The SABI (Iberian Balance Sheet Analysis System) database was the main source for obtaining the data on the companies. This database is generated by Bureau van Dijk and contains financial information about Spanish companies. Many previous studies use activity headings in accordance with the Statistical Classification of Economic Activities (Hirsch & Gschwandtner, 2013). Given the specific nature of this research, the primary activity of the firms analysed is the ‘Raising of swine/pigs’ (code NACE 0146) and they were located in any province throughout Spanish territory during the 2003-2018 period inclusive. This classification was made up of pig farming companies, from birth until the animals are slaughtered and were therefore focused on their production. All kinds of companies were considered, of all sizes, whether or not they were family-owned, thus increasing the level of generality of the results obtained.

Initially, a total of 1,810 companies were observed, with a total of 28,960 observations. We covered a 56% of the population of companies in the swine sector. Subsequently, the database was cleaned, deleting atypical data of the sample so as to prevent biases in the results obtained and increase their robustness. In particular, the observations with profitability below -100% and above 100% were eliminated. Furthermore, the companies with a number of missing data equal to or greater than four of the variables at company level during the analysed period were also deleted. For the remaining cases, the missing values in both the variables at company level and the profitability itself were imputed, if sufficient information was available. The imputation process was carried out with the R package missMDA (Josse & Husson, 2016), which performs a principal component analysis (PCA) with incomplete data and selects their number by minimising the mean squared error of prediction of the data observed, based on the estimated data using the calculated components. For this purpose, the imputePCA function was employed, which uses a regularised iterative PCA algorithm described in Josse & Husson (2012). After carrying out these depuration processes, the definitive sample contained 1,802 companies with a total number of 19,617 observations.

The definitions of the analysed variables, as well as their descriptive statistics are displayed in Table 1.

The company’s profitability was analysed by the ROA (return on assets), a ratio expressed in a percentage that is calculated as earnings before interest and taxes (EBIT) divided by the total assets. Due to the high availability of this variable and its informative capacity, its use is very common in literature when analysing performance in any sector, including the agri-food sector (Grau & Reig, 2015). ROA gives us information about the amount of money that the company obtains in exchange for each euro invested in the business and shows the ability of the business to produce profit using its assets. In some sectors, the ROA is higher than in others because the amount of capital invested in assets varies (Panigrahi & Vachhani, 2021). In addition, ROA has proven especially useful for comparing pig farms, because it shows the rate of return on assets and, in some cases, the use of financial leverage to support the productive activity of a farm.

For the purposes of this paper, ROA is a better indicator than other alternatives such as Return on Equity (ROE). ROA reports the profitability of all company assets, while the ROE only determines the rate of return on capital invested by its owner.

To explain the evolution of profitability and with comparative purposes, the following explicative or independent company-level variables, most of which have been considered previously in literature, were used: the size of the company measured by using the natural logarithm of the total assets (in thousands of euros); the age of the company (in years); the increase in sales (percentage); the liquidity ratio measured with the logarithm of the quotient between current assets and current liabilities; the natural logarithm of the debt ratio, which divides the total debt (both long- and short-term) between the company’s total assets; the natural logarithm of the annual emissions of CO₂ (in tons of CO₂eq).

Some of the original variables were log-transformed in order to increase their normality and weaken the influence of outliers.

