This study applies a case study approach at regional level, with one case per region, located five in European countries: Kostelec in Czechia (CZ); Upper Rhine Valle in Germany (DE); Istria in Croatia (HR); Kainuu in Finland (FIN); and Galicia in Spain (ESP; Figure 1). This procedure captures specific contexts and practices, allowing to explore how they influence enterprise operations and decision-making processes across Europe. These CSs were selected to cover major biogeographic areas, and diverse forest management systems and FVC structure across Europe. In each country, a representative forest enterprise was selected as an exemplar, taking into account the processing capacities, which reflect the distinct structure and scale of the forest-based enterprises. Considering that, CZ, DE, and HR have mid-scale wood processing industries, where the use of timber and forest products is mostly confined to the region, and the export capabilities of the forest-based industries are limited. These forest-based sectors often served domestic or nearby markets, constrained by logistical infrastructure and certain production technologies. While FIN and ESP feature as large-scale wood processing industries, they showcased diverse processing industries and greater integration with international markets. The main disturbances varied between case studies along the considered period, with a prominent role of bark beetle outbreak in CZ, windstorms in DE and FIN, ice-storm and windstorms in HR and wildfires in ESP. The disturbances were reflected in the salvage logging volume, which was used as a proxy of the volume of damaged timber. The volume and quality of damage timber was affected by disturbance agent prevalent in particular regions. This approximation is commonly used in literature (Butry et al., 2001; Prestemon and Holmes, 2004; González-Gómez et al., 2013). The type of processing capacity and its versatility not only increases the potential for value recovery after disturbances, but also market outlets and the recovery of damaged timber.

Data was gathered according to García-Jácome et al. (2025), considering annual data for the period 2001 to 2021. The dataset included information on case studies in terms of harvesting systems, technologies, timber production and market price of sawlogs, pulpwood and energy wood.

In this study, revenue was used as a system variable, representing a quantitative indicator of the performance of the FVCs social-ecological system in response to disturbances. A system variable is a measurable attribute that responded to external pressures and characterizes the system’s structure and function over time (Nikinmaa et al., 2023; Baho et al., 2017). We followed Asada et al. (2023) to obtain the system variable, the revenue for sawlogs, pulpwood and energy wood, where the source of the data was UNECE/FAO (2021) and United Nations (2021). However, to ensure cross-country comparability, all prices were kept in USD. Additionally, we applied inflation corrections using the World Bank (2025), allowing us to isolate real price changes from general inflation or currency effects and ensure an accurate reflection of economic trends over time.

To assess the resilience dynamics of revenue across the case studies, we identified 11 key resilience predictors (sensu Lloret et al., 2024; Table 1). These predictors offer insight into specific mechanisms or components that enhance a system’s ability to absorb disturbances and adapt to change (Standish et al., 2014). This approach will help tailor to local needs while informing broader European strategies for sustainable forest-based economies.

For the resilience assessment, a forest enterprise was considered sustainable and resilient when it was able to generate stable revenue over time while remaining within ecological limits, represented by the economic upper threshold (Ec.UTH). In turn, consistently generating revenues covered operational costs, defined by the economic lower threshold (Ec.LTH).

The Ec.UTH was calculated by first determining the total available timber supply at the upper threshold level (TS.UTH). As it is described in García-Jácome et al. (2025) “UTH is the upper resilience threshold, defined as the mean annual logging prescriptions according to forest management plan rom year 1 (2001) to year n (2021) of the observed period; and n is the number of years within the observed period.”

First, the Timber supply UTH was calculated as:

UTH = ∑ 1 n Prescribed Logging Volume Number of years (García-Jácome et al., 2025).

This volume was then divided into different assortments (sawlogs, pulpwood and energy wood) based on their respective shares obtained from appropriate timber yield tables. Each assortment’s volume is multiplied by its mean annual market price (MP.SP, MP.PP and MP.EP), and the results are summed up to calculate the total economic value, referred to as Ec.UTH.

While Ec.UTH provides a guideline for sustainable revenue, it should not be considered as a strict rule. Factors like price fluctuations can temporarily increase revenue without necessarily leading to overexploitation.

