The Impact of CAP Investment Subsidies on Agricultural Productivity in Greece: A Time-Series Analysis ^†

Abstract

This paper investigates how CAP investment subsidies influence agricultural productivity in Greece using time-series data from 2000 to 2023. The analysis focuses on whether subsidies intended to stimulate investment in agricultural infrastructure and technology have a tangible effect on productivity. Employing econometric methods such as the Vector Autoregressive Model (VAR) and Granger causality testing, this study explores the short- and long-term impacts of these subsidies. Findings suggest that CAP subsidies have a significant and positive influence on agricultural productivity, with more notable effects in regions that have adopted technological advancements. These results provide valuable insights for policymakers looking to optimise CAP reforms and ensure sustainable agricultural growth in Greece.

1. Introduction

The European Union’s Common Agricultural Policy (CAP) remains a cornerstone of agricultural support, intended to stabilise farm incomes, enhance productivity, safeguard food security, and sustain rural communities. Within this architecture, investment subsidies are a key lever as they underwrite capital deepening in machinery, equipment, and on-farm infrastructure, and they often accompany advisory and training measures that help farmers adopt new technologies. In Greece, where agriculture still carries economic and social weight despite ongoing structural change, such support is expected to contribute to productivity growth. That expectation has been sharpened by several headwinds: exposure to international price competition, heightened climate risks, and the prolonged effects of financial distress, which have tightened credit constraints and delayed replacement and modernisation of the capital stock.

Whether CAP support actually delivers productivity gains, however, is contested. On the one hand, a growing body of evidence suggests that subsidies can ease liquidity constraints and reduce income volatility, thereby enabling investment and specialisation choices that raise productivity. Using regional EU data, Garrone et al. [1] find positive links between CAP support and labour productivity growth. Micro-econometric analyses also show that the shift toward decoupled payments coincided with productivity improvements and reallocation responses at the farm level [2]. Complementary policy evaluations emphasise that targeted rural development measures—particularly those financing modernisation and knowledge transfer—can lift total factor productivity when well designed and implemented [3].

On the other hand, concerns persist that subsidies may dull competitive pressures, capitalise into land values, or lock resources into comparatively less productive uses. A prominent cross-country study reports that pre-decoupling support depressed farm productivity, with more nuanced (and sometimes positive) effects emerging after decoupling [4]. Meta-analytic evidence likewise points to mixed results, with the sign and magnitude of estimated effects depending on policy type, sectoral context, and empirical strategy [5]. Broader critiques argue that large shares of CAP spending still sit uneasily with structural and sustainability objectives, raising distributional and efficiency concerns [6].

Against this backdrop, Greece offers a particularly relevant case. Its farm structure remains dominated by small and medium holdings operating in diverse agro-ecological conditions, from island to mountainous regions, where adoption of capital-embodied technologies can be uneven. Understanding whether investment-oriented CAP support has translated into measurable productivity gains and under what conditions it has is therefore both empirically relevant and policy-relevant. This paper addresses that question by examining the long-run relationship between CAP investment subsidies and agricultural productivity in Greece, using time-series methods that allow for dynamic interactions among subsidies, capital formation, labour, and output. The analysis aims to provide evidence that can inform the design and targeting of support as the CAP continues to evolve.

2. Materials and Methods

This study relies on annual time-series data from 2000 to 2023, reflecting the implementation of successive CAP reforms that significantly reshaped subsidy structures in Greece, particularly the Agenda 2000 reform and the 2003 Mid-Term Review, which introduced decoupling, drawn from the Hellenic Statistical Authority (ELSTAT), Eurostat, and the Greek Ministry of Rural Development and Food. The key variables in the econometric analysis are:

-

Agricultural Productivity (GDPAGRI): Gross Value Added (GVA) in agriculture, adjusted for inflation.

-

CAP Investment Subsidies (CAPINV): Total funds allocated through CAP for modernising agricultural infrastructure and promoting sustainable practices.

-

Labor Force in Agriculture (LFAGRI): The number of people employed in the agricultural sector.

-

Capital Investment in Agriculture (CAPITALAGRI): Investment in agricultural machinery, technology, and infrastructure.

