015	1.067	0.016
AVG.	1.013	1.102	0.945	1.007	1.109	0.058	1.029	1.217	0.903	1.011	1.183	0.099	1.016	1.051	0.795	1.002	1.307	0.136	1.019	1.379	0.787	0.957	1.007	1.070	1.345	0.110
S.D.	0.018	0.172	0.061	0.015	0.093	0.052	0.031	0.246	0.057	0.018	0.143	0.071	0.016	0.124	0.154	0.012	0.285	0.104	0.010	0.257	0.148	0.042	0.008	0.052	0.277	0.068

This table summarizes technological change across OECD countries, capturing shifts in the production frontier due to innovation and technological advancement. For each period (pre-, during-, post-COVID, and total), the table provides average, cumulative, minimum, median, maximum, and standard deviation values, highlighting disparities in technological adaptation and resilience.

summarizes technological change across the pre-, during-, and post-COVID periods, reflecting shifts in the frontier driven by innovation.

Pre-COVID period (2018Q1–2020Q1): During the pre-COVID period, technological progress—as measured by frontier shifts—varied widely across OECD countries. Latvia, Colombia, and Estonia led in average technological change, with indices above 1.05, reflecting strong innovation and adoption of productivity-enhancing technologies. Latvia’s sustained gains and Colombia’s late-period surge highlight proactive policy and industrial advancements.

Countries like Ireland and Lithuania also recorded consistent gains, likely driven by digital transformation and integration into global value chains. In contrast, more advanced economies such as Austria, Belgium, and Germany showed modest or fluctuating frontier shifts, suggesting slower adaptation or innovation saturation. Overall, the period revealed early signs of divergence in technological progress—an important precursor to post-pandemic resilience and productivity outcomes.

During-COVID period (2020Q1–2022Q1): During the COVID-19 crisis, technological shifts varied widely across OECD countries, reflecting both disruptions and adaptive innovation. Countries such as Switzerland, Sweden, and Norway led the frontier shift, with average indices exceeding 1.08. These gains were likely driven by accelerated digital transformation, automation in manufacturing, and strong institutional support for innovation during lockdowns.

Austria and Denmark also ranked among the top performers, though their frontier shifts showed high volatility. Each experienced sharp dips in 2020 followed by significant rebounds, highlighting the uneven adoption of technology and recovery in sectors like health, logistics, and public services. Despite the pandemic’s constraints, many countries leveraged the crisis to push the technological frontier, though the path was non-linear. Several economies faced mid-period declines—such as Belgium and Canada—before regaining momentum, reflecting sectoral disruptions and gradual adaptation.

Overall, the data suggests that the pandemic acted as both a stress test and a catalyst for innovation, with clear divergence in how countries moved the production frontier. Those with robust digital infrastructure and flexible policy frameworks adapted more rapidly, laying a foundation for post-pandemic productivity gains.

Post-COVID period (2022Q1–2024Q4): In the aftermath of the pandemic, OECD countries exhibited highly varied patterns of technological progress. The average frontier shift remained positive for most countries, suggesting continued innovation momentum. However, the post-COVID period was marked by high volatility, with notable spikes and dips reflecting uneven recovery and adaptation.

Slovenia, Greece, and Portugal stood out with the highest average technological progress, each exceeding an index of 1.04. These gains were driven by sharp increases in late 2024, possibly reflecting delayed implementation of digital transformation initiatives or EU-funded modernization programs. However, these countries also recorded some of the lowest quarterly values—e.g., Greece’s 0.49 and Portugal’s 0.52—highlighting an unstable innovation trajectory. Meanwhile, countries like Canada and Australia showed steadier, more moderate gains, with averages close to 1.00 and relatively low standard deviations. Their consistent performance may signal stronger institutional support and smoother post-pandemic transitions in innovation ecosystems.

Overall, the post-COVID period emphasized a divergent recovery in technological change, with some economies surging ahead in bursts while others adopted more incremental innovation paths. The variation underscores the lasting impact of the pandemic on national innovation capacity, policy readiness, and sectoral adaptability.