The size is expected to have a positive and significant influence on the ROA, given that larger hog confinements exhibit increasing returns to scale, consistent with the dramatic increase in market share of very large farms in recent years (Yu & Orazem, 2013). Likewise, the age of the company is expected to have a negative relationship with the ROA, as older companies have less capacity to react to technological change (Baráth et al., 2021). In addition, it is expected that growth could help improve employee motivation, thus achieving greater productivity and leading to an increase in the financial profits. Liquidity is expected to have a positive and significant effect on the financial profits while debt should have negative implications, due to companies use their resources reducing financial expenses, thus increasing their profits and making more resources available to renew obsolete investments that could be more productive. Finally, an environmental variable (CO₂ emissions) was included because the swine industry is one of the major polluters in the food industry. To that aim, each year while the study was underway we obtained data on the tons of CO₂ emitted overall by farms to obtain one ton of pork. We believe that the reason certain farms are able to produce lower CO₂ emissions is partly due to their investment in fixed assets for animal housing, and having waste treatment facilities, such as biodigesters that transform biological waste into biodiesel. For this reason, we have taken each farm’s investment in these tangible fixed (measured as fixed assets/sales), as a corrector for the average CO₂ emissions obtained from the sector. The corrected amount of CO₂ emissions has been incorporated into the model to see its effect on ROA. We expect the effect of this variable to be positive, that is, the higher the level of pollution, the greater the profitability. This is due to the fact that to reduce pollution, it is necessary to invest in non-productive fixed assets which, very likely, will have a negative impact on profitability.

With the purpose of analysing the impact of the characteristics at an industry/subsector level, we considered their concentration, size and growth. Based on previous studies, the following independent variables at this level were chosen: the market concentration level measured by the Herfindahl-Hirschman Index (HHI), thus a high (reduced) value of the index is a sign of a market with a high (low) concentration and un- (very) competitive; the natural logarithm of the sales in the sector (in thousands of euros); and the increase in the number of companies in the sector.

It can be seen (see Fig. 3 below) that there was an increase of fairly large farms, particularly the biggest pig producers, although there was a considerable drop in the number of small farms, leading to a decrease in the total number of pig production companies. However, increased production in certain financial years could have resulted in sporadic increases in the total number of farms.

For this reason, a significant negative relationship between the industry concentration index (subsector) and the company’s profitability was expected, due to these companies’ predisposition to group together. Furthermore, the increase in production is expected to lead to a growth in the industry’s sales and this will benefit profitability.

At a geographical or territorial level, the following independent variables were used, related to previous literature: the natural logarithm of the provincial unemployment rate; the level of education estimated by using the regional (denominated Comunidad Autónoma in Spain) educational training percentages; and the rate of foreigners measured through the proportion of foreign-born people among the total population in Spain.

These three variables were extracted from the National Statistics Institute (INE, 2018). The following was also analysed: population density or number of inhabitants per square metre referring to the municipality in which the company operated, whose data come through the Digital Atlas of Urban Areas in Spain (MITMA, 2022).

We used geographical variables instead of other kind of variables (macroeconomic, political variables) in line with previous studies (e.g. Usai & Paci, 2003; Ollinger et al., 2005; Webber et al., 2009; Schiefer, 2011; Bekeris, 2012; Fairlie, 2013; Fearne et al., 2013; García-Alvarez-Coque et al., 2013; Zouaghi et al., 2017) that provide information about the composition of the labour force and its quality.

The unemployment rate was expected to have an effect related to the workforce in the area where the company was located, and this could affect its financial profits. Likewise, for the reason described in the previous paragraph, both the level of education of the town where the companies were located and the rate of foreigners were expected to exercise a growing influence on profitability. Finally, it was highly likely that the population density would have a positive influence on performance, as labour costs would possibly be lower in populated areas.

In addition, and in order to know the degree of persistence of financial profits over time, the ROA variable was delayed for a period.

Fig. 1 displays the evolution of the mean profitability throughout the analysed period. It was observed that after a drop in the mean profitability in 2007, during which mean profits of -1.55% were reached as a consequence of the crisis, there was a growing trend after that year with an estimated average compound annual growth rate of 3.29% cumulated between 2007 and 2018. The reasons for this evolution obeyed a notable restructuring in the sector in recent years, during which there was a marked decrease in the number of small pig farms. At the same time, there was an increase of large pig producers due to takeovers and/or mergers of those that already existed, which led to an increase in production and the census, as well as the average profitability for the period (MAPAMA, 2020).

