The Threshold Effect of Overfishing on Global Fishery Outputs: International Evidence from a Sustainable Fishery Perspective

Abstract

Using global fishery outputs from 62 countries during the period 2001–2020, this study applies the threshold panel data model of Hansen (2000) to empirically explore the nonlinear relations between the number of fishing vessels, fishers, and fishery production, with distinct overfishing stocks (overfishing) and fish caught by trawling or dredging techniques (trawling) as the threshold variables. Our findings show that the levels of fishery outputs could be increased by different combinations of primary factors, such as the number of fishing vessels and fishers. More specifically, when the number of fishing vessels and fishermen operating in countries with a high ratio of overfishing (overfishing > 4.4456) increased, their fishery outputs significantly increased as compared with countries with a low percentage of overfishing. As overfishing activities increase, they can result in significant shortages and imbalances of fishery resources, directly influencing fish biomass, biodiversity, and sustainability of fisheries, as well as making destructive fishing practices worse for marine ecosystems. Countries with a relatively high proportion of trawler net usage (trawling > 4.5516) would have significantly lower fishery outputs, as more fishing vessels are invested in them, compared with countries with a low proportion of trawling net use. This implies that trawler net usage is a non-sustainable fishing technique and harms fishery resources. The policy implications of this study indicate that taxing overfished species and implementing trawling regulations could significantly improve species richness and site-based abundance, thus helping to sustain fishery outputs.

1. Introduction

The production of food from the ocean has a crucial position in supply, provides resources, and ensures food security for our society [1,2], but human population growth has contributed to a decline in biologically sustainable fishing stocks, approximately from over 89% to under 70% in recent years. Specifically, marine fish consumption increased by more than 60% between 1990 and 2018 [3]. Regardless of the scenario, the global primary production and presence of top predators are anticipated to decline significantly [4,5]. An ensemble of ecosystem models predicts for a long period from 2006 to 2100 claim that fish biomass will decrease by an average of 17% under high emission conditions [6]. The effects of overfishing vary significantly between species and geographical areas [7,8].

The term overfishing refers to situations where insufficient fishing and defaunation result in declining fish populations [9]. High overfishing damages community structures and fish sizes, resulting from the selective harvesting of species and the bycatching of other species, along with habitat modification as a result of changes in biomass, species composition, and size structure [10,11,12]. The FAO [13] reported that more than 85% of global fish stocks are either overfished or fully exploited, and overfishing or lower biomass limits are responsible for the 40% reduction in Europe’s fish stocks [14,15]. Several fisheries in the European Union (EU) have a current fleet size and capacity that exceed sustainable levels by two to three times [16].

Overfishing reduces the adaptability of fish stocks and aquatic resources to climate change [15,17]. Moreover, overfishing is directly related to a variety of detrimental fishing practices, including illegal, unreported and unregulated (IUU) fishing, bycatching, and harmful subsidies [18,19]. Overfishing could occur if harmful fishing techniques such as bottom trawling are continued, as this destroys the ocean’s seafloor and affects both some species that are specifically targeted and those that are not [20,21]. Furthermore, harmful subsidies lead to overfishing by supporting fleets with excessive numbers, resources, and technology [18,22].

Therefore, the research objective of this study is to empirically investigate the threshold impacts of overfishing stocks (SDG14 overfishing) and fish caught by trawling or dredging techniques (SDG14 trawling) on global fishing outputs, by controlling for the number of fishing vessels, number of fishers, and national income.

2. Materials and Methods

Once fisheries are unable to keep up with the growing demand for fishery products, aquaculture has expanded and now exceeds fish production [13]. The more prominent role of aquaculture has resulted in more fish production, competition between wild and aquaculture fish production, and increased trade. The high demand for seafood products inadvertently creates opportunities for fishermen to overfish, due to the high economic value associated with maximizing the discounted value of present and future net benefits [23]. Overall, overfishing actions can cause a significant reduction in fishery outputs.

2.1. Study Area

In this study, we used a global fishing output database, covering 2001 to 2020, derived from the fisheries’ statistical yearbooks published by the FAO website https://www.fao.org/fishery/en/statistics, accessed on 17 August 2022. We have also collected data on sustainable development goal scores (SDG14) from the website https://dashboards.sdgindex.org/rankings, accessed on 17 August 2022. Finally, we obtained the GDP annual growth rate and poverty rate as control variables provided by the World Bank statistics (available website: https://databank.worldbank.org/source/world-development-indicators, accessed on 17 August 2022). We then aggregated the data to enable us to analyze the more specific target levels, which include the global fishery outputs and the goals of SDG14 (fish grabbed from overfishing or tumbled stocks and fish grabbed by trawling or dredging).

This study primarily focuses on the global fishery outputs from 2001 to 2020 using the data from 62 countries (see Figure 1). Overall, the fishery outputs caught in Asia fluctuated from 1,504,057 to 50,100,000 (tons of live weight). As shown in Figure 1, we found that Chinese fishermen caught an average of 50,100,000 tons of live fish each year, followed by Indonesia (8,326,168 tons), India (4,424,787 tons), and Vietnam (2,658,570 tons). There is an interesting difference between the fishery production in the Americas and Asia during the period 2001–2020. Asia region grabbed almost as much freshwater and anadromous fish, ranging from 18,314,993 to 47,260,766 tons live weight, in comparison to the Americas region’s 1,250,587 and 1,516,139 tons live weight.

Overfishing’s impact on global fishery outputs. Assemblage overfishing, a common phenomenon, refers to fishing down marine (or aquatic) food webs [10,24]. The phenomenon of “fishing down” is when fisheries target large fish in an ecosystem, and their abundance is reduced relative to that of smaller fish [25]. Additionally, overfishing has become increasingly evident, causing indirect changes in habitat structure and processes [26,27,28,29]. Overfishing is a longer-term issue that results from an imbalance between the resources used for harvesting (the resulting current level of output) and the resources required (corresponding output) for harvesting at an “optimal” scale. Overcapacity and excessive fishing inputs have been a leading concern around the world since the recent debate about managing fishing capacity began. According to the Code of Conduct for Responsible Fisheries proposed by The Food and Agriculture Organization (FAO), excessive fishing capacity threatens fisheries’ resources and their ability to sustainably provide catches as well as benefits to fishers and consumers. However, the FAO reported in Article 6.3 section and cautioned that “Nations could take steps to avoid overfishing and excess fishing capacity and implement management measures to ensure that fishing effort is adequate for the productive capacity of fishery resources and their long-term consumption.”