Figure 4 presents the evolution of cluster structures based on Technological Change (TC), capturing how the production frontier shifted across OECD countries during three key periods. In the pre-COVID phase, clustering indicates a moderate degree of convergence. Countries such as Korea, Canada, Germany, and the U.S. form tight clusters, likely reflecting consistent investment in R&D and digital infrastructure. Meanwhile, Greece, Chile, and Israel are more distant, pointing to structural or sectoral limitations in advancing technological capabilities.

During the COVID period, cluster patterns become more dispersed. Countries like Japan, Italy, and France shift positions, reflecting uneven capacity to sustain innovation during crisis conditions. A notable cluster of Germany, the U.S., and Korea persists, signaling resilience in tech-driven sectors. In contrast, countries such as New Zealand and Czech Republic appear isolated, suggesting a lag in adapting production technologies during the disruption.

In the post-COVID recovery period, clustering becomes more polarized. Advanced economies such as Germany, Korea, the U.S., and Canada continue to lead, forming a stable core of technological advancement. However, countries like Greece, Chile, and Slovakia remain outliers, highlighting persistent gaps in innovation ecosystems or digital readiness. These evolving clusters underscore growing disparities in how countries shifted their production frontiers during and after the pandemic—pointing to long-term implications for global competitiveness and convergence.

The resilience and post-pandemic acceleration of technological change can be traced to several interconnected factors.

• The pandemic dramatically accelerated digital adoption: Firms and households rapidly embraced remote work tools, e-commerce platforms, cloud computing, and automation technologies to maintain operations amid stringent lockdowns and mobility restrictions (Bloom et al., 2025). Digital-intensive economies such as Ireland and Estonia experienced notable surges in ICT services exports, reflecting how digital transformation supported productivity and technological advancements even during the height of the crisis.
• The reorganization of production processes played a critical role: In response to global supply chain disruptions, many firms adopted leaner and more automated systems to reduce reliance on labor-intensive operations. For instance, Germany and Japan leveraged robotics, digital supply chain management, and Industry 4.0 technologies to enhance flexibility and efficiency, driving a forward shift in the production frontier.
• Specific industries experienced innovation surges that contributed disproportionately to TC: The pharmaceutical sector underwent unprecedented R&D acceleration, particularly in the development and deployment of mRNA vaccines. Similarly, ICT, logistics, and healthcare sectors introduced breakthrough innovations to meet the demands of the crisis, which strengthened technological progress at the macroeconomic level.
• Sectoral resource reallocation contributed to the observed advancements: Labor, capital, and research efforts shifted away from low-productivity sectors such as tourism and hospitality towards high-productivity sectors like digital services, healthcare, and technology-intensive manufacturing. This reallocation supported structural adjustments that sustained technological gains.
• The adaptive policy measures of 2021 appear to have reinforced these trends: As shown in our regression robustness checks (Section 5.5), 2021 stringency measures were positively associated with TC, suggesting that more targeted and flexible interventions not only minimized economic disruptions but also encouraged firms to invest in digital transformation and process innovations. This pattern highlights the dual role of policy in both mitigating immediate public health challenges and indirectly fostering long-term technological resilience.

Our results on technological change are consistent with evidence presented by Bloom et al. (2021), who showed that pandemic-related challenges accelerated innovation in remote-work technologies and digital infrastructure. Similarly, Baqaee and Farhi (2022) emphasize that sectoral shifts toward digital and knowledge-intensive industries helped offset productivity losses in traditional sectors, which is mirrored in our findings on the resilience of TC during the pandemic.

The clustering analysis based on MPI, EC, and TC (Figure 2, Figure 3 and Figure 4) reveals clear groupings of OECD countries that reflect differences in macroeconomic performance, technological adaptation, and efficiency dynamics before, during, and after the pandemic. The clusters illustrate how the pandemic accelerated structural divergence across economies.