Fig. 2 displays the annual evolution of the mean values of the explicative company-level variables described in the data and variable section. During the 2007-2015 period, a stable trend is observed in the evolution of the mean values of the variables, in terms of the size of the companies, followed by a growing trend from 2016 to 2018 (Fig. 2). The sales increase was stable and positive (around 10%), except in recent years, when they dropped by 5%. A growing trend in liquidity was also seen from 2008, accompanied by a falling trend in the debt ratio. Indeed, the rise in pork consumption in the domestic market enabled companies to increase their cash flow entries on their balance sheets, thus improving their liquidity and showing a lower dependence on bank debt. Finally, it can be appreciated that the average values of CO₂ emissions remained more or less stable throughout the period analysed.

Fig. 3 displays the annual evolution of the concentration, size and growth of the swine subsector. It is observed that there was a clear rising trend in the concentration of the sector (especially from 2007 onwards) and a higher level of sales, and a stationary trend in the new number of companies, with consecutive rises and falls that fluctuated at an average of around -8.86% per year (see Table 2). This is due to the fact that in the recent decades the pig industry had been through a restructuring process in the number of farms.

Fig. 4 displays the annual evolution of the geographical variables described in the data and variables section. In 2013 the unemployment rate reached a maximum before dropping significantly. However, the population’s cultural levels displayed a clear growing trend, which is expressed by a greater percentage of people with a secondary level of studies or higher, accompanied by drop in the percentage of people with primary education. The importance of this growth in the educational level makes workers tend to obtain higher educational achievements, which translates into higher productivity levels and, consequently, in higher profitability levels. In addition, from 2008 onwards, the percentage of foreigners was stable and the same thing happened with the population density, which fell slightly in 2018.

The total number of pig producers analysed was N=1,802 with T=16, giving rise to 19,617 observations of the type of business per year. The estimated model showed an adequate goodness of fit to data since there were no over-identifying problems (the difference with the Sargan-Hansen test was not significant) nor the second-order residual autocorrelation. The results are displayed in Table 2 and show the existence of significant influences on profitability in terms of both companies and sector/industry, and the territorial and/or geographical factors in which the company carried out its activity. It should be noted, however, that the results did not include data from individual farmers or self-business since both of them are not collected by SABI, and this is a limitation to the research.

The size of the company exercised a significant U-shaped influence, reaching minimum profits in companies with total assets of €1,776,000. Therefore, it was observed that up to a certain level of assets, expansion could have negative impacts on profits. Nevertheless, this trend changed as soon as the company’s total assets reached a figure of nearly €1.8 million, thus becoming more profitable as it increased its size.

The reason for this type of effect lies in the growth of the global demand for meat, as new investments are necessary to keep abreast of the latest developments and larger facilities are required to ensure profitability. However, the investment for creating farms and/or extending those that already exist implies a cost that not everyone can afford (Wijnands et al., 2007; Chaddad & Mondelli, 2013; Hirsch et al., 2014). Consequently, companies with a higher turnover will have greater capacity to adapt and mitigate market risks, but small entities are more exposed to these risks, which explains why there was an increasing trend to either change to the integration system and/or disappear (Domínguez & Daudén, 2018). In contrast, expanding by diversifying towards new markets is usually restricted by the capacity of the company’s existing management team, as analysed in the first administrative theory of business literature (Penrose, 1995; Marris, 1964). Consequently, the management team may condition the company’s trend as regards its size, as traditional livestock farmers can manage their growth up to a certain extent, but afterwards they will need a larger management team that can cover the businesses expansion.

The company’s age had a significant negative effect on profitability, in line with previous research (Agarwal & Gort, 2002), with a drop in profitability of 0.28% for each year’s increase in the age of the company. This could be due to diverse reasons: rigid organisational structures, slow growth, outdated assets (Loderer & Waelchli, 2010; Hirsch et al., 2014; Zouaghi et al., 2017), lack of quality in business innovation or inertia or bureaucracy (Majumdar, 1997), which can help reduce the ability to react to financial circumstances.

