#### 3.1. Data Collection and Research Plan

The article analyzed a sample of 22 companies operating in the biogas plant sector in Italy, with the plants located in the Lombardia, Emilia Romagna and Veneto regions. All 22 companies in the sample operate under the ATECO 2007 code “351100 Production of electric energy”. The firms in the sample are based in three regions of northern Italy. Fourteen companies are located in the Lombardia region (7 in the Province of Cremona, 3 in the Province of Brescia, 2 in the Province of Mantova, 1 in the Province of Milano and 1 in the Province of Pavia); 5 are located in the Emilia Romagna region (3 in the Province of Bologna and 2 in the Province of Ferrara) and 3 are located in the Veneto region (2 in the Province of Rovigo and 1 in the Province of Verona). All plants in the analysis are located in the Po Valley, which is the Italian area with the largest concentration of biogas plants. The plants are located in lowland areas at altitudes of between 35 meters above sea level and 210 meters above sea level. Vegetable biomass (primarily corn, sorghum and other cereals) and livestock effluent (particularly that resulting from dairy cattle and pig breeding) are fed to the plants. The production of cereals and the rearing of bovine and porcine livestock are especially prominent agricultural activities in Italy’s Po Valley. The facilities are all fairly new, having begun operations in the period of 2007–2009. Five of them began operating in 2007, 14 began operating in 2008 and 3 began operating in 2009. For all of the plants, 2010 was a year of normal operations, and the associated data are considered normal operating data. The firms had a median installed electric capacity of 999 kW. In fact, 18 of the enterprises had 999 kW of installed power, 2 of them had 800 kW of installed power and 2 of them had 600 kW of installed power. Companies with an installed electrical capacity exceeding 1 MW each were not included in the sample. Such large companies, if included in the sample, would have had data that was not comparable to that associated with the firms in the sample. All the electricity produced at the plants considered in this article was fed into the national electric grid, and the company, Gestore Servizi Energetici (GSE) Spa, purchased it based on an incentive tariff of 280 € per MW per hour. The firms in the sample are legally considered as agricultural firms; in fact, Article 1 of the Decree of May 18, n. 228, “Orientation and modernization of the agricultural sector”, redefines Article 2135 of the Italian Civil Code where the use of renewable energy is classified as agricultural activity; under the roles of Italian bankruptcy law, agricultural firms are not subject to bankruptcy. Data derive from “Computerized analysis of Italian firms” database (AIDA). The data analysis has been performed with SPSS statistical package (issue 19). The data cover a five year period, from 2010 to 2014, with 110-year data. In the article, the analysis is developed as follows: (a) We reclassify the annual accounts of 22 biogas plants firms included in the sample, applying descriptive statistics to balance sheets, income statements, and cash flow statements; (b) we test whether there are statistically significant correlations between economic margins (variables EBITA, EBIT, and Π) and financial margins (CF, OCF, UFCF, and FCFE); (c) we test whether there are statistically significant differences between economic margins (EBITA, EBIT, and Π) and financial margins (CF, OCF, UFCF, and FCFE); (d) we apply two linear regression models to determine the explanatory variables of ROE and FOE in the biogas plants firms of the sample.