• High-Productivity Cluster: This cluster consists of countries that consistently outperform others in terms of MPI and TC, particularly Ireland, South Korea, the United States, and the Nordic countries (Denmark, Sweden, and Finland). These economies benefitted from robust digital infrastructure, significant investment in R&D, and advanced technological capabilities. For example, Ireland’s strong ICT sector and South Korea’s rapid digital transformation allowed them to weather the pandemic with minimal productivity losses and even record gains in TC. These countries also implemented targeted fiscal measures and flexible labor market policies that enabled quick adaptation to shifting economic conditions.
• Medium-Productivity Cluster: The second cluster includes economies such as France, Germany, Canada, and Italy, which experienced moderate productivity improvements during the post-COVID period. While these countries implemented effective pandemic responses, they faced constraints in fully leveraging technological change. In Germany, for instance, the automotive and manufacturing sectors rebounded slowly due to global supply chain disruptions, despite strong pre-pandemic industrial bases. These countries demonstrate steady, if slower, recovery trajectories compared to the high-productivity group.
• Low-Productivity Cluster: The third cluster primarily consists of countries with persistent efficiency and technological challenges, including Spain, Greece, Portugal, and some Central European economies. These countries are characterized by heavy reliance on tourism and traditional services, sectors that were severely impacted by prolonged restrictions. Higher unemployment and inflation rates in these economies contributed to lower MPI scores, and technological improvements have been slower due to structural rigidities and weaker digital infrastructure.
• Temporal Dynamics: The cluster differentiation became more pronounced during the pandemic (2020Q1–2022Q1) and persisted into the post-COVID recovery (2022Q1–2024Q4). Countries in the high-productivity cluster not only maintained their lead but also widened the gap due to accelerated adoption of digital technologies and proactive policy interventions. In contrast, countries in the low-productivity cluster struggled to recover lost efficiency and lagged in implementing structural reforms.

These results highlight the structural and policy-driven commonalities within each cluster. High-productivity economies share characteristics such as strong digital infrastructure, innovation-driven industries, and flexible labor markets, while low-productivity countries exhibit economic structures that are more vulnerable to external shocks. This analysis suggests that pandemic-era productivity outcomes were heavily influenced by pre-existing economic fundamentals and the speed with which countries adapted to new technological and organizational paradigms. These cluster patterns align with findings from Andrews et al. (2021), who documented that countries with robust digital infrastructure and high innovation intensity were better able to sustain productivity-enhancing reallocation during the pandemic. Our clustering results further suggest that countries like Ireland and South Korea, which prioritized digital transformation prior to COVID-19, maintained stronger technological momentum, supporting the broader argument that pre-pandemic investments in intangibles were key to resilience (Haskel & Westlake, 2021).

A comprehensive overview of productivity patterns across pre-, during-, and post-COVID periods is provided in

Table 1. Cross-Period Summary of Productivity Dynamics in OECD Countries.