Overfishing and habitat destruction have depleted one-third of the world’s fish stocks [17]. Overexploitation has had long-term effects on marine ecosystems and poses severe risks to ocean health, such as biodiversity, food security, and coastal livelihoods and economies [17,30,31,32]. Overfishing frequently has significant environmental impacts [33,34], and has even been recognized as a driver of shifts in ecosystem regimes [35]. Overfishing directly impacts fish biomass, biodiversity, and the sustainability of fisheries, as well as exacerbating destructive fishing approaches’ impact on marine ecosystems (e.g., bottom trawlers). As a result of illegal, unreported, or unregulated fishing, overfishing is usually performed with high-impact fishing gear, such as bottom trawlers, which negatively affect the benthic substrate [36]. Overfishing activities over time can lead to a significant scarcity of aquaculture resources. Evidence shows that up to 70% of fish stocks in European waters are currently overfished, or at their lower biomass limits [37]. While some encouraging actions are being taken, several regulations have been implemented to control overexploitation. The buyback of fishing vessels, or the setting of tax rates by species in proportion to the extent of overexploitation, have been performed [38]. The level of tax related to overfished species should be determined by how much effort is being focused on each species—more severely overfished species will incur a higher rate [38]. Taxing overexploited fish species would increase the price for consumers, resulting in a reduction in fish consumption. In response to lower demand, processors and wholesalers offer fishers lower prices upon landing, resulting in lower fishing efforts.

Trawling or dredging. Trawling accounts for approximately 25% of the global capture of fishing [39], and it is increasingly being considered as an unsustainable fishing technique [40,41,42]. It has two impacts on benthic ecosystems. First, by disrupting epibenthic sediments, fishing gear causes losses of habitat complexity and the resuspension of sediments into the water column [43,44,45], leading to lower organic matter contents in sediments [46]. The second consequence of trawling is that it disrupts the structure of benthic communities by selectively removing large-bodied species, which results in small r-selected species dominating in benthic communities [21]. The management of fisheries and the preservation of aquatic life have long been affected by bottom trawling (BOT) [47,48]. BOT results in bycatching and the discarding of non-target species, such as endangered or protected species [49,50]. From 1950 to 2014, BOT accounted for 23% of landed marine catches and 60% of fisheries’ discards globally [51]. However, Roda et al. [52] showed that this percentage has dropped to 46% in recent years (2010–2014). Since most of its effects are felt in coastal waterways, pressure is being placed on those ecosystems [53,54]. A few countries prohibit bottom trawling because of the harm it causes to marine life [55,56]. For example, since 2013, Chinese policymakers have undertaken several steps to safeguard marine fisheries and tighten regulations on bottom trawling, such as prohibiting the authorization and production of bottom trawlers [56]. Additionally, the authorities of the Hong Kong Special Administrative Region also banned trawling to aid in the recovery of fishery resources and associated benthic ecosystems. Their findings indicate that a trawl ban can significantly improve species richness and site-based abundance, which implies that it is an efficient approach for conserving biodiversity in tropical coastal seas [20].

Sustainable development goal 14 (SDG14). The primary target of SDG14 is to protect and utilize the oceans, seas, and marine ecosystems sustainably. To accomplish this, all maritime stakeholders must participate in the various aspects of SDG14. Most efforts have been directed towards the first four targets: sustainable management, marine pollution, sustainable fishing, and ocean acidification. It has been more difficult to eliminate fishery subsidies, as stated in target 14.6.

SDG—Sustainable management. Sustainable development goal (SDG) 14.2 aspires to effectively manage and safeguard marine and coastal habitats, increase their resilience, reduce harmful effects, and attain productive and healthy oceans (United Nations, 2018). Management plans that consider all human activities are necessary if we are to achieve this aim. However, all these steps are administered by different entities, leading to extremely fragmented ocean management [57,58,59]. The project manager of marine resources is a critical contributor to ocean management. Most of these resources are managed via a specific species approach, which neglects interspecies interactions, habitat influences, and ecosystem dynamics, leading to overfishing, habitat damage, and changes in the aquatic community [60]. Integrating ecosystem-based management (EBM) into fisheries management is one efficient way to achieve comprehensive management [61,62,63]. Executing ecosystem-based management is a long-term goal that requires much more effort be applied at different levels. The fishing industry could provide a valuable input by participating in EBM discussions and decision making [64]. Bycatch and other effects of fishing represent additional avenues of engaging with this goal. Generally, the involvement in environmental management of fishing enterprises will be critical to restoring marine ecology and reducing overfishing [65]. While EBM has several benefits, implementation has been relatively slow due, primarily, to training challenges [62].

SDG—Subsidies. Subsidies aggravate the problem of overexploitation and fleet overcapacity. They are monetary payments made by the government to private recipients in the fishing industry [66,67]. According to Sumaila et al. [68] and Da-Rocha et al. [69], subsidized overcapacities can harm the marine environment in the long run, and directly impact its productivity. Sustainable Development Goal 14.6 aims to prevent subsidies that significantly exacerbate these problems (United Nations 2018). However, eliminating the use of harmful subsidies in the fishing industry is made complicated by political and social concerns. Thus, if the management institution is to address this issue, it may require the careful consideration of financing for income supplements and transitional measures to minimize immediate losses [70]. Research by Da-Rocha, García-Cutrín, Prellezo and Sempere [69] demonstrated that cutting subsidies leads to lower societal costs than anticipated. Diversifying and governing days at sea are two methods for reducing the number of vessels, a common strategy that the government could use to reduce harms and prevent overfishing [71].

2.2. Threshold Panel Data Model

The study examined the impact of overfishing on global fishery outputs, and the dependent variable is the fishing outputs. In order to reduce heteroscedasticity, logarithmic processing was performed, which referred to prior study [72]. Fishery outputs can be used to reflect the intensity of fishery resources’ development, and will be affected by many factors, such as the number of fishers, and sustainable development goal scores for ocean-related issues (SDG14).

The independent variables in this study are calculated by the logarithm of the number of fishing vessels (LogNFV) and the number of fishers (LogNF). In terms of control variables, we have utilized a dummy income variable (high income, lower-middle income, and upper-middle income). These data were obtained from the fisheries’ statistical yearbooks published by the FAO with time series data from 2001 to 2020. All variables are reported clearly in Appendix A.