Then, the Ec.UTH was then calculated by:

The Ec.LTH represents the minimum total revenue required for a forest enterprise to sustain its essential operations. It was calculated by taking the minimum observed annual revenue (Min Revenue) and the expected profitability ratio for each case study, according to literature-based values (Annex I).

The Ec.LTH was calculated as

where x% denotes the expected profitability (expressed as a decimal). This formulation assumes that when total revenue falls below Ec.LTH, the enterprise becomes economically unsustainable, that is, unable to cover its essential operating costs once profit margins are accounted for.

If the system variable, revenue, stays within the defined economic upper and lower thresholds, then the exemplar can be considered economically resilient. This means it’s generating enough revenue to cover costs without exceeding ecological limits.

To assess the robustness of the Ec.LTH estimates, a sensitivity analysis was performed by adjusting the Ec. LTH values by ±10 and ±50%. These scenarios tested the influence of moderate and extreme deviations in the profitability assumptions that would affect the classification of resilience states across the case studies.

We evaluated the role of salvage logging on the operational economies of forest enterprises. So, we calculated the proportion of salvage logging relative to total timber supply from 2001 to 2021. This metric allows us to identify and quantify long-term trends in salvage logging dependence and its impact on forestry markets.

The proportion of salvage logging was calculated as:

Where the percentage of salvage logging is relative to total timber supply (Pct.HS.SL) represents the share of volume of unplanned timber extraction due to disturbances. The total timber supply was used as the denominator because the goal was to show the proportion of salvage logging within the total volume extracted, regardless of planning. This approach helped to assess the relative impact of disturbances on overall supply. However, in some case studies, salvage logging even exceeds the reported total timber supply, suggesting it may not have been completely placed on the market within that year, thus spilling over to subsequent year.

We also analyzed the price trend for sawlogs, pulpwood and energy wood adjusted for inflation and expressed in USD per CS. This analysis helped track long-term price dynamics, detecting market disruptions, and allowed exploring links with revenue fluctuations.

We used Spearman correlations to explore the bivariate relationship between economical revenue and the different predictors (Table 1), Then we employed PCA to reduce the dimensionality of predictors (Gower et al., 2011; Ficko et al., 2019; Riccioli et al., 2020) in order to visualize regional economic patterns and detect temporal trends in economic resilience, in terms of economic revenue. The PCA was conducted using data from 2001 to 2021, and its Dim1 accounted for 70% of the total variability. This PCA was performed by R Studio (version 4.3.1).

Following the explanatory PCA that identified general spatial and temporal trends, we employed a panel regression model to estimate the specific effects of the predictors on forest based economic revenue, while controlling for unobserved heterogeneity across countries and years.

We estimated a two-way fixed effects panel regression, which controls for both country-specific ( μ i ) and year-specific ( λ t ) effects. On one hand, country fixed effects capture time-invariant characteristics of each national forestry system, such as forest resource, management structure, or institutional condition. On the other hand, year fixed effects absorb external shocks common to all countries (e.g., economic shocks, disturbance years, pandemic). This specification isolates the within-country, overtime variation in the explanatory variables that drives changes in real timber revenue.

With many related predictors, multicollinearity was a major concern. To minimize this and ensure stable estimation, variables were pre-processed (Annex II), and the transformed variables were used in the subsequent models.

Model A “Scale model,” examined how total planned harvest, salvage intensity, mechanization and overall market price influenced revenue:

Where revenue ( Revenu e it ) in country i in year t is explained by the volume of planned harvested ( HS _ Planne d it ), the proportion of salvage logging to total supply ( Pct _ HS _ SL 01 it ); the degree of mechanization ( HS _ CT L it ), and the aggregate price level for timber assortment ( Price _ Inde x it ).

Log-transformed variant was also estimated to interpret coefficients as elasticities and to reduce heteroscedasticity:

Model B “Composition model,” replaced total harvest with type of assortments volume of sawlogs, pulpwood and energy wood to test whether changes in harvest composition affected revenue:

Model significance was tested using F-statistics for joint significance of regressor and Wald Test for time and group effects. The models were estimated using the plm package (Croissant and Millo, 2008) in RStudio (version 4.3.1) (RStudio Team 2020). Heteroscedasticity-robust standard errors were computed and clustered by both country and year to ensure inference robustness with a small cross-sectional sample (five case studies). The FE was used because unobserved structural characteristics of each country are likely correlated with the explanatory variables (Wooldridge, 2010).