	Pre-COVID (2018Q1–2020Q1)	During-COVID (2020Q1–2022Q1)	Post-COVID (2022Q1–2024Q4)
Malmquist Productivity Index (MPI)	Most OECD countries experienced steady productivity growth (MPI > 1.00), especially advanced economies like Belgium, Austria, and Australia with low variability (SD < 0.01). Volatile high performers included Colombia (MPI = 1.130, SD = 0.219) and Costa Rica.	Productivity trends diverged: Australia, Canada, and US maintained MPI > 1.01. Ireland (MPI = 1.068, SD = 0.174) and Estonia (MPI = 1.066, SD = 0.239) were volatile. Costa Rica and Czechia declined below MPI = 1.00, reflecting pandemic impacts.	Japan (avg MPI = 1.066) and Colombia (1.038) led recovery, with substantial gains. Czechia, Finland, and New Zealand showed balanced performance (avg MPI ~1.014). Costa Rica (0.982) and Lithuania (0.988) lagged, indicating structural challenges.
Efficiency Change (EC)	Countries like Colombia (EC = 1.606), Sweden, and Switzerland showed strong efficiency gains. Moderate performers included France and Netherlands. Some countries, like Germany, Ireland, and Mexico, recorded no change (EC = 1.000).	Efficiency declined across many countries. Estonia (EC = 1.514), Czechia (1.275), and Latvia (1.229) showed resilience. Others like Sweden (0.642), Austria (0.657), and Finland (0.693) experienced steep drops. Several maintained flat EC = 1.000.	Japan (EC = 1.444), Czechia (1.206), Chile (1.194), and Portugal (1.161) were among the top efficiency improvers. Estonia and Slovenia regressed (EC < 0.80), while Germany, Ireland, and Mexico showed flat growth (EC = 1.000).
Technological Change (TC)	Technological progress was led by Latvia, Colombia, and Estonia (TC > 1.05). Ireland and Lithuania showed consistent gains. Many advanced economies had stable but modest frontier shifts, indicating early divergence in innovation capacity.	Countries like Switzerland, Sweden, and Norway had strong frontier shifts (TC > 1.08), driven by digital transformation. Austria and Denmark showed rebound patterns. Others, like Belgium and Canada, had mid-period declines before partial recovery.	Slovenia, Greece, and Portugal exceeded TC > 1.04 but with high volatility. Greece’s TC ranged from 0.49 to 1.2. Canada and Australia achieved steady TC gains. Divergence widened between innovation leaders and structurally lagging countries.

, which details the evolution of Malmquist Productivity Index (MPI), Efficiency Change (EC), and Technological Change (TC) across OECD countries. Figure 5 visualizes country-level trends in productivity dynamics across the pre-COVID, during-COVID, and post-COVID periods. Using a heat map, it highlights changes in Malmquist Productivity Index (MPI), Efficiency Change (EC), and Technological Change (TC), facilitating cross-country comparisons of performance and resilience.

5.4. Impact of Government Stringency

This subsection investigates how variations in government stringency policies during the COVID-19 crisis (2020–2022) influenced cumulative productivity performance in the post-pandemic year spanning 2023Q1 to 2024Q1. Using two regression specifications—one with annual indicators and one with quarterly time dummies—we estimate the impact of past policy intensity on three outcome variables: the Malmquist Productivity Index (MPI), efficiency change (catch-up effect), and technological change (frontier shift), each measured cumulatively over the 2023–2024 period. By focusing on this specific post-crisis window, we capture not just immediate shocks but the delayed and potentially enduring effects of pandemic-era interventions on long-run economic adaptability and innovation. This approach rests on the premise that government-imposed restrictions, while necessary for public health, may have altered production structures, investment flows, and efficiency paths in ways that only became visible once countries transitioned into a relatively stable recovery phase. The results below in

Table 2. Regression Estimates of Quarterly and Annual Effects of Government Stringency on Productivity, Efficiency, and Technological Change.