#### 3.2. Annual Account Data Analysis

The majority of the plants using bio energy (i.e., biomass, biogas and bio liquids) in Italy at the end of 2014 were small plants, with each one producing less than 1 MW of power as the Statistical Report of GSE Spa for the year 2014 (entitled “Energia da fonti rinnovabili in Italia”) indicated. The 2014 report surveyed 2482 plants in total: Altogether, they produced 4044 MW of power and generated 18,732 GWh of gross electric energy. A large proportion of the plants (2104 of them) produced less than 1 MW of power each; altogether, these plants produced 1261 MW of power and 7700 GWh of energy. Moreover, 313 power plants produced between 1 MW and 10 MW of power each; altogether, they produced 893 MW of power and 3009 GWh of energy. Last but not least, 65 plants produced more than 10 MW of power each, giving a total power production of 4043 MW and a total energy production of 8024 GWh. In 2014, in Italy, the bio-energy production was 17,732 GWh, 15.5% of the total production of renewable energies. Also in 2014, out of a total of 2482 plants, there are 321 plants using solid biomass (municipal waste or other biomass), with a total power of 1610 MW and 6130 GWh of energy produced, 1796 plants producing biogas with a total power of 1406 MW and 8199 GWh of energy produced, 526 plants using bioliquids, with a total power of 1027 MW and 4341 GWh of energy produced. Among the 1796 plants producing biogas, 360 utilize waste (401 MW and 1638 GWh), 74 use sludge (44 MW and 121 GWh), 421 use animal manure (203 MW and 989 GWh) and 941 using scraps of agriculture and forestry (758 MW and 5451 GWh), thus confirming the impact of agricultural activity on bio-energy production. Between 2001 and 2014, the number of biomass plants increased from 202 to 2482, with an increase in installed power from 740 to 4044 MW; from 2009 it was noted, in particular, an increase in smaller systems, less than 1 MW, which benefited from higher public incentives, in the form of feed-in tariffs established by the Ministerial Decree (DM) of 18 December 2008. The geographical location, out of a total of 2482 plants (4044 MW and 18,702 GWh, of which 8199 GWh from biogas), 75.1% are in the northern regions of Italy, where the Lombardia region is the first region in Italy to have plants (657 plants) for power (918 MW) and energy output (4249 GWh of which 2702 GWh is from biogas), followed by the Veneto region, with 345 plants, totaling 359 MW of power and 1899 GWh of energy produced, of which 1158 GWh is from biogas, and the Emilia Romagna region with 289 plants, totaling 613 MW of power and 2759 GWh of energy produced, of which 1272 GWh is from biogas.

In total, 1291 plants (51.02% of the total) are located in the first three regions of Italy (according to the concentration of plants). Altogether, they produce 1890 MW of power (46.74% of the total). Moreover, they produce 8907 GWh of energy (47.63% of the total), of which 5132 GWh comes from biogas (62.59%), confirming the territorial concentration of Italian energy production in regions characterized by a greater intensity of agricultural activities.

The analysis of the 22 sample firms first considers annual account data (

Table 1), which confirm the high level of capital intensity required for biogas plants activities (the median value of TA/S is 2.642 and NIC/S is 2.335); sector firms are capital intensive, particularly considering fixed assets (the median value of FA/S is 2.278). Capital absorption is relevant in fixed assets (the median value of FA is 86.20% of TA), and this confirms that biogas plants firms are characterized by relevant investments in fixed assets, particularly for Bfa

^{tan} (86.09% of TA). FA investments have an effect on increasing the capital needed to finance long-term investments, to be covered with ET or Df

^{>12m}. To cover their financial needs in FA, firms in the sample use Df

^{>12m} as the first source of capital, given the fact that the median value of Df

^{>12m} is 67.64% of TA while ET = 13.15% of TA and Df

^{<12m} = 9.45% of TA. The values of financial debt are quite symmetric (.269 for Df

^{<12m} and −0.312 for Df

^{>12m}). Stable sources of capital (ET + Df

^{>12m}) are 80.79% of TA, while FA is 86.20% of TA, and stable sources of finance are not able, in median values, to completely cover financial needs to finance FA investments. A part of FA investments is then financed with short-term loans, expressing a typical matter of financial risk, particularly with Df

^{<12m}.

In

Table 2, we express net invested capital (NIC) as the sum of FA and NWC. Data show that FA is 97.52% (the mean is 96.38%) of NIC, while NWC is 2.48% of NIC (the mean is 3.62%). These data are particularly interesting because they confirm that biogas plants do not have financial absorption to finance NWC investments. Out of 110 cases in the sample, NWC ≥ 0 in 64 and NWC < 0 in 56. ET is 14.88% of NIC (the mean is 19.20%), while NFP is 85.12% of NIC (the mean is 80.88%) as the main source of firms’ capital. An analysis of balance sheets shows that the data have a high level of positive and negative skewness and kurtosis for the majority of values, therefore, the Kolmogorov–Smirnov D statistic on normality of distribution shows that the balance sheet values do not follow a normal distribution.