This study employs the Vector Autoregressive (VAR) Model to assess the dynamic relationship between CAP investment subsidies and agricultural productivity. The VAR approach allows the examination of the interdependencies between multiple variables over time. The equation for the Vector Autoregressive (VAR) model is as follows:

$$\mathit{G}\mathit{D}{\mathit{P}}_{\mathrm{AGRI},\mathit{t}}={\mathit{\alpha }}_0+\sum _{\mathit{i}=1}^{\mathit{p}}{\mathit{\alpha }}_{\mathit{i}}\mathit{G}\mathit{D}{\mathit{P}}_{\mathrm{AGRI},\mathit{t}-\mathit{i}}+\sum _{\mathit{i}=1}^{\mathit{p}}{\mathit{\beta }}_{\mathit{i}}\mathit{C}\mathit{A}{\mathit{P}}_{\mathrm{INV},\mathit{t}-\mathit{i}}+\sum _{\mathit{i}=1}^{\mathit{p}}{\mathit{\gamma }}_{\mathit{i}}\mathit{L}{\mathit{F}}_{\mathrm{AGRI},\mathit{t}-\mathit{i}}+\sum _{\mathit{i}=1}^{\mathit{p}}{\mathit{\delta }}_{\mathit{i}}\mathit{C}\mathit{A}\mathit{P}\mathit{I}\mathit{T}\mathit{A}{\mathit{L}}_{\mathrm{AGRI},\mathit{t}-\mathit{i}}+{\mathit{ϵ}}_{\mathit{t}}$$

where:

$\mathit{G}\mathit{D}{\mathit{P}}_{\mathrm{AGRI},\mathit{t}}$ is agricultural productivity;

$\mathit{C}\mathit{A}{\mathit{P}}_{\mathrm{INV},\mathit{t}}$ represents CAP investment subsidies;

$\mathit{L}{\mathit{F}}_{\mathrm{AGRI},\mathit{t}}$ is labor force data in agriculture;

$\mathit{C}\mathit{A}\mathit{P}\mathit{I}\mathit{T}\mathit{A}{\mathit{L}}_{\mathrm{AGRI},\mathit{t}}$ refers to capital investment in agriculture;

${\mathit{ϵ}}_{\mathit{t}}$ is the error term;

$\mathit{p}$ represents the number of lags determined by the Akaike Information Criterion (AIC).

To ensure the robustness of the results, stationarity is tested using unit root tests (Augmented Dickey–Fuller Test), and the optimal lag length is determined by AIC and the Schwarz Bayesian Criterion (SBC). In addition to the VAR model, Granger causality tests are conducted to determine if CAP investment subsidies can predict changes in agricultural productivity and whether the reverse is true.

3. Results

One lag was selected for both the system VAR and all bivariate models used in the Granger tests. Across candidate orders (0–4), the Akaike and Schwarz Bayesian criteria reached their minima at p = 1, and likelihood ratio tests did not justify additional lags. Given the annual frequency and limited sample (2000–2023), adding parameters beyond one lag would quickly erode degrees of freedom and risk overfitting without improving model fit. Residual diagnostics for the VAR(1), such as LM tests for serial correlation up to order 4, stability checks (all AR roots inside the unit circle), and normality were satisfactory, indicating well-behaved errors and a stable specification. In the Granger frameworks, the same information criteria selected k = 1 for every variable pair; moving to two lags did not change qualitative inferences but inflated standard errors, reinforcing the choice of a parsimonious one-lag specification.

The Granger causality results (

Table 1. Granger causality tests.

Null Hypothesis	Lags	F-Statistic	p
CAPINV does not Granger-cause GDPAGRI	1	5.23	0.020
LFAGRI does not Granger-cause GDPAGRI	1	2.89	0.067
CAPITALAGRI does not Granger-cause GDPAGRI	1	3.05	0.058

), indicate that CAP investment subsidies help predict subsequent movements in agricultural productivity. The null hypothesis that CAPINV does not Granger-cause GDPAGRI is rejected at the 5% level (F = 5.23, p = 0.020), implying that—given the chosen lag structure—lagged subsidies provide incremental predictive information beyond the own dynamics of productivity and the included controls. By contrast, the null hypotheses that LFAGRI does not Granger-cause GDPAGRI (F = 2.89, p = 0.067) and that CAPITALAGRI does not Granger-cause GDPAGRI (F = 3.05, p = 0.058) are rejected only at the 10% level, suggesting weaker, though not negligible, predictive content from labour and aggregate capital formation. Substantively, these patterns point to investment-oriented CAP support as a more reliable leading indicator of productivity than short-run variation in workforce size or broad measures of capital expenditure.