Model I	MPI					Efficiency Change (Catch-Up Effect)					Technological Change (Frontier-Shift)
Term	Coef	SE Coef	T-Value	p-Value	VIF	Coef	SE Coef	T-Value	p-Value	VIF	Coef	SE Coef	T-Value	p-Value	VIF
Constant	1.0970	0.1570	6.99	0.000		1.1640	0.1190	9.79	0.000		0.9298	0.0774	12.01	0.000
2020	−0.0110	0.0040	−2.76	0.009	2.15	−0.0074	0.0030	−2.45	0.020	2.15	−0.00474	0.00196	−2.42	0.021	2.15
2021	0.0073	0.0044	1.66	0.106	2.53	0.0045	0.0033	1.36	0.184	2.53	0.00454	0.00217	2.09	0.044	2.53
2022	0.0029	0.0037	0.78	0.439	1.27	0.0001	0.0028	0.05	0.962	1.27	0.00102	0.00181	0.56	0.578	1.27
Model II	MPI					Efficiency Change (Catch-Up Effect)					Technological Change (Frontier-Shift)
Term	Coef	SE Coef	T-Value	p-Value	VIF	Coef	SE Coef	T-Value	p-Value	VIF	Coef	SE Coef	T-Value	p-Value	VIF
Constant	1.0390	0.1750	5.93	0.000		1.1380	0.1410	8.05	0.000		0.8688	0.0869	9.99	0.000
2020 Q1	0.0077	0.0039	1.97	0.060	1.30	0.0041	0.0032	1.29	0.210	1.30	0.0046	0.0019	2.36	0.027	1.30
2020 Q2	−0.0069	0.0031	−2.20	0.038	2.87	−0.0048	0.0025	−1.89	0.070	2.87	−0.0025	0.0016	−1.63	0.116	2.87
2020 Q3	0.0022	0.0027	0.83	0.417	4.08	0.0022	0.0022	1.01	0.322	4.08	0.0005	0.0013	0.36	0.725	4.08
2020 Q4	−0.0065	0.0035	−1.85	0.076	4.93	−0.0053	0.0028	−1.89	0.071	4.93	−0.0015	0.0017	−0.88	0.388	4.93
2021 Q1	−0.0002	0.0042	−0.05	0.961	7.75	−0.0001	0.0034	−0.03	0.973	7.75	−0.0010	0.0021	−0.49	0.630	7.75
2021 Q2	0.0100	0.0040	2.49	0.020	5.61	0.0074	0.0032	2.28	0.032	5.61	0.0053	0.0020	2.63	0.015	5.61
2021 Q3	−0.0052	0.0041	−1.28	0.212	5.14	−0.0041	0.0033	−1.27	0.218	5.14	−0.0028	0.0020	−1.40	0.174	5.14
2021 Q4	0.0099	0.0040	2.45	0.022	6.19	0.0067	0.0032	2.07	0.050	6.19	0.0054	0.0020	2.71	0.012	6.19
2022 Q1	−0.0090	0.0040	−2.26	0.033	4.60	−0.0063	0.0032	−1.97	0.060	4.60	−0.0041	0.0020	−2.07	0.050	4.60
2022 Q2	0.0017	0.0051	0.33	0.742	5.13	−0.0001	0.0041	−0.01	0.990	5.13	0.0016	0.0025	0.65	0.524	5.13
2022 Q3	0.0069	0.0070	0.99	0.332	6.50	0.0050	0.0056	0.89	0.381	6.50	0.0003	0.0035	0.09	0.927	6.50
2022 Q4	−0.0008	0.0061	−0.13	0.901	4.36	−0.0008	0.0049	−0.15	0.880	4.36	0.0009	0.0030	0.30	0.768	4.36

shed light on how early, mid-, and late-pandemic stringency measures shaped the trajectory of post-COVID productivity outcomes across countries.

The regression estimates presented in this section examine the cumulative changes in productivity, efficiency, and technological progress over the one-year period from 2023Q1 to 2024Q1. The dependent variables—Malmquist Productivity Index (MPI), efficiency change (catch-up effect), and technological change (frontier shift)—capture total changes during this specific post-pandemic year. By regressing these 2023–2024 cumulative outcomes on government stringency measures from 2020 to 2022, we aim to understand how earlier policy responses shaped countries’ longer-run productivity capacity once the immediate crisis had passed.

This design is motivated by the recognition that there are time lags between policy interventions and observable productivity impacts. Stringent measures taken during 2020–2022 affected firms’ and governments’ operational decisions, technology investments, and labor dynamics in ways that likely manifested more fully once pandemic-related volatility subsided. The period from 2023Q1 to 2024Q1, by contrast, represents a relatively stable phase in which countries operated without acute COVID disruptions—making it ideal for assessing the residual or delayed consequences of prior policy choices.

The regression results show that higher stringency in 2020 is significantly associated with lower cumulative productivity in 2023–2024. The MPI coefficient for 2020 is negative and statistically significant (−0.0110, p = 0.009), driven by both a reduction in technical efficiency (−0.0074, p = 0.020) and a decrease in technological progress (−0.0047, p = 0.021). These findings suggest that while early stringency may have been necessary from a public health standpoint, it came at a cost: it created frictions that suppressed post-pandemic productivity performance, possibly by crowding out investment, delaying structural change, or creating scarring effects in labor and capital allocation.