Further information on the typical characteristics of firms in the sector results from an analysis of economic data in

Table 3. The median value of S amounts to 1,901,637 , and the major production factors are raw materials (Mc), 531,086 , 27.93% of S, and services (Sc), 371,328, 19.53% of S). EBITDA has a median value of 811,401 (42.67% of S) and a mean of 799,577 Ac + Dc absorbs a median value of 312,438% of S, and EBIT then has a median value of 498,963 (26.24% of S) and a mean of 494,542. The median values of EBITDA and EBIT are slightly higher than the mean values, as expressed by a comparison with mean values, having EBITDA ≥ 0 in 104 cases out of 110 and EBIT ≥ 0 in 98 cases. Financial management (SF) absorbs a median value 10.79% of S (i.e., 41.12% of EBIT and 25.29% of EBITDA) in mean values, and the data highlights that SF ≥ 0 in 10 cases out of 110, contributing to profit generation; Π has a median value of 274,013 (14.41% of S) and a mean of 272,881 , and Π ≥ 0 in 85 cases out of 110. It is useful noting that 5 cases of negative Π are concentrated in one firm, and 3 cases of negative Π are concentrated in a single other firm, thus confirming firms’ capacity to generate income via biogas plants management in a large majority of cases. An analysis of income statements also shows that the Kolmogorov–Smirnov D statistic on normality of distribution highlights that income statements data do not follow a norma distribution.

Given the high level of investment required to access the sector, this analysis of the differences between income and financial margins highlights necessary considerations to evaluate firms’ capacity to cover the costs of financial debts and to ensure NFP repayment via mortgage plans. In fact, EBITDA and EBIT, as economic margins, are frequently applied to assess the sustainability of a business cycle and to approximate cash flow, particularly regarding interest coverage ratios (ICRs) application [

64], even if some researchers have highlighted that ICRs could be improved applying a financial approach [

65,

66]. The analysis of cash flow statements (

Table 4) calculated from 110 years of data highlights some typical management characteristics of firms in the biogas plants sector: (1) Income margin profit (Π) generates a significant amount of cash (274,013 as a median value, that is, 133.92% of FCFE); (2) CF, because of the high values of Dc + Ac, is relevant and amounts to 791,644 as a median value, that is 386.92% of FCFE; (3) the dynamic of NWC investment does not absorb a significant amount of liquidity; (4) OCF is then 386.92 as a median value, that is 379.78% of FCFE; (5) the dynamic of FA investments absorbs a relevant part of Dc + Ac, as expressed by UFCF values, making a median value of 607,802 , that is 297.07% of FCFE; (6) SF absorbs a great part of 195.86% of FCFE with a median value of 403,200. Given these results, the analysis shows that CF ≥ 0 in 109 cases out of 110, OCF ≥ 0 in 107 cases out of 110, UFCF ≥ 0 in 102 cases out of 110, and FCFE ≥ 0 in 63 cases out of 110. In the 110 considered cases, EBITDA ≥ 0 in 104 cases, EBIT ≥ 0 in 98 cases, and Π ≥ 0 in 75 cases. The analysis shows in several cases that the sample firms are not able to cover the cost of debt and FCFE available for NFP repayment is consequently reduced. In fact, the case in which FCFE < 0 expresses the inability of firms in the sample, on average, to proceed to a distribution of profits and eventually repay NFP.