Figure 2 illustrates the percentage of timber supply sourced from salvage logging across the five case studies. A higher proportion of salvage logging indicates a stronger reliance on reactive harvesting rather than planned timber extraction. The fluctuating trend reflects the severity and frequency of disturbances.

In CZ and DE, there was a clear upward trend in the proportion of timber coming from salvage logging, particularly from 2015 onwards, likely related to the bark beetle crisis in CZ and extreme weather like drought in DE. In DE specifically, this increase represented a shift from the earlier declining trend and was marked by a considerable year-to-year variation, which increased unplanned timber extraction. In FIN, there was a notable spike around 2018, coinciding with a major windstorm and snow damage. In HR, the relative percentage of salvage logging remained consistently low throughout the period under consideration, though it showed some fluctuations. ESP maintained the lower, more stable proportions of salvage logging, indicating that their timber supply was primarily sourced from planned operations.

For CZ (Figure 3), all prices of timber grades showed an upward trend until around 2007–2008 (global financial crisis), followed by a decline and fluctuations in later years. Sawlog and pulpwood prices followed a similar trajectory, with notable fluctuations occurring from 2010 to 2020, with a slight recovery only for sawlogs in 2021. The energy wood prices were consistently lower than the other two assortments, with a drop in the late 2010s. In contrast, sawlog prices in HR remained relatively stable when considering the whole 2001–2021 period, but experienced periodic peaks, possibly reflecting sporadic market disruptions. As for pulpwood, there is an increase from 2006 to 2009, where it remained stable and dropped in 2021. In the case of energy wood, the same as in CZ can be observed, it remained lower than the other two assortments but with less fluctuations than CZ. In the case of DE, the sawlog prices declined with fluctuations over time, while pulpwood prices exhibited an increasing trend until 2011–2013, declining again afterwards. Wood energy prices remained relatively low and stable, with some moderate fluctuations. In FIN, price volatility was observed, particularly for sawlogs and pulpwood, with peaks around 2007–2010 and subsequent declines. Energy wood prices remained more stable, but with a remarkable drop after 2007 and recovering by 2020–2021 Finally, in ESP, sawlogs remained relatively stable over the period. However, pulpwood and energy wood showed important fluctuations, with important peaks during 2007–2008 for pulpwood and 2009–2010 for energy wood. Across the observed period, sawlog prices tended to be more stable over the long term (HR and ESP). Notably, CZ and FIN experienced market fluctuations, particularly for sawlogs and pulpwood prices. In contrast, DE and HR exhibited more moderate price movements. Finally, energy wood consistently remained the lowest-priced assortment.

This section evaluates economic resilience by examining revenue behavior relative to the established upper (UTH) and lower thresholds (LTH) in response to natural disturbances, estimated by salvage logging (m3) (Figure 4). In CZ, the revenue exhibited a rather steady behavior after 2007, but presented several notable breaches over the UTH, with a maximum value of 6.1 million USD. Although salvage logging fluctuated, and followed peaks of revenue in 2007 and 2021, it does not appear to significantly buffer revenue losses, such as in 2009 or after 2011. In HR, the revenue initially remained within the thresholds, but it rose over the UTH after 2016, reaching a peak of 247.07 million USD, indicating a temporary surge in activity, likely as a compensatory harvesting following disturbance events. However, the moderate previous levels of salvage logging suggests that additional economic or policy mechanisms likely played a role in stabilizing revenue.

The German case study exhibited prolonged breaches of the UTH. It was significantly affected by Storm Lothar in 1999/2000, which led to a substantial reduction in timber stock. This decline resulted in a situation where standard harvesting levels necessitated an increase in harvesting intensity beyond sustainable limits. This disruption is reflected in the highest observed deviation from the UTH, reaching 1.8 million USD, which coincides with a peak in salvage logging following the storm event. Overall, sustained high revenue suggests a robust market response, supported either by increased timber prices or intensified harvesting efforts.