By contrast, stringency in 2021 appears to have supported technological advancement, with a positive and statistically significant effect on frontier shift (+0.0045, p = 0.044), even though its effects on MPI and efficiency are not statistically different from zero. This pattern suggests that while 2021 may not have directly improved short-term output or operational efficiency, it was a critical year for setting the stage for future innovation—perhaps through digital infrastructure investments, supply chain diversification, or adaptive policy learning.

Stringency in 2022, however, is not associated with any significant changes in cumulative productivity, efficiency, or technology in 2023–2024. All coefficients are small and statistically insignificant across models, indicating that by 2022, most of the long-term productivity implications of COVID-era policy choices had already been locked in. This reinforces the idea that 2020 and 2021 were the pivotal years for shaping post-pandemic trajectories.

The quarterly regression results provide additional nuance to this interpretation. Periods of high stringency, particularly 2020Q2 and 2020Q4, align with significant quarterly declines in both MPI and efficiency. In contrast, quarters with moderate relaxation—such as 2021Q2 and 2021Q4—coincide with productivity rebounds and measurable gains in both technical efficiency and frontier advancement. These transitions suggest that the balance and timing of policy stringency—rather than its absolute level—played a decisive role in whether countries emerged from the pandemic with strengthened or weakened productivity foundations.

These results resonate with the findings of Calligaris et al. (2023), who highlighted how policy interventions, such as job retention schemes, preserved productive capacity during the pandemic. Our analysis adds to this literature by showing that while early, severe lockdowns were associated with reduced post-pandemic productivity, adaptive policies in 2021 appeared to promote technological advancement, potentially by encouraging firms to adopt digital solutions and reorganize production processes.

Taken together, these findings indicate that the productivity performance of countries during 2023–2024 was not simply a function of recovery capacity, but also a reflection of policy choices made during the preceding years. While early stringency slowed efficiency and innovation, more adaptive and targeted measures in 2021 may have fostered durable technological gains. By 2022, however, diminishing returns set in. The cumulative effects observed in the post-crisis year highlight the importance of aligning emergency interventions with forward-looking strategies that support resilience and innovation in the medium run.

5.5. Robustness and Endogeneity

A key limitation of our regression analysis is the potential endogeneity of government stringency measures. Stringent policies were implemented primarily in response to pandemic severity, which itself had a direct negative impact on productivity. Therefore, the observed relationship between stringency and post-pandemic productivity may partly reflect unobserved confounding factors. To address this concern, we conducted a set of robustness checks aimed at reducing simultaneity bias and testing the stability of our results.

First, we replaced the annual average stringency index for 2020 with its lagged value from 2020Q2, the quarter when the initial and most severe lockdown measures were implemented. This approach ensures that the predictor (2020Q2 stringency) is temporally distant from the post-pandemic productivity outcomes (2023–2024) and less likely to suffer from reverse causality. The results, reported in

Table 3. Robustness Checks on the Impact of Government Stringency on 2023 MPI.

Model	Coefficient (β)	SE	T-Value	p-Value	R-sq (%)
Lagged Stringency (2020Q2)	−0.00633	0.00210	−3.02	0.005	20.7
2020 Stringency (Year Average)	−0.01174	0.00382	−3.07	0.004
2021 Stringency (Year Average)	0.00886	0.00391	2.27	0.030	21.8

Notes: The dependent variable is 2023 MPI (Malmquist Productivity Index). The first model uses stringency from 2020Q2 as a lagged predictor, while the second model includes year-specific measures for 2020 and 2021 stringency. Both models confirm that early lockdowns are associated with lower post-pandemic productivity, while later adaptive policies positively correlate with technological advancements.