In

Table 5, we show some ratios that analyze economic dynamics (ROE, FOE, ROS), turnover of capital (TURNOVER), cost of debt (ROD), level of financial indebtedness (NFP/E), and duration of the working capital cycle, given by average number of days of trade receivables (AR_DAYS) and trade payables (AP_DAYS). Analysis of the data indicates that companies in the sample had a profitability (ROE) median of 0.4147 (and a mean of 0.3279). This profitability was much higher than the average yield on Italian government bonds, which was equal to 0.0135 for the year 2014 (and 0.0208 for the year 2013). A recent survey [

67] quantified the required return on equity in the Italian market at 0.070 for the year 2014. In the sample, the yield was still high, but it was lower in terms of FOE, with a median value of 0.3097 and a mean of 0.2374. It is confirmed that the biogas plants firms have high operating profitability (0.2624 as median value and 0.2616 as mean value) while the turnover of capital is below the net asset value (0.3785 and 0.3717 as the median value as the mean value). The return on capital employed (ROA) has the median value of 0.1123, higher than the median cost of debt, 0.0541, thus ensuring, on median values, that the leverage effect of debt has convenience. The ratio between NFP and ET (DER) is 5.722 (the mean is 4.208), expressing a high level of financial indebtedness, thus confirming the relevance of financial debt in biogas plant firms’ capital management. Biogas plants firms have short delay in collecting their accounts receivable (Cwc

^{ar<12m} + Cwc

^{ar>12m}) but quite often have large delays in payment to suppliers. AR_DAYS, calculated as (Cwc

^{ar<12m} + Cwc

^{ar>12m}) × 365/S. has a median length of 50.421 days (the mean is 60.759 days). Accounts payable (Dwc

^{ap<12m} + Dwc

^{ap>12m}) are also an important source of capital (the mean value is 6.10%, and the median value is 5.89% of TA). AP_DAYS, calculated as (Dwc

^{ap<12m} + Dwc

^{ap>12m}) × 365/S, has a median length of 119.73 days (the mean is 59.902 days). The capital generation due to the length of AP_DAYS confirms the bargaining power of biogas plants firms with regard to their suppliers.

#### 3.4. Multiple Regression Analysis

The aim of the research is then to quantify the causal relationship between a variable to be explained (the dependent variable) and one or more explanatory variables (independent variables), as exposed in the models. In the article, we calculate the determinants of economic (ROE) and cash flow (FOE) ratios available for equity holders, in order to provide useful information for managing firms in the biogas plant firms. First, the research aims to analyze if there is a relationship between a financial return on equity capital for a given period, t (FOE

_{t}), and some independent variables. FOE expresses the amount of cash available for equity holders as expressed in the methodological part of the article. To achieve this aim, we consider the explanatory capacity of a linear regression model (first model). The model, as expressed in Equation (9), considers FOE

_{t}, which expresses the financial return available for equity holders, as an independent variable for a given time (t). In the first regression models, the constant term is α, the variables are: TO (turnover), AR_DAYS, AP_DAYS and DER. The model then considers EBITDA, EBIT, and Π as explanatory variables, considered in values for the years t and t−1 (EBITDA

_{t} and EBITDA

_{t−1}, EBIT

_{t} and EBIT

_{t−1}, and Π

_{t} and Π

_{t−1}, respectively). At the same time, CF, OCF, and UFCF are considered explanatory variables and considered in their values for years t and t−1, giving then another six explanatory variables (CF

_{t} and CF

_{t−1}, OCF

_{t} and OCF

_{t−1} and UFCF

_{t} and UFCF

_{t−1}, respectively). The model could be expressed as follows:

The first model tries to explain actual FOE

_{t} (at a given time, t) considering a set of explanatory variables that express capital intensity (TO), working capital cycle duration (AR_DAYS, AP_DAYS), debt level (DER), operative profitability (ROS), actual income margins (EBIT, EBITDA and Π), and their respective values considered at t−1 (EBIT

_{t−1}, EBITDA

_{t−1}, and Π

_{t−1}), even considering actual financial margins (CF, OCF, and UFCF) and their respective values considered at t−1 (CF

_{t}, OCF

_{t}, and UFCF

_{t}). Unless otherwise specified, all the explanatory variables are taken at a certain time, t. The first regression model, as expressed in Equation (9), is analyzed in