In FIN, the revenue showed dramatic spikes, from 2006 to 2012, peaking at 1.12 million USD. After this period, fluctuations were less extreme but still generally remained above the UTH. In contrast, the largest spike in salvage logging occurred in 2018, driven by a major disturbance event involving ice and storm damage. This disturbance resulted in only a moderate increase in revenue in 2019. However, the delayed revenue response in subsequent years suggests that surplus salvage timber was stored and gradually introduced to the market, contributing to a more stable revenue stream over time.

Finally, ESP presented multiple breaches of the UTH; particularly a sharp increase in 2018 coincided with major disturbance events, likely linked to 2017 wildfire season in Galicia as reflected in the peak of salvage logging. The highest revenue exceeded 1.9 million USD. Salvage logging events in ESP in the study period (2001–2021) were often followed by a slight increase in revenue, however, these increases were typically temporary, with revenues declining in the years that followed.

ESP and CZ experienced the highest exceedances, suggesting that strong economic activity followed disturbance events. In contrast, DE and FIN showed moderated but sustained deviations, likely reflecting market adjustments such as increased salvage logging, changes in harvesting intensity, or pricing strategies to mitigate revenue losses. Finally, HR showed the lowest revenue exceedances, suggesting a more stable structure with less dependence on disturbance-driven harvesting.

The sensitivity analysis (Figure 4) tested how adjustments of the Ec.LTH by ±10 and ±50% influenced the classification of economic resilience across the five case studies. Results showed that moderate deviations (±10%) had minimal effect on the resilience status, as revenue trends generally remained within the defined Ec.UTH and lower bounds. In contrast, extreme adjustments (±50%) meaningfully affected the classification only in case where revenues were consistently near the lower threshold, such as HR and ESP, where minor declines pushed revenues below sustainability limits. The overall shape of the trajectories and distance between Ec.UTH and Ec.LTH remained stable, confirming that the threshold definition is robust under realistic profitability assumptions.

The Spearman correlation (Table 2) showed that revenue had a strong positive correlation with timber supply (HS.TS) in HR and a moderate correlation in CZ and FIN. Salvage logging only had a weak but significant correlation in HR. The percentage of salvage logging (Pct.HS.SL) showed significance only in HR, indicating this predictor had limited direct influence on revenue across most regions. For harvesting systems, the use of cut-to-length harvesting systems (HS.CTL) showed a strong positive correlation with revenue in the CZ and a moderate correlation in DE.

Regarding timber production variables, sawlog volume (TS.SP) displayed a strong positive correlation with revenue in HR and a moderate correlation in DE and FIN. Pulpwood volume (TP.PP) showed a significant correlation in HR. Market prices demonstrated significant correlations across several regions. Sawlog prices (MP.SP) showed a strong positive correlation in CZ and HR. Pulpwood prices (MP.PP) and Energy wood prices (MP.EP) were significantly correlated with revenue in HR, DE and ESP. For more detailed information about the correlation between the system variable, revenue, and the predictors, please go to Annex III.

The PCA results across the five CS reflect varying levels of integration between disturbance-driven harvesting, silvicultural practices and market dynamics, offering insight into how each region responds to economic incentives and ecological pressures. The first two principal components explained substantial variance in each country (CZ: 79.3%, HR: 77.5%, DE: 76.1%, FIN: 70.2%, ESP: 69.0%; Figure 5), allowing for robust interpretation of the underlying resilience mechanisms affecting revenue streams. A table summarizing the PCA loadings for each case study is provided in Annex IV to highlight the relative importance of each variable and to support the interpretation of the results.

According to the temporal patterns, CZ and ESP showed a transition from conventional forest management to a disturbance-driven system. Both countries exhibited a gradual shift where early years were market-oriented, while later years became increasingly dominated by salvage logging operations. Then, HR exhibits a divergent trajectory, some years strongly influenced by market prices while others are dominated by salvage logging, suggesting potentially competing management strategies. DE showed a cyclic pattern where salvage logging was important in the early years, followed by a regular period, and then intensification of salvage operations in the last years. Finally, FIN showed a distinct multi-phase pattern with an initial year’s focus on energy wood, followed by a conventional sawlog orientation, then product-oriented practices with increased salvage logging, and finally, intensified salvage operations.