Remarkably, the threshold panel data model put forward by Hansen (2000) was combined with the Log of SDG14 overfishing (the proportion of grabbed from overfishing or tumbled stocks (% of total catch)) and SDG14 trawling (the proportion of fish grabbed by trawling or dredging) as the threshold variables to evaluate the function parameters of fishery outputs. The single-regime threshold panel data model for cross-country panel data is represented in Equation (1), where k_i stands for the threshold variables, λ₁* stands for the optimal threshold value, and f(.) is an indicator function. If the relationship in (.) is true, then f(.) equals 1; otherwise, f(.) is 0. The parameters α₁, α₂, γ₁, γ₂, θ₁, θ₂, ξ₁, and ξ₂ are to be estimated:

$$\begin{array}{cc}\hfill LogGF{O}_{i,t}=& \left[{\alpha }_1{\left(Technological progress\right)}_{i,t}+{\gamma }_{1}LogNF{V}_{i,t}+{\theta }_{1}LogN{F}_{i,t}+{\xi }_{1}Incom{e}_{i,t}\right]\times f\left(k_{i,t}\le {\lambda }_1^{*}\right)\hfill \\ \hfill +& \left[{\alpha }_2{\left(Technological progress\right)}_{i,t}+{\gamma }_{2}LogNF{V}_{i,t}+{\theta }_{2}LogN{F}_{i,t}+{\xi }_{2}Incom{e}_{i,t}\right]\times f\left(k_{i,t}>{\lambda }_1^{*}\right)\hfill \end{array}$$

(1)

where Log refers to the logarithm; GFO_i is global fishing outputs as a proxy for fishing production performance (tons of live weight); (Technological progress)_i,t presents time trend, represented as by the years between 2001 and 2020; NFV_i,t is the global number of fishing vessels (motorized vessels propelled by engines); NF_i,t is the global number of fishers; and Income_i is categorized into distinct levels (HIC = high income, LMIC = lower-middle income, UMIC = upper-middle income).

In this equation (Equation (1)), the threshold parameter λ (SDG14 overfishing and SDG14 trawling) as sample-splitting (or threshold) variables. The impact of technological progress, the number of fishing vessels, fishers, and income level on global fishery outputs will be α₁ and α₂ for countries with a low or high regime, respectively. It is obvious that under the hypothesis, α₁ = α₂; γ₁ = γ₂; θ₁ = θ₂; and ξ₁ = ξ₂ in the model.

Our estimate procedure compares the threshold model in Equation (1) to the linearity null hypothesis, H₀: α₁ = α₂. The fact that the threshold parameter λ was not detected when the null hypothesis was kept caused a non-standard inference concern. Both Wald and Lagrange multiplier tests failed to follow their typical chi-square patterns [73,74]. Instead, conclusions are reached by generating Wald or Lagrange multiplier (score) statistics for each possible value of λ, and then making inferences based on the sum of the Wald or Lagrange multiplier potential λ_s. This statistic’s limiting distribution is non-standard and relies on several model-specific nuisance factors. Instead of tabulations, inferences are produced using a bootstrap model whose validity and features have been demonstrated by Hansen (1996). The slope parameters’ estimations come naturally as β(λ) and γ(λ) after an estimate of λ has been acquired (as the minimizer of the residual sum of squares computed over all possible values of λ). The failure of the test demonstrates that the threshold effect exists. Secondly, for robustness, we also follow the procedure proposed by Hansen [74] for investigating the possibility of multiple threshold effects. We posit the detailed hypotheses: the single threshold corresponds to the linear and single-threshold models (hypotheses H₀ and H_a). In contrast, the double threshold equals single and double-threshold models (including hypothesis H₀ and H_a), respectively, and so forth.

3. Results

3.1. Preliminary Analysis

The global fishery outputs in tons of live weight are reported in

Table 1. The descriptive statistic variables by year, region, and country from 2001 to 2020.