, show that 2020Q2 stringency remains negatively and significantly associated with 2023 MPI (β = −0.0063, p = 0.005), explaining 20.7% of the variation.

Second, we explored year-specific stringency effects by including separate measures for 2020 and 2021. As shown in Table 3, 2020 stringency retains a negative and significant coefficient (β = −0.0117, p = 0.004), while 2021 stringency shows a positive and significant effect (β = +0.0089, p = 0.030). These results suggest that while stringent early lockdowns likely caused lasting disruptions to productivity, the more adaptive measures in 2021 may have facilitated digital transformation, reorganization of production processes, and sectoral reallocation, supporting long-term technological change.

Finally, while these robustness checks enhance confidence in our findings, the results must still be viewed as associative rather than causal. The severity of the pandemic and other unobserved factors (e.g., sectoral composition, fiscal stimulus intensity) likely influenced both stringency levels and productivity outcomes. Future research could address this limitation through instrumental variable (IV) approaches—for example, using early COVID-19 exposure in neighboring countries or pre-pandemic healthcare capacity as instruments—or through structural models designed to isolate exogenous policy effects.

6. Conclusions

This study examined macroeconomic productivity dynamics across 37 OECD countries between 2018Q1 and 2024Q4, offering an in-depth decomposition of total factor productivity (TFP) into efficiency change (catch-up effect) and technological change (frontier shift). Using a Slack-Based Measure Data Envelopment Analysis (SBM-DEA) framework with the Malmquist Productivity Index (MPI), the study traced productivity developments across three distinct phases—pre-COVID, during-COVID, and post-COVID—and linked them to the intensity and timing of government policy responses during the pandemic.

The analysis reveals that, while productivity gains were broadly sustained across OECD countries, the sources, stability, and magnitude of these gains varied significantly over time and across economies. Prior to the pandemic, most countries demonstrated steady improvements in TFP, with several Eastern European members—such as Latvia, Lithuania, and Slovakia—showing strong catch-up growth supported by structural reforms and EU integration. Mature economies like Australia, France, and Belgium achieved stable performance through balanced investment and institutional strength. Conversely, emerging economies such as Colombia and Costa Rica experienced more volatile productivity trajectories, driven by external demand fluctuations and structural transformations.

The COVID-19 period introduced sharp divergence in productivity patterns. Although many countries maintained MPI values above 1.00, reflecting resilience, the underlying efficiency dynamics were far more heterogeneous. While Estonia and Ireland achieved notable gains, reflecting agility and digital adaptation, several others—particularly in Southern and Eastern Europe—suffered sharp declines in efficiency, likely due to their reliance on disrupted sectors like tourism, rigid labor structures, or constrained fiscal space. Technological change proved more robust than efficiency change during this period, as countries with strong digital infrastructure (e.g., Sweden, Korea, Switzerland) leveraged the crisis to accelerate frontier-shifting innovation.

Post-pandemic data indicate a broad but uneven recovery in productivity. Countries like Japan and Colombia registered strong growth, while others—such as Germany, Ireland, and Luxembourg—recorded flat or declining efficiency, pointing to varying capacities for operational adaptation in the recovery phase. The frontier continued to advance in most cases, though patterns were inconsistent. While some economies experienced sustained innovation momentum, others saw progress in short bursts, reflecting differences in investment capacity, institutional readiness, and exposure to global headwinds.

The dendrogram analyses of MPI, EC, and TC reveal growing divergence in productivity dynamics across OECD countries before, during, and after the COVID-19 pandemic. Prior to the crisis, most countries clustered closely, reflecting shared institutional strengths and synchronized growth. However, the pandemic introduced sharp fragmentation—especially in efficiency and technological change—exposing disparities in countries’ capacity to adapt to disruption. While some advanced economies (e.g., Germany, Korea, the U.S.) maintained resilience through strong innovation systems, others (e.g., Estonia, Mexico, Greece) showed more erratic or isolated trajectories. Post-COVID, a partial reconvergence is observed in efficiency among select Eastern European countries, but technological gaps have widened, with a small group of advanced economies pulling ahead. These patterns underscore the importance of investing in innovation, digital infrastructure, and institutional adaptability to support sustained recovery and long-term convergence.