Regarding the market dynamics and price relationships, CZ and ESP showed sawlog price (MP.SP) separated from pulpwood and energy wood prices (MP.PP and MP.EP) likely reflecting market saturation from salvage logging volume depressing sawlog values. Then, HR showed market prices clustering together but somewhat separated from timber production variables, indicating distinct market dynamics that maintained partial independence from disturbance-driven production changes- a potential buffer against revenue volatility. The DE case study has market prices clustered in one quadrant during middle years and reduced influence during intensified salvage periods. FIN maintained more coherent market price dynamics despite changing production patterns. Potentially due to well-established international trade relationships and industrial capacity to absorb varying product qualities.

The disturbance response mechanisms are different in each of the case studies, CZ showed disturbance-driven harvesting as the primary structuring force with tight integration between salvage operations and production systems; HS.SL, HS.TS and HS.CTL, cluster together, indicating a unified response to disturbances. The salvage logging (HS.SL) in HR aligned with sawlog volume (TP.SP), suggesting salvage operations focus on recovering sawlog material, while total timber supply (HS.TS) aligned with pulpwood and energy wood production (TP.PP and TP.EP). Then, DE showed early engagement with salvage-focused management, with variables like total timber supply (HS.TS) and cut to length (HS.CTL) closely aligned. Uniquely, FIN separated the total timber supply (HS.TS) from salvage logging (HS.SL), indicating different dynamics in how disturbances affect total supply versus focused salvage operations.

From the technological adaptation, CZ, ESP and DE showed strong contribution of cut-to-length (HS.CTL) with disturbance vectors, implying technological adaptation optimized for efficiency in degraded stands. The HR case study with the use of horses (HS.H) was correlated with sawlog prices.

Finally, all five case studies have experienced shifts toward disturbance-driven forest management, but with distinct trajectories and adaptation mechanisms. However, the economic implications of this convergence varied considerably in the last years: CZ showed an association with energy wood, indicating a shift toward lower-value products, but maintained revenue streams. The patterns in HR diverge and suggest regional or operational differences in economic adaptation strategies. Due to the longer-term engagement with disturbance management in DE, it showed a cyclic pattern rather than a single transformation. FIN maintains more separation between different product categories despite increasing disturbance influence, and ESP demonstrated a strong-salvage logging focus without clear product differentiation.

The two-fixed effect models were estimated to explain within-country variation in the revenue across the five European case studies from 2001 to 2021 (Table 3). Model A “Scale effects” (Table 4) explained 56% of within-country variation in revenue (R² = 0.562). All coefficients were positive and statistically significant (p < 0.01). Planned harvest volume had the strongest impact: each additional cubic meter increased annual revenue by about 42 USD (±10). A 10 percentage-point rise in the proportion of salvage logging increased revenue by approximately 189,000 USD, indicating that large disturbance events temporarily boost output and sales. Mechanization (use of cut-to-length) also raised revenue, adding about 50 USD per additional percentage-point increase. Price index showed a strong positive effect (≈ 0.5 million USD per unit change), confirming the central role of market conditions. The variance inflation factor (VIF) below 2 indicated no multicollinearity.

Model B “Composition effects” (Table 3) replacing total harvest with the volume of each assortment the explanatory power decline (R² = 0.46). Volumes of sawlogs, pulpwood and energy wood were not statistically significant once price and mechanization were controlled for (p > 0.10). Only the price index remained significant (β = 509403; p < 0.001), suggesting that short-term revenue fluctuations are driven mainly by market prices rather than shifts in harvest composition. VIFs (≤9) confirmed moderate correlation among product categories but acceptable model stability.

The log-scaled version of Model A (Table 4) yielded consistent results and interpretable elasticities: a 1% increase in planned harvest raised revenue by 0.22%, a 10 pp. increase in salvage share by ≈ 11%, and a one-standard deviation rise in the price by ≈ 30%. All effects remained significant at p < 0.01, confirming the robustness of the results. Then, across all specifications, harvest scale and market prices emerged as the dominant drivers of the system variable, revenue. Mechanization contributed positively, while the composition of harvest assortments showed limited influence. These results indicate that maintaining harvesting capacity and stable price levels is more critical to short-term economic resilience. Product composition plays a limited role once a total harvest, and price effects are controlled for.