Panel A: The mean value of variables by year.
Year		Observations		Global Fishery Outputs		Overfishing		Trawling		NFV (Number of Fishing Vessels)		NF (Number of Fishers)		GDP Growth Rate		Poverty Rate
2001		62		716,897		74.33		80.18		39,653		527,542		47.67		18.07
2002		62		766,203		74.29		79.81		40,292		528,153		48.79		18.02
2003		62		812,125		73.33		80.74		40,124		537,879		49.20		17.68
2004		62		881,533		71.94		80.77		38,535		557,982		50.77		17.23
2005		62		935,467		70.31		81.88		40,880		541,681		50.18		16.30
2006		62		995,938		67.93		80.98		40,030		564,652		50.87		16.02
2007		62		1,048,080		66.17		81.49		40,641		530,685		50.71		15.61
2008		62		1,109,958		65.92		81.28		41,920		543,191		48.49		14.92
2009		62		1,168,047		65.57		81.42		42,913		558,465		45.21		14.16
2010		62		1,234,563		65.22		81.57		41,097		571,106		49.76		13.17
2011		62		1,292,839		65.76		82.09		41,275		577,512		49.25		12.87
2012		62		1,397,460		65.65		82.94		40,776		563,890		48.73		11.30
2013		62		1,505,387		64.97		81.64		38,495		561,428		48.67		11.13
2014		62		1,578,629		65.77		82.22		42,363		579,659		48.72		11.31
2015		62		1,647,904		68.16		83.04		41,624		579,574		48.25		10.64
2016		62		1,715,094		69.02		82.19		41,379		576,410		48.06		10.96
2017		62		1,778,825		69.82		81.87		41,824		577,776		48.60		10.49
2018		62		1,836,535		70.07		81.75		40,657		576,055		48.20		9.75
2019		62		1,897,315		69.24		81.69		39,214		572,649		47.70		10.15
2020		62		1,941,143		68.92		81.77		39,721		559,559		41.49		9.92
Panel B: The mean value of variables by world region.
Region		Observations		Global Fishery Outputs		Overfishing		Trawling		NFV (Number of Fishing Vessels)		NF (Number of Fishers)		GDP Growth Rate		Poverty Rate
Africa		280		29,353		73.61		87.90		9150		191,327		50.92		39.15
Europe and Central Asia		40		74,129		60.04		99.91		1004		85,834		50.13		5.42
South East Asia		260		5,535,225		81.11		72.97		120,028		2,199,454		51.71		12.15
Latin America and the Caribbean (LAC)		140		120,545		64.97		95.24		13,525		157,712		48.77		6.83
The Middle East and North Africa (MENA)		80		297,380		81.30		72.63		29,505		211,453		49.49		4.08
OECD		420		326,311		56.35		76.95		29,423		49,295		44.28		0.68
Oceania		20		2829		86.03		105.00		514		248,000		50.09		40.90
		Note: GFO is the global fishing outputs (tons live weight); Overfishing is the fish caught from overexploited or collapsed stocks (% of the total catch); Trawling is the fish caught by trawling or dredging (% of fish caught by trawling); NFV is the global number of fishing vessels (number of motorized vessels propelled by engines); NF is the global number of fishers (number); GDP growth rate is the gross domestic product annual growth rate (annual %); Poverty rate is the proportion of poverty headcount at national poverty lines (% of population). All variables are reported clearly in Appendix A.
Panel C: The mean values of variables by country.
Country	Observations	Global Fishery Outputs	Over-Fishing	Trawling	NFV (Number of Fishing Vessels)	NF (Number of Fishers)	GDP Growth Rate	Poverty Rate	Country	Observations.	Global Fishery Outputs	Over-Fishing	Trawling	NFV (Number of Fishing Vessels)	NF (Number of Fishers)	GDP Growth Rate	Poverty Rate
Angola	20	553	88.52	68.65	2467	84,676	51.17	17.08	Mozambique	20	1735	83.66	105.00	1096	224,555	52.49	70.95
Argentina	20	2727	58.79	53.11	1490	19,887	47.61	2.34	Myanmar	20	748,168	85.57	54.63	15,090	2,909,199	55.47	3.47
Bangladesh	20	1,525,550	98.27	90.67	27,206	1,500,570	52.07	19.35	Namibia	20	426	75.59	68.42	218	19,906	49.27	22.60
Belize	20	5147	44.69	104.95	586	3074	48.08	5.51	Netherlands	20	55,836	48.22	75.79	840	2526	47.24	0.10
Brazil	20	419,035	81.80	91.02	31,591	848,764	48.09	6.83	New Zealand	20	103,553	51.07	58.47	1420	1847	48.71	0.15
Cambodia	20	110,637	69.95	33.11	56,417	501,575	53.17	33.95	Nigeria	20	184,074	85.72	96.20	36,779	1,122,825	51.40	44.92
Cameroon	20	1235	48.25	96.41	8669	173,570	50.10	25.99	Norway	20	1,034,438	76.66	82.06	6619	12,746	47.54	0.19
Canada	20	169,108	55.17	80.02	19,795	41,199	48.28	0.22	Oman	20	333	97.55	98.19	15,089	42,455	48.88	10.58
Chile	20	962,062	60.00	104.94	13,218	53,883	49.33	2.07	Pakistan	20	127,439	62.93	105.00	13,502	371,384	50.09	13.94
China	20	50,100,000	81.34	55.21	592,406	8,709,106	54.78	12.67	Panama	20	7495	52.61	104.84	8463	36,459	50.95	4.84
Congo, DP	20	3072	43.64	105.00	19,424	275,254	51.28	81.69	Papua New Guinea	20	2829	86.03	105.00	514	248,000	50.09	40.90
Denmark	20	37,026	59.95	66.71	2686	2300	47.21	0.21	Peru	20	73,430	94.71	104.63	4181	85,841	50.17	9.56
Ecuador	20	311,597	64.85	105.00	16,444	63,653	49.00	7.37	Philippines	20	2,149,586	83.65	97.81	183,824	1,927,922	50.86	10.39
Egypt	20	937,234	66.93	52.32	4583	139,358	50.33	2.30	Poland	20	36,934	38.38	66.82	839	4534	49.56	0.60
France	20	213,387	74.21	76.70	5845	15,120	46.93	0.29	Portugal	20	9526	22.01	68.65	7178	17,423	46.38	0.43
Georgia	20	999	49.90	98.97	26	2200	51.05	10.51	Russian Federati	20	147,259	70.18	100.86	1982	169,469	49.21	0.33
Germany	20	40,757	45.51	88.83	1660	2954	47.08	0.01	Senegal	20	534	63.50	87.15	7199	65,597	50.09	36.97
Ghana	20	25,459	63.04	97.16	12,689	234,408	51.99	34.84	Sierra Leone	20	62	98.25	88.09	7442	63,570	51.58	46.98
Guinea	20	315	90.71	82.08	1398	33,406	50.68	34.37	South Africa	20	6294	61.19	80.89	1784	17,748	48.21	22.13
Iceland	20	11,300	50.24	85.72	1731	4633	48.68	0.05	Spain	20	282,526	54.54	59.32	10,357	32,710	47.13	0.81
India	20	4,424,787	94.49	53.51	118,084	7,521,681	52.01	27.82	Sri Lanka	20	17,187	84.94	105.00	25,568	223,890	50.73	5.05
Indonesia	20	8,326,168	81.64	63.16	386,143	3,609,297	50.99	17.52	Sweden	20	10,154	65.13	77.89	1382	1646	48.03	0.48
Iran	20	250,713	70.38	103.38	79,420	552,670	48.96	0.94	Taiwan	20	315,074	81.97	87.39	21,872	243,565	49.68	0.46
Ireland	20	46,859	58.59	91.89	1764	5547	11.22	0.26	Tanzania	20	8592	80.33	105.00	14,149	174,718	52.29	48.82
Italy	20	158,380	37.03	50.60	12,029	30,454	45.85	1.11	Thailand	20	1,107,688	52.24	88.24	18,243	182,686	49.53	2.64
Japan	20	1,140,500	51.09	85.39	266,099	196,766	46.56	0.27	Turkey	20	196,146	56.96	73.39	17,813	62,021	11.84	0.98
Korea, Rep.	20	1,504,057	71.60	67.56	75,352	145,059	49.80	0.22	Uganda	20	70,290	60.96	76.21	6464	80,487	52.31	50.07
Malaysia	20	381,815	79.55	72.61	44,522	120,701	50.43	0.61	United Kingdom	20	199,784	60.65	78.95	6504	12,389	47.26	0.26
Mauritania	20	108,299	87.13	74.29	8328	107,849	49.99	10.68	United States	20	477,764	68.92	82.05	76,140	149,443	47.79	1.00
Mexico	20	162,436	77.41	94.20	88,624	239,990	47.50	4.52	Venezuela	20	24,384	57.30	103.13	31,922	46,306	47.53	11.39
Morocco	20	1240	90.34	36.62	18,929	111,330	49.80	2.52	Vietnam	20	2,658,570	97.92	42.25	57,484	771,327	52.48	10.06

Note: GFO is the global fishing outputs (tons live weight); Overfishing is the fish caught from overexploited or collapsed stocks (% of the total catch); Trawling is the fish caught by trawling or dredging (% of fish caught by trawling); NFV is the global number of fishing vessels (number of motorized vessels propelled by engines); NF is the global number of fishers (number); GDP growth rate is the gross domestic product annual growth rate (annual %); Poverty rate is the proportion of poverty headcount at national poverty lines (% of population). All variables are reported clearly in Appendix A.