Table 8 and assumes a significant statistical capacity to explain FOE

_{t} values; the F statistic has high significance (F = 0.000);

R^{2} is 0.9310, while adjusted

R^{2} has a value of 0.8421, expressing the capacity of the model to explain a great part of the variability of FOE

_{t}; the statistic DW is 2.118; and the majority of the variables are significant. First, TO has a positive effect on FOE values, expressing that an increase in turnover (then a decrease in the capital-intensive structure of assets) has a positive effect on the FOE value. The explanatory variables of FOE generation are, in particular, values expressing the duration of the working capital (WC) cycle. AR_DAYS has a negative sign, expressing that an increase in WC durations has a negative effect on the FOE result. AP_DAYS has a positive sign on FOE, expressing the opposite situation. DER has a positive sign on FOE given that an increase in debt could generate cash. Even ROS is particularly important in increasing the FOE value. Income and financial margins at a certain time, t, have an effect on FOE at the same time, t (particularly OCF

_{t} and PROFIT

_{t}). Income and financial margins at t−1 have even effect on FOE, particularly PROFIT

_{t−1}, OCF

_{t−1} and UFCF

_{t−1} margins, even with a relation significant only at the 0.05 level (two-tailed).

The research then considers a second regression model to analyze if there was a relation between economic return on equity capital for a given period, t (ROE

_{t}), and a set of independent variables. ROE expresses the return available for equity holders as expressed in the methodological part of the article. It then proposes an explanatory linear regression model (second model). In the second regression model, the constant term is α, and are considered the explanatory variables: TO (turnover), AR_DAYS, AP_DAYS, DER. The model then considers EBITDA and EBIT as explanatory variables, considered in values for the years t and t−1 (EBITDA

_{t} and EBITDA

_{t−1}, EBIT

_{t} and EBIT

_{t−1} respectively). At the same time, CF, OCF, UFCF and FCFE are considered explanatory variables and considered in their values for years t and t−1, giving then another six explanatory variables (CF

_{t} and CF

_{t−1}, OCF

_{t} and OCF

_{t−1}, UFCF

_{t} and UFCF

_{t−1}, FCFE

_{t} and FCFW

_{t−1} respectively). Obviously, PROFIT is not considered as an explanatory variable. The set of explanatory variables is the same as those considered in Equation (10), with the exception of FCFE instead of Π. The model could be expressed as follows:

The second model tries to explain actual ROE

_{t} (at a given time, t) considering a set of explanatory variables that express capital intensity (TO), working capital cycle duration (AR_DAYS, AP_DAYS), debt level (DER), operative profitability (ROS), actual income margins (EBIT, EBITDA), and their respective values considered at t−1 (EBIT

_{t−1}, EBITDA

_{t−1}), even considering actual financial margins (CF, OCF, UFCF and FCFE) and their respective values considered at t−1 (CF

_{t}, OCF

_{t}, UFCF

_{t}, FCFE

_{t−1}) Unless otherwise specified, all the explanatory variables are taken at a certain time, t. The second regression model, as expressed in Equation (10), is analyzed in

Table 9 and assumes a quite good statistical capacity to explain ROE

_{t} values; the F statistic has high significance (F = 0.000);

R^{2} is 0.8861, while adjusted

R^{2} has a value of 0.7455, expressing the capacity of the model to explain a good part of the variability of ROE

_{t}; the statistic DW is 2.204; and the majority of the variables are significant. First, TO has a positive effect on ROE and values expressing duration of the working capital (WC) cycle (AR_DAYS) has a positive sign, expressing that an increase in WC durations has a positive effect on the ROE result. Even ROS is particularly important in increasing the ROE value. Income and financial margins at a certain time, t, have an effect on ROE at the same time, t (particularly OCF

_{t} and FCFE

_{t}). Income and financial margins at t−1 have, instead, an effect on ROE but are limited to FCFE

_{t−1}, while EBIT

_{t−1} and UFCF

_{t−1} margins are significant only at the 0.05 level (two-tailed).