Sales growth is considered to be an indicator related to the company’s ability to compete and protect itself from cyclical fluctuations in the market (Rassier & Earnhart, 2015), as well as being a synonym for business success. Previous studies (Pál & Ferrando, 2010; Delmar et al., 2013; Pattitoni et al., 2014; Zouaghi et al., 2017) showed that the growth of sales is associated to the probability of survival, as it represents an increase in the size of the company, and the size lowers the risk of leaving the market. It is also a strengthening process, where previous growth leads to future growth (Delmar et al., 2013), and the dynamics of this growth are an incentive for company employees who will not feel at risk of losing their jobs (Pattitoni et al., 2014; Zouaghi et al., 2017). After the growth, there will be an increase in both productivity and profitability, understanding sales growth as a proxy for investment opportunities and the increase of the size of the company (Pál & Ferrando, 2010). Greiner (1997) considered that this sales growth variable could have negative effects in the event of a break-up in informal relationships between employers due to increased competitiveness.

We also found a U-shape and clearly significant relationship, achieving a minimum profitability of €3.9 million. Given that the corresponding coefficients were significantly positive we conclude that Sales growth is positively related to Profitability but with increasing returns to scale. This multiplying effect is explained by the rise in production that took place in the sector at a time when there was a drop in the number of farms, which had turned into economies of scale, making it possible to increase profitability. This fact, coupled with the rise in the price of pigs in origin, had a positive impact on profitability.

As regards the effect of the financial risk, the influence of the liquidity ratio and debt ratio were analysed. The liquidity ratio, as an indicator of the company’s capacity to meet short-term payment obligations (Rees, 1995), was expected to exercise a significant, positive effect on profitability as when companies have a lower risk they have greater capacity to face their short-term debts, and are more profitable in the long run. In our case, this effect had a U-shape and clearly significant relationship. Given that the estimated coefficients were significantly positive, we conclude that liquidity is positively related to profitability but with increasing returns to scale. The companies have greater ability to adapt to changes as a consequence of having the necessary resources to deal with unexpected situations and short-term financial obligations (Goddard et al., 2005). The impact of the financial risk (measured as the opposite of liquidity) was mainly negative and significant in the Spanish agri-food sector (Zouaghi et al., 2017). The results obtained were in the same line as other empirical studies (e.g., Gschwandtner, 2005; Enqvist et al., 2014; Hirsch et al., 2014; Pattitoni et al., 2014), based on the paradox that good business practices can increase the ROA and, at the same time, reduce financial risk (Bowman, 1980). Therefore, when liquidity is managed properly, short-term payment commitments can be addressed and new investments can be carried out with greater security.

Continuing with the financial risk analysis, the company’s debt ratio had relevant implications for profitability. It was shaped like an inverted U, reaching a maximum in 0.2390. According to these results, taking on debt had a positive effect on profitability until the maximum was reached but, from that point onwards, it tended to exercise a negative effect on profitability. The higher the level of indebtedness, the stronger the effect. The negative effect is due to the fact that an increase in the financial risk reduces profitability and implies lower profits; therefore, fewer resources are generated (Bowman, 1980). For this reason, a company with high debt levels could decide not to renew new productive investments for the company’s performance (Garvey, 1992). In line with previous studies, companies that are not vulnerable to negative financial situations, corresponding to those that do not experience significant reductions in their financial profits during these periods, all have in common the growths of their assets and reductions in their levels of indebtedness (Grau & Reig, 2015).

With respect to the influence of the CO₂ emissions produced by the farms, a significant positive influence on their profitability was observed, in such a way that when CO2 emissions increase by 1%, profitability can be expected to grow by 0.69% (see Table 2). Therefore, as anticipated, the control of emissions did not have an economic benefit in companies in the pig sector, but on the contrary, the costs of adapting the facilities reduced profitability without compensating the investment effort for environmental purposes. These results denote the lack of sufficient incentives in companies, in the period analysed, to undertake investment projects aimed at environmental improvement.