Regression results add further nuance by linking cumulative productivity outcomes in 2023–2024 to the intensity of government stringency measures during the crisis. High stringency in 2020 was significantly associated with lower post-pandemic productivity, both in terms of efficiency and technology, suggesting that early restrictive measures—while necessary for public health—had lingering adverse effects on economic adaptability. In contrast, stringency in 2021 correlated positively with frontier shifts, indicating that strategic interventions during the second pandemic year may have laid the groundwork for longer-term innovation through digital infrastructure expansion and supply chain diversification. By 2022, however, the impact of stringency on productivity outcomes had largely dissipated, reinforcing the idea that early policy responses had the most durable consequences.

These findings yield several important policy insights. First, while swift containment measures are critical during crises, their long-term economic costs must be mitigated through proactive investment in digitalization, institutional flexibility, and labor market adaptability. Countries that balanced short-term health priorities with forward-looking innovation strategies—such as digital infrastructure investment and targeted fiscal stimulus—emerged more resilient. Second, governments should embed productivity-enhancing reforms within their emergency policy frameworks to ensure that crisis responses do not merely stabilize output but also strengthen long-term efficiency and technological capacity. Third, the experience of the pandemic underscores the need to diversify economic structures and reduce reliance on vulnerable sectors. Countries that demonstrated greater sectoral resilience, institutional adaptability, and coordinated policy execution were more likely to sustain and amplify their productivity trajectories during and after the crisis.

Finally, the study highlights the value of multidimensional, efficiency-based assessments of macroeconomic performance that go beyond GDP alone. Policymakers should routinely monitor productivity through frameworks that integrate both desirable (e.g., output growth) and undesirable (e.g., inflation, unemployment) outcomes, allowing for more precise targeting of reforms. While this study is limited by data constraints and the aggregate nature of national indicators, future research should build on these findings by incorporating firm- or sector-level data, environmental dimensions of productivity, and the role of human capital development in sustaining innovation-led growth.

In conclusion, productivity resilience is not an automatic consequence of economic recovery—it is shaped by a country’s structural strengths, policy choices, and capacity to adapt. As countries move beyond the pandemic, the imperative is not only to regain lost output but to create a more efficient, equitable, and innovation-driven growth model equipped to handle future global disruptions.

This study has several limitations. First, the DEA methodology is deterministic and sensitive to outliers and measurement errors. Efficiency scores are relative to the best-performing countries within the sample. Second, cross-country data consistency is a challenge, and some data for late 2024 may be provisional. Most notably, the Oxford Stringency Index ends in 2022, limiting post-COVID policy analysis. Third, the second-stage regression may suffer from endogeneity and omitted variable bias, as factors like fiscal stimulus size or structural changes are not directly captured. To ensure robustness, we conducted additional analyses (Table 3), including lagged stringency measures (2020Q2) and year-specific effects (2020 vs. 2021). Both checks confirm that strict early lockdowns (2020) were correlated with lower post-pandemic productivity, whereas adaptive measures in 2021 were positively related to technological change, potentially due to the accelerated adoption of digital technologies, structural reorganization, and sectoral resource shifts. Nevertheless, these results must be interpreted as associative rather than causal, since pandemic severity simultaneously influenced both policy stringency and productivity outcomes. Future research could strengthen causal inference by employing instrumental variable approaches or structural models to disentangle the exogenous effects of policy interventions.

Future work could apply this framework to industry-level data to uncover sector-specific dynamics. Incorporating variables on fiscal policy, digitalization, or human capital would deepen insights. Alternative approaches like Stochastic Frontier Analysis (SFA) could help separate inefficiency from random shocks. Lastly, examining non-linear policy effects and the role of institutional context would enhance understanding. The SBM-DEA and MPI framework offers a strong foundation for such expanded macroeconomic assessments.