. Panel A of Table 1 demonstrates the mean value of variables by year regarding the global fishery outputs, the global number of fishing vessels, the number of fishers, SDG14 overfishing, SDG14 trawling, GDP growth rate, and poverty ratio for the years 2001 to 2020. Specifically, the average global fishery outputs ranged from 716,897 to 1,941,143 (tons of live weight). Connecting Panel C, we found that China’s highest average global fishing output was 50,100,000 (tons live weight), followed by Indonesia, India, and Vietnam with the mean value of 8,326,168, 4,424,787, and 2,658,570 (tons live weight), correspondingly. All remaining variables are shown in the details of Panel A and C. Panel B demonstrates the mean value of all variables by region, the global fishing outputs in the Southeast Asia region have a mean value of 5,535,225 tons live weight, which is the highest region in comparison to Africa and Oceania regions of 29,353 and 2829 (tons live weight), respectively. Interestingly, the percentage of overfishing in Oceania, the Middle East, North Africa, and South East Asia is higher than in the other regions. At the same time, Oceania, Europe, and Central Asia have the most remarkable rate of fishing by trawling and dredging. Other variables are clearly described in Panel B of Table 1.

We present the groups of species of the global fishery outputs by continent from 2001 to 2020 (see Appendix B: Groups of species from FAO by continent), including Aquatic Animals NEI, Aquatic Plants, Crustaceans, Demersal Marine Fish, Freshwater and Diadromous Fish, Marine Fish NEI, Molluscs excluding Cephalopods, and Pelagic Marine Fish. Specifically, fishery outputs in Asia almost related to Freshwater and Diadromous Fish (ranging from 18,314,993 to 47,260,766, tons live weight), in comparison to the Americas region (between 1,250,587 and 1,516,139, tons live weight) in grabbing Freshwater and Diadromous Fish. Furthermore, Demersal Marine Fish species range from 100,753 to 453,845 in Africa where European region shows fishery outputs between 116,212 and 205,799 (tons live weight). In terms of Aquatic Plants species and Molluscs without Cephalopods, the fishery outputs in Asia range from 10,781,141 to 34,916,316 (tons live weight) and between 9,219,565 and 16,158,709 (tons live weight).

Moreover, the species groups of worldwide fishing outputs by income classification based on the World Bank are shown in Appendix C. It found that lower-middle-income nations yield more freshwater and diadromous fish, ranging from 4,675,907 to 22,060,390 tons, compared to upper-middle-income nations’ 14,044,467 to 27,839,930 tons. According to Mollusks excluding Cephalopods, the value of upper-middle-income nations (8,350,940–15,231,576 tons) shows substantially higher than that of low-income countries (60,000–62,420 tons). Similarly, the outputs of Aquatic Plants in upper-middle-income countries present the highest in other countries, ranging from 8,423,150 to 21,075,219 tons.

Figure 2a shows that the nonlinear effects of overfishing (SDG14 overfishing) negatively impact global fishing outputs. Similarly, the results of trawling net (SDG14 trawling) negatively impact fishing outputs, which indicates that overfishing and trawling activities can lead to a decrease in the global fishery outputs (see Figure 2b).

3.2. Summary Descriptive Statistics

Table 2. Descriptive statistics of 62 countries, 2001–2020.

Variables	Measurement Unit	Obs.	Mean	Std Dev.	Minimum	Maximum
Log of global fishing outputs (GFO)	Tons live weight	1240	10.754	3.029	0.693	18.071
Log of global number of fishing vessels (NFV)	Number of motorized vessels propelled by engines	1240	9.047	1.867	3.258	13.453
Log of global number of fishers (NF)	Number of fishers	1240	11.184	2.082	5.298	16.182
Log of SDG14_ overfishing	% of total catch	1240	4.171	0.366	2.194	4.604
Log of SDG14_trawling	% of fish caught by trawling	1240	4.361	0.312	1.609	4.654
High income (HIC)	Dummy variable	1240	0.323	0.468	0.000	1.000
Lower-middle income (LMIC)	Dummy variable	1240	0.339	0.473	0.000	1.000
Upper-middle income (UMIC)	Dummy variable	1240	0.242	0.428	0.000	1.000
Technological progress (Time trend)	Year from 2001–2020	1240	10.5	5.768	1.000	20.000
Log of GDP growth rate	% of annual	1240	3.851	0.319	0.270	4.284
Log of Poverty rate	% of population	1240	2.024	1.192	0.693	4.538

Note: Sample countries include Bangladesh, Sierra Leone, Vietnam, Oman, Peru, India, Guinea, Morocco, Angola, Mauritania, Papua New Guinea, Nigeria, Myanmar, Sri Lanka, Mozambique, Philippines, Taiwan, Brazil, Indonesia, China, Tanzania, Malaysia, Mexico, Norway, Namibia, France, South Korea, Iran, Russian Federation, Cambodia, United States, Egypt, Arab Rep., Sweden, Ecuador, Senegal, Ghana, Pakistan, South Africa, Uganda, United Kingdom, Chile, Denmark, Argentina, Ireland, Venezuela, Turkey, Canada, Spain, Panama, Thailand, Japan, New Zealand, Iceland, Georgia, Cameroon, Netherlands, Germany, Belize, Congo, Dem. Rep., Poland, Italy, and Portugal.

demonstrates the descriptive statistics about the global fishing outputs, the global number of fishing vessels, the number of fishers, SDG14 overfishing, and SDG14 trawling for the years 2001 to 2020. Specifically, the average global fishing outputs were 10.754 (1,312,997 tons of live weight). We can see that China’s highest average global fishing output was 50,100,000 (tons live weight), followed by Indonesia with a mean value of 8,326,168 (tons live weight). The mean global number of fishing vessels was 9.047 (40,671 motorized vessels propelled by engines), while the average global number of fishers was 11.184 (559,292 number of fishers).

Regarding the sustainable development goals, the average value for SDG14 overfishing in the period 2001 to 2020 was 4.171 (68.619% of total catch). The countries with the highest proportions of total catch fish stocks were Bangladesh and Sierra Leone, with percentages of 98.273 and 98.253, respectively. Furthermore, the average SDG14 trawling was 4.361 (81.566% of fish caught by trawling).

The correlation matrix analysis for the variables employed in the analysis is presented in

Table 3. Pearson correlation results.