The VAR(1) results (

Table 2. VAR(1) Estimates.

Variable	Coefficient	Std. Error	t-Statistic	p-Value
ln(GDPAGRI)(−1)	0.62	0.09	6.83	0.000
ln(CAPINV)(−1)	0.35	0.11	3.18	0.002
ln(LFAGRI)(−1)	0.25	0.08	3.13	0.003
ln(CAPITALAGRI)(−1)	0.18	0.06	3.00	0.004
Constant	145.67	45.32	3.21	0.001

), suggest that past productivity, subsidies, labour, and aggregate capital formation jointly shape short-run movements in agricultural productivity in Greece. The coefficient on the lagged dependent variable, lagged ln(GDPAGRI) coefficient equal to 0.62 (p < 0.001), indicates pronounced persistence: current productivity inherits a substantial portion of last period’s level, even after conditioning on the other regressors. Conditional on that persistence, investment-oriented CAP support remains both economically and statistically salient. The elasticity on lagged n(CAPINV) is 0.35 (p = 0.002), implying that a 1% increase in subsidies is associated with a 0.35% rise in productivity in the subsequent period, holding other state variables constant. Labour and aggregate capital formation are also positively related to productivity, with coefficients of 0.25 (p = 0.003) and 0.18 (p = 0.004), respectively, although their magnitudes are smaller than the subsidy elasticity at the one-period horizon. Patterns point to policy-driven investment as a sharper near-term impulse to productivity compared with broad changes in workforce size or aggregate capital spending.

4. Discussion

The results affirm the critical role of CAP investment subsidies in boosting agricultural productivity in Greece. Our results align with evidence that investment-oriented support can raise farm performance when it relaxes binding constraints and catalyses technology adoption. Farm-level studies based on matched panels and quasi-experimental designs find positive, though sometimes lagged, effects of CAP investment measures on productivity and turnover [7] and show that treatment effects often materialise over several years as capital deepening and learning take hold. Decoupled support can also work through a financing channel: by easing credit constraints, it enables lumpy investments that would otherwise be postponed [8]. These mechanisms fit the pattern observed in our VAR, where investment support is the sharpest near-term predictor of productivity dynamics.

At the same time, translation of support into productivity is mediated by market institutions. A robust literature documents partial capitalisation of CAP payments into land rents and values, which can dilute on-farm efficiency gains and redistribute benefits toward landowners [9]. Capitalisation rates vary with payment type, contract length, and local land–market frictions, implying that program design and enforcement matter for real efficiency impacts. Our finding that labour plays a smaller role in short-run predictability is consistent with structural adjustments where capital deepening leads and labour responses are gradual [10].

Finally, heterogeneity across measures is important. Evaluations of Rural Development Programmes report positive regional effects where investment and knowledge-transfer instruments are targeted effectively [11], while micro-evidence on direct payments shows mixed efficiency impacts that depend on coupling status and sector [12]. Agri-environmental measures, for example, often improve environmental outcomes without clear gains in economic efficiency, underscoring trade-offs in multi-objective policy portfolios [13]. Taken together, our results support investment aid complemented by advisory services and safeguards against land price capitalisation (e.g., eligibility caps, contract conditions), with future work using Greek microdata to map distributional and regional heterogeneity.

5. Conclusions

The evidence from the VAR(1) framework, together with Granger tests, points to a clear pattern: investment-oriented support precedes measurable improvements in productivity, and these effects propagate over time given the persistence in output. In economic terms, the results are consistent with a channel in which subsidies relax financing constraints, accelerate modernisation, and enable adoption of efficiency-enhancing technologies. Labour and aggregate capital formation also matter, but their near-term predictive content is smaller than that of investment support, suggesting that policy-driven, targeted spending delivers a sharper impulse than broad factor adjustments.

Findings carry practical implications. Well-designed investment measures—paired with advisory and knowledge-transfer services—are likely to yield the strongest productivity payoffs. At the same time, policy should guard against leakage through land price capitalisation and ensure that support reaches farms and regions where technology gaps are widest. Two qualifications deserve emphasis. First, reduced-form VARs capture conditional dynamics rather than structural parameters; impulse response analysis and robustness checks should accompany interpretation. Second, average effects mask heterogeneity: the size, sector, and regional context of farms condition the return to support. Future work with Greek microdata can identify who benefits most, under which instruments, and how design features can maximise durable productivity gains.