The variables measured in terms of industry and/or subsector exercised a significant influence on the evolution of the companies’ profitability. Thus, lower levels of concentration (higher number of companies) and higher sales levels in the sector were directly associated to profitability. This was justified by the synergies created by the proximity between companies. This situation led to reductions in transaction and transport costs, due to the fact that it was easy to work with suppliers in the area. Furthermore, as the integrating businesses needed to put the animals on multiple farms in order to attend to the existing demand, it led to higher markups for the farmers, which helped increase profitability. This result did not coincide with the results presented in other sectors that find a positive and significant impact between concentration and profitability (Bhuyan & McCafferty, 2013; Delmar et al., 2013; Hirsch et al., 2014; Zouaghi et al., 2017). It must be pointed out that literature also shows that strong business dynamism can lead to instability and high volatility in the environment, which has a negative effect on profits (Misangyi et al., 2006).

In our case, the size of the industry acted as an indicator of a heavy demand and high profits (Zouaghi et al., 2017). However, this was a circumstantial result and could have changed had there been a variation in the circumstances of the international market. Therefore, it is advisable to take these results with due precaution.

In relation to territorial variables, significant influences on profitability of a different sign could be observed, with the sole exception of the percentage of foreigners in the area.

As regards the unemployment rate, our results showed a significant positive paradoxical relationship with profitability, with the opposite sign to the one found in previous literature. According to previous evidence, an increase in the regional unemployment rate reduces profitability, especially in small businesses (Bekeris, 2012). High unemployment can make businesses enter the market, with the purpose of completing their staff, which increases competition and causes profitability to drop (Fairlie, 2013). However, in our case, greater competition generates higher financial profit in this kind of company. The reason for this relationship is associated to the workforce, as the companies within this sector have a constant need to hire staff. High unemployment in a specific area can make the company’s profitability increase as a result of hiring unqualified unemployed staff who probably demand lower salaries.

Previous research sustains that a higher level of education can lead to a rise in productivity, greater competitiveness and, consequently, an increase in profitability. As a result, the companies located in areas where there is a high educational level can be expected to be more productive and competitive (Usai & Paci, 2003). Nevertheless, it can have the opposite effect; in other words, when the population has a lower educational level, companies tend to be more profitable, due to the existence of workers with lower qualifications (Ollinger et al., 2005; Schiefer, 2011). In our case, a significant relationship with a negative sense was observed, regarding both the percentage of population with higher studies and the percentage of population with primary educational levels, which implies a positive relationship with the percentage of population with secondary level studies. Bearing in mind that, in general terms, the population’s educational level had increased (see Fig. 4), these results suggested that the level of education had an overall positive effect on profitability, and an excessively qualified workforce was not required.

In respect to foreign-born population, it was observed that it did not have a significant effect. This could have been due to the fact that foreign manpower is related to other economic activities, such as the meat industry, slaughterhouses or activities related to agriculture (Zouaghi et al., 2017).

It was also considered relevant to examine the effect of density on the ROA, as it would determine the environment in which the most profitable companies were located and would provide information about which areas would be the most appropriate for setting up farms when investing in the business. Our results indicated that this variable exercised a slightly significant negative influence on companies’ profitability, with rural areas being the most profitable. In general, companies dedicated to livestock production were located on land not suitable for development near agricultural areas, due to the proximity for accessing raw materials and the subsequent cutting of costs this implied. This feature acts as an element that leads the population to settle in the rural environment, consequently reducing the rural depopulation phenomenon (García-Moreno, 2020). The result was similar to previous studies which affirm that especially in micro-enterprises and small and medium-sized enterprises, the rural environment is not perceived as a limitation for profitability (Fearne et al., 2013; García Alvarez-Coque et al., 2013).

Finally, we observed a persistence over the years that was significantly positive, although not very marked, in the evolution of the profitability, with a first-order autocorrelation coefficient of a medium-low type, with a value of around 0.17 (see Table 2). Therefore, in respect to the ROA, effects of the past affect the present day (Hirsch et al., 2013), as it is a sector that does not directly suffer the consequences of external negative situations and its revenue has a certain guarantee of security due to the static demand.