Variables	GFO	NFV	NF	SDG14 Overfishing	SDG14 Trawling	HIC	LMIC	UMIC	GDP Growth Rate	Poverty Rate
GFO	1
NFV	0.601	1
NF	0.419	0.720	1
SDG14 Overfishing	−0.563	0.216	0.437	1
SDG14 Trawling	−0.205	−0.151	−0.822	−0.138	1
HIC	0.139	−0.981	−0.549	−0.363	−0.706	1
LMIC	0.204	0.193	0.484	0.304	−0.137	−0.494	1
UMIC	0.336	−0.538	0.201	−0.0949	0.129	−0.390	−0.404	1
GDP growth rate	−0.305	0.678	0.190	0.124	−0.331	−0.144	0.179	−0.100	1.000
Poverty rate	−0.409	0.170	0.381	0.296	0.149	−0.620	0.406	−0.476	0.228	1

Note: This table represents the Pearson correlation coefficients of major variables for the full sample of countries. The results of primary Pearson correlations are significant at the 5 percent level or higher, excluding the Pearson correlations in italics. GFO = global fishing outputs; NFV = global number of fishing vessels; NF = global number of fishers; SDG14 Overfishing = sustainable development goals—fish stocks; SDG14 Trawling = sustainable development goals—trawling or dredging; HIC = high income; LMIC = low-middle income; UMIC = upper-middle income, GDP growth rate = the gross domestic product annual growth rate, and Poverty rate = poverty headcount ratio at national poverty lines.

. The table demonstrates that the threshold groups (SDG14 overfishing; SDG14 trawling) are negatively associated with global fishery outputs. For example, the correlation between global fishery output and SDG14 overfishing is −0.563, whereas that between global fishery outputs and sustainable development SDG14 trawling is −0.205. In addition, the remaining variables (fishery fleet; global number of fishers; income levels) are positively correlated, whereby the correlation between global fishery outputs and number of fishing vessels is 0.601. The Pearson correlation result between SDG14 overfishing and upper-middle income is negative but insignificant, with a value of −0.0949, while at the same time, SDG14 overfishing is strongly positive with low-middle income of 0.304, GDP growth rate of 0.124, and the poverty rate of 0.296.

3.3. The Threshold Effect of Overfishing on Global Fishery Outputs

Table 4. Threshold estimates of using different threshold variables.

Sample Split	SDG14 Overfishing	SDG14 Trawling
	Model 1	Model 2
The Lagrange multiplier test for no threshold	139.600	118.718
Bootstrap p-value	0.0004	0.0001
Threshold estimate	4.4456	4.5516
95% Confidence Interval	(4.396, 4.455)	(4.5077, 4.5523)

displays the results of formula (1) using two distinct SDG14 threshold variables: overfishing and trawling. The bootstrap method with 5000 replications and a 15%-trimmed mean was utilized to estimate the p-value, which was used to assess the statistical significance of the threshold value. The bootstrap p-values demonstrate that the test of the no-threshold effect can be rejected in all models. Thus, we divided the sample into two distinct regimes. For example, in models 1 and 2, logarithm (NFV) is the global number of fishing vessels (motorized vessels propelled by engines) and logarithm (NF) is the global number of fishers, and the empirical results favor a threshold model. Based on model 1, an overfishing threshold of 4.4456 is estimated, with a 95 percent confidence level of 4.396–4.455. The finding implies that countries categorized as low-SDG14 overfishing clusters (i.e., having a low percentage of fish caught from overexploited groups) are those with threshold values of less than 4.4456, while those classified as high-SDG14 overfishing clusters are those with threshold values greater than 4.4456 (high percentage of fish caught from overexploited groups). Additionally, the threshold value for SDG14 trawling in model 2 has an estimate of 4.5516 and a 95% confidence range from 4.5077 to 4.5523. The finding implies that nations with threshold values below 4.5516 are categorized as belonging to the low-SDG14-trawling clusters (i.e., having a low percentage of fish captured by trawling or dredging), whereas nations with higher values are classified as belonging to the high-SDG14-trawling clusters (with a high rate of fish caught by trawling or dredging).

In

Table 5. Regression results using different threshold variables. Dependent variable: global fishery outputs.

Variables	Panel GLS without threshold	Threshold Model 1 = SDG14 Overfishing		Threshold Model 2 = SDG14 Trawling
		Regime 1	Regime 2	Regime 1	Regime 2
		Overfishing ≤ 4.4456	Overfishing > 4.4456	Trawling ≤ 4.5516	Trawling > 4.5516
Constant	−4.402 **	−2.548 **	−10.46 **	−5.106 **	−2.981 **
	[−9.150]	[−5.332]	[−10.524]	[−14.489]	[−4.869]
Logarithm (NFV)	0.552 **	0.412 **	0.776 **	0.728 **	0.283 **
	[10.023]	[7.501]	[7.399]	[18.473]	[7.170]
Logarithm (NF)	0.605 **	0.623 **	0.927 **	0.576 **	0.599 **
	[9.967]	[10.237]	[7.786]	[16.919]	[9.160]
High income (HIC)	4.557 **	4.017 **	−0.358	4.400 **	4.515 **
	[17.689]	[14.543]	[−0.596]	[13.925]	[12.504]
Lower-middle income (LMIC)	1.755 **	0.951 **	1.634 **	1.932 **	1.985 **
	[7.867]	[3.814]	[4.385]	[5.944]	[12.962]
Upper-middle income (UMIC)	3.031 **	2.221 **	3.400 **	2.515 **	3.513 **
	[13.079]	[8.872]	[7.913]	[8.040]	[22.514]
Technological progress (Time trend)	0.0574 **	0.0659 **	0.0255	0.0435 **	0.0809 **
	[5.482]	[6.554]	[1.184]	[10.515]	[8.786]
Observationss	1240	905	335	829	411
Wald χ2	7848 **	4489 **	3731 **	6797 **	1186 **
Prob > χ2	0.0000	0.0000	0.0000	0.0000	0.0000

Notes: the t-statistics results are shown in square parentheses (white-corrected for heteroscedasticity). The findings correspond to a 15%-trimmed mean. ** indicate 1% level of statistical significance, respectively.

, 1240 observations are grouped into two regimes. The first regime (SDG14 overfishing) gave 905 observations, and the second included the following 335 observations. The Lagrange multiplier test for no threshold was rejected. The overfishing threshold variable shows nonlinear statistically significant correlations with overall fishing outputs. Furthermore, to gauge the threshold effect, we presume that the single threshold corresponds to the linear and single-threshold models (hypotheses H₀ and H_a). In contrast, the double threshold equals single and double-threshold models (including hypothesis H₀ and H_a), respectively. The results for regime 1 (SDG14 overfishing ≤ 4.4456) show that with a higher number of fishing vessels and fishers (coefficient value of 0.412 and 0.623), global fishery outputs may significantly increase. Similarly, the income level and time period are also significantly positively related with fishing production. Specifically, the coefficients for income levels are as follows: high income = 4.017, lower-middle income = 0.951, upper-middle income = 2.221. The results for Regime 2 (SDG14 overfishing > 4.4456), on the other hand, indicate that with a higher number of fishing vessels, the number of fishers may significantly increase global fishery outputs. This finding maintains the sense, but the coefficient values of NFV (0.776) and NF (0.927) are significantly higher than in regime 1. Furthermore, the effects of high income level and time are nonexistent when SDG14 overfishing > 4.4456.

Regarding the result of the SDG14 trawling threshold estimation, there were 829 observations in regime 1 and 411 observations in regime 2. Similarly, the Lagrange multiplier (LM) test for no threshold thus rejected the null hypothesis. In other words, the global fishery outputs and the trawling threshold variable exhibit statistically significant nonlinear correlations. The results remains the same when using SDG14 overfishing. More precisely, a higher fishery fleet threshold (SDG14 trawling ≤ 4.5516) significantly increases the global fishery product (coefficient value of the number of fishing vessels = 0.728) compared with lower thresholds (SDG14 trawling > 4.5516; the coefficient value of the number of fishing vessels = 0.283). In contrast, the number of fishers with the threshold shown in regime 1 (SDG14 trawling ≤ 4.5516) increases the fishery outputs significantly less notably (coefficient value of the NF = 0.576) compared to using regime 2 (SDG14 trawling > 4.5516; the coefficient value of the number of NF = 0.599). According to the income level, and replicating the result of threshold SDG14 overfishing, in regime 1 (SDG14 trawling ≤ 4.5516), the income level and time positively affect global fishery outputs significantly less notably (high income = 4.400; lower-middle income = 1.932; upper-middle income = 2.515) than regime 2 (SDG14 trawling > 4.5516) (high income = 4.515; lower-middle income = 1.985; upper-middle income = 3.513). This is consistent with our expectation that countries with a high income level will drive greater demand for marine food, leading to increased global fishery outputs.

3.4. Additional Analysis by Controling for the Country GDP Growth and Property Rate

We add the GDP growth rate and poverty ratio for controlling variables and specify our second model (Equation (2)) as follows to generate new estimated threshold values.

$$\begin{array}{cc}\hfill LogGF{O}_{i,t}=& \left[{\alpha }_1{\left(Technological progress\right)}_{i,t}+{\gamma }_{1}LogNF{V}_{i,t}+{\theta }_{1}LogN{F}_{i,t}+{\xi }_{1}LogGD{P}_{i,t}+{\psi }_{1}LogPoverty rati{o}_{i,t}\right]\times f\left(k_{i,t}\le {\lambda }_1^{*}\right)\hfill \\ \hfill +& \left[{\alpha }_2{\left(Technological progress\right)}_{i,t}+{\gamma }_{2}LogNF{V}_{i,t}+{\theta }_{2}LogN{F}_{i,t}+{\xi }_{2}LogGD{P}_{i,t}+{\psi }_{2}LogPoverty rati{o}_{i,t}\right]\times f\left(k_{i,t}>{\lambda }_1^{*}\right)\hfill \end{array}$$

(2)

Table 6. Threshold estimates of using different threshold variables after adding country control variables (GDP growth rate and poverty rate).

Sample Split	SDG14 Overfishing	SDG14 Trawling
	Model 1	Model 2
The Lagrange multiplier test for no threshold	139.600	96.704
Bootstrap p-value	0.0004	0.0001
Threshold estimate	4.4456	4.4038
95% Confidence Interval	(4.396, 4.455)	(4.1596, 4.4073)

demonstrates the results of Equation (2) utilizing two distinct threshold variables of SDG14 overfishing and SDG14 trawling. The threshold value’s statistical significance was evaluated based on the p-value, estimated using the bootstrap technique with 5000 replications and a 15% trimming percentage. For an additional test, we divided our sample into two separate regimes. The point estimate of the threshold value of SDG14 overfishing is 4.4456, with a 95 percent confidence level ranging from 4.396 to 4.455 for Model 1. Besides, the threshold value for SDG14 trawling in model 2 has an estimate of 4.4038, with a corresponding 95% confidence interval from 4.1596 to 4.4073. Taken together, the number of fishers and fishing vessels in

Table 7. Regression results using different threshold variables after adding country control variables (GDP growth rate and poverty rate). Dependent variable: global fishery outputs.

Variables	Panel GLS without Threshold	Threshold Model 1 = SDG14 Overfishing		Threshold Model 2 = SDG14 Trawling
		Regime 1	Regime 2	Regime 1	Regime 2
		Overfishing ≤ 4.4456	Overfishing > 4.4456	Trawling ≤ 4.4038	Trawling > 4.4038
Constant	2.326 **	3.901 **	−5.234 **	−0.0769 **	4.826 **
	[3.234]	[16.980]	[−7.584]	[−0.295]	[18.108]
Logarithm (NFV)	0.493 **	0.423 **	0.488 **	0.966 **	0.243 **
	[10.422]	[20.701]	[7.710]	[28.849]	[9.677]
Logarithm (NF)	0.602 **	0.457 **	1.203 **	0.215 **	0.699 **
	[13.014]	[25.262]	[17.205]	[7.593]	[21.714]
Logarithm of GDP growth rate	0.00163	0.164 **	−0.274 *	0.392 **	−0.331 **
	[0.009]	[3.327]	[−1.911]	[7.926]	[−4.678]
Logarithm of poverty rate	−1.433 **	−1.322 **	−1.096 **	−0.861 **	−1.438 **
	[−25.124]	[−58.985]	[−18.631]	[−19.457]	[−38.016]
Technological progress (Time trend)	0.0117	0.0202 **	−0.00578	0.0379 **	−0.00385
	[1.195]	[5.644]	[−0.515]	[8.648]	[−0.789]
Observations	1240	905	335	558	682
Wald χ2	6932 **	6216 **	3137 **	5044 **	5705 **
Prob > χ2	0.0000	0.0000	0.0000	0.0000	0.0000

Notes: The t-statistics results are shown in square parentheses (white-corrected for heteroscedasticity). The findings correspond to a 15% trimming percentage. * and ** indicate 5% and 1% level of statistical significance, respectively.

keep the same results compared to baseline models (Equation (1)).

As shown in Table 7, interestingly, the results for regime 1 (SDG14 overfishing ≤ 4.4456) show that with a higher GDP growth rate (coefficient value of 0.164), global fishery outputs would be significantly increased, compared to SDG14 overfishing > 4.4456 (Regime 2), as GDP growth rate in these countries increases, their fishery production will reduce (coefficient value of the GDP growth rate = −0.274). We also duplicated the result of threshold SDG14 trawling in regime 1 (SDG14 trawling ≤ 4.4038) and found that the GDP growth rate positively affects global fishery outputs (coefficient value of 0.392) in comparison to countries with a high proportion of trawl net use (SDG14 trawling > 4.4038), meaning that fishery resources decrease significantly when using more trawling nets in countries.

In terms of the poverty rate for both threshold values of SDG14 overfishing and SDG14 trawling, we discovered that poverty rates are significantly negative to fishery outputs. This implied that as the poverty ratio in these countries increased, their fishery production decreased significantly compared with countries with a low proportion of overfishing and trawling. Overfishing and trawling net use are much more complex concerns in countries with high poverty ratios, and poverty drives poor fishers to deplete stocks far below levels at which bioeconomic balance would generally be reached.

4. Discussions

Regarding countries with a high proportion of overfishing activities (SDG14 overfishing > 4.4456), as the number of fishing vessels and the number of fishermen operating in these countries increased, their fishery production also increased significantly compared with countries with a low proportion of overfishing, but at the same time, this also led to an increase in overfishing or the presence of fish species lacking their required resources. Prior studies have proposed taxing overfished species or limiting the operation of fishing vessels as policy management tools [38,75,76,77]. Furthermore, implementing a wide range of policies such as subsidies and vessel buybacks has been suggested by Milazzo (1998) [78] as a “green subsidy”. The aim of the vessel buyback agendas is to lessen fleet overfishing, which improves the sustainability of the fishing industry and resource conservation [78]. Switching capacity-enhancing subsidies from fishing for reduced fish stocks to sustainable activities, such as promoting “fishing for plastic” instead of depleted fish stocks, is a high priority that would still allow fishers to receive subsidies [68].

On the other hand, in countries with a relatively high proportion of trawl net use (SDG14 trawling > 4.5516), as the number of fishing vessels increases in these countries, their fishery outputs are significantly reduced in comparison to countries with a low proportion of trawl net use, meaning that the use of trawl nets has a negative impact on fishery resources. For example, Hong Kong’s trawl ban policy, which aims to restore marine biodiversity and fisheries resources, offers an excellent opportunity to mitigate the adverse effects of trawling [79]. After two and a half years of trawling in Hong Kong waters, significant biodiversity improvements were shown [20,79]. Additionally, there have been attempts at implementing similar trawl bans in Canada and Italy, with mixed results [80,81].

We also found that in countries with a high overfishing ratio (SDG14 overfishing > 4.4456), such as low-income countries, fishery outputs are significantly higher than in middle-income and high-income countries, and specifically, low-income countries display a lower production of fishing vessels, often investing more manpower in increasing fishery production. The physical risk in fishery production will increase significantly and cause biodiversity to decrease and marine ecosystems to suffer damage. People suffer from fishery income and employment problems when the ocean ecosystem is damaged [82,83]. Furthermore, regarding countries with a high proportion of overfishing and trawling activities (threshold values: SDG14 overfishing > 4.4456; SDG trawling > 4.4038), as the GDP annual growth rate in these countries increased, their fishery production also decreased significantly compared with countries with a low proportion of overfishing and trawling, but at the same time, this also led to a decrease in overfishing and trawling net use. Interestingly, we also found that overfishing and trawling net use are much more complex problems in countries with high poverty ratios, and that poverty drives poor fishers to deplete stocks far below levels at which bioeconomic equilibrium would usually be achieved, which is consistent with prior studies [84,85]. Hence, providing appropriate fisheries management and alleviating poverty policies in these areas should be a priority, aiming to protect the vulnerable ecosystems in these regions [84,86].

Advances in science and technology enable humans to collect more data more quickly [87,88], and also to detect changes in the status of fish groups in a timely manner, so that they can be intervened in effectively. By using catch control principles, overfishing can be avoided when fish stocks decline, regardless of the cause of the decline. In the face of the problem of ocean productivity, it may be feasible to use a strategy such as this. Several areas of the United States already employ these types of fishing control principles [89]. In fact, fisheries have successfully used this method for capturing unstable stocks such as sardines and anchovies [90], as have fisheries focused on long-lived demersal fish [89].

International agreements, such as the Parties to the Nauru Agreement (PNA) ship-related scheme, have been used to regulate and control a variety of fish stocks (tuna) and thus reduce disputes and accommodate changes in fish populations and distributions. When the abundance or distribution of tuna changes, the quantity of tuna catch allocated to each PNA member country will change [91]; the catch allowance can also be shared between countries to avoid overfishing when the fish stocks change. In addition, there are precedents for compensatory payments (e.g., the US/Canada Pacific Salmon Treaty) and the trading of fishing rights (e.g., between Norway and Russia), which act as countermeasures between economic zones [92]. Introducing management systems or measures that encourage fishermen to reduce their catches of specific fish stocks can reduce the risk of fish stock depletion, ecosystem damage and fishery closures. For example, the change in the catch sharing system for bottom fish can reduce the fishing of endangered species [93].

5. Conclusions

Using panel data from 62 countries, this study applied a threshold panel data model to empirically explore the possible threshold effects of SDG14. It used distinct overfishing (SDG14 overfishing) and fish caught by trawling or dredging techniques (SDG14 trawling) as the threshold variables to examine the nonlinear interaction between the factors affecting the fishery fleet, fishers, and fishery outputs. Regarding overfishing and bottom trawling impacts, the study supports a single-threshold effect. The results demonstrate that various combinations of the primary components may enhance the levels of global fishery output (the number of fishers, fishing vessels, and income levels). The threshold value effect has various implications for fishermen regarding their fishing output.

The regime 1 results (low SDG14 overfishing) indicate that the numbers of fishery fleets and fishers, and income levels, positively increase fishery outputs, which effect remains the same when using the alternative threshold of SDG14 trawling. In regime 2 (high overfishing), the high income level and time factors are insignificant and inconsistent with baseline regression. In contrast, the other variables retain a similar level of significance, but their coefficient values are significantly greater than those of low fishing stocks. As for the threshold variable low bottom trawling, we find that the number of fishers, income levels, and time positively affect global fishery outputs slightly less significantly than in countries with high bottom trawling. In contrast, the result regarding fishing vessels is also significantly positive, but this coefficient is lower in cases with high trawling. This finding implies that countries with greater investments in trawl nets will endanger fishery resources due to the bycatching and discarding of non-target species; this is consistent with prior reports that bottom trawling accounted for approximately 23% of landed marine catches and 60% of fisheries’ discards globally, from 1950 to 2014 [50,51].

Finally, we discovered that in countries with high rates of overfishing, such as low-income nations, their fishery outputs are significantly higher than those of middle- and high-income countries. In particular, substantial investments in staffing or subsidies in low-income nations may lead to overfishing and impact global fishery production in the long term [94,95].