Comparing Benchmarking Measures: An Application to Retail Fertilizer Firms

Michael A. Boland and Jay T. Akridge

Michael Boland is an assistant professor in the Department of Agricultural Economics, Kansas State University, and Jay Akridge is a professor in the Department of Agricultural Economics, Purdue University. The Tennessee Valley Authority provided funds to assist in collecting the retail fertilizer plant data used in this study. Comments and suggestions of two anonymous reviewers are greatly appreciated. This is Contribution No. 99-185-J of the Kansas Agricultural Experiment Station, Manhattan, Kansas.

Abstract

The objective of this research was to compare financial ratios and a productive efficiency measure to determine their appropriate use in benchmarking. Return on sales, return on assets, and expense-revenue ratios are compared to productive efficiency measures for use in benchmarking the performance of retail fertilizer plants. Different measures assess different areas of performance. For example, financial ratios incorporate information concerning how a manager markets outputs and purchases inputs (i.e., incorporate "price effects"), while productive efficiency measures incorporate information on a single variable input (i.e., remove "price effects"). The results suggest that financial ratios are not good substitutes for a measure of productive efficiency when measuring short-run variable cost efficiency.

Key words: agribusiness, benchmarking, efficiency measurement, financial ratios

Article <top>

Comparing firm performance against other firms is a standard method of measuring firm or managerial efficiency. For example, a firm’s earnings per share is commonly used to compare firm performance vis-à-vis other firms in the same industry. Other financial ratios and other measures such as cost per unit of output have long been used to measure a firm’s relative performance (French). To make even more meaningful comparisons, firms are sometimes separated into alternative categories (two categories based on the top and bottom 50th percentile, four categories based on quartiles, etc.) based on their performance on profitability, cost, or productivity ratios. These measures are then used to help explain differences between the categories of firms. Examples of this approach from the literature include Babb and Keen; Schrader, Babb, Boynton, and Lang; Van Dyne and Rhodes; Plumley and Hornbaker; Akridge (1991); Holmes; Ginder and Henningsen; and Harris and Fulton.

Managers, board directors, and investors use such information to "benchmark" a firm’s performance relative to firms with similar production technologies who are recognized as industry leaders because of their ability to achieve lowest average costs or highest average profits. There are many sources of such benchmarking information. For example, the American Seed Trade Association (ASTA), in conjunction with Purdue University’s Center for Agricultural Business, collects accounting data from its members and develops industry standards for factors such as sales volume, gross margin, profitability, debt levels, and asset investment.

Benchmarking measures such as return on sales or cost per unit of output provide a firm with some evidence for comparing its performance against other firms. Efficiency measurement is another method used to evaluate a firm’s performance. Efficiency measures provide an indication of the increased profit or cost savings associated with a firm’s ability to operate as efficiently as the industry’s highest-profit or lowest-cost firm. Advantages of efficiency measures include controlling for output, input price, and fixed input effects for the sample of firms being analyzed. By controlling for such factors, firms are able to compare or benchmark their performance against firms producing the same level of output. In addition, these measures can be calculated for a single input, and managers can identify potential profit gains or cost savings for improving the efficiency of that input. Recent studies using such efficiency measures include Whittaker and Morehart; Featherstone and Rahman; and Preckel, Akridge, and Boland.

The objective of this research is to compare financial ratios and a productive efficiency measure, and determine their appropriate use in benchmarking. A sample of midwestern retail fertilizer plants is used for comparing both approaches. Firms are ranked according to their performance on each ratio or measure. Then each method for ranking firms is compared to determine whether there is any association between the ranking methods and whether a firm’s ranking in one year is related to its ranking in another year. Results should provide insight into the type of information provided by these two approaches, which may assist managers in making better informed decisions as to what types of performance measures may be most useful under alternative circumstances.

Alternative Benchmarking Methods <top>

Three common financial ratios used for benchmarking purposes are return on sales, return on assets, and the expense-revenue ratio. These ratios are used in reporting profitability and productivity. Return on sales (ROS) is net income (defined as total sales minus cost of goods sold and operating expenses) divided by total sales. Return on total assets (ROA) is net income divided by the accounting book value of total assets. The expense-revenue ratio (ERR) is net expenses before interest and taxes divided by total sales.

The relationships associated with these three ratios can be seen by using the cost of goods sold (COGS) to revenue ratio (cost of goods sold divided by total sales) and total asset turnover (total sales divided by accounting book value of assets, or TAT). Return on assets is equal to

(1)

and return on sales is equal to

(2)

Increases (reductions) in ERR reduce (increase) ROS, which then reduces (increases) ROA. In general, superior performance relative to the firm’s past performance is defined as a "high" ROS and ROA, and a "low" ERR. Calculating and evaluating these ratios provides a useful first step as a manager or board director analyzes the firm’s performance.

Farrell proposed a method whereby the total efficiency of a firm could be decomposed into allocative and technical efficiency components using a primal approach requiring knowledge of the firm’s production function. Kopp, and Kopp and Diewert described a dual method for determining these three efficiency measures without requiring information on the production function. Productive efficiency can be described as how near a firm comes to achieving some "best practices" or "theoretically optimal" standard. For example, the base objective might be the cost level achieved by a firm which produced the same amount of output at the lowest cost for the industry. Productive efficiency measures the cost savings that a firm could have potentially realized had it been as productively efficient as the firm with the lowest cost.

A firm is technically inefficient if it uses more inputs to produce the same level of output than the firm with the lowest input use. Two concepts are needed to define technical efficiency. Technically efficient cost (TC) is defined as the least cost input level required to produce a given level of output, holding the mix of inputs constant at the inefficient firm’s current input mix. The firm’s actual cost of inputs used to produce that same level of output is defined as OC. Technical efficiency (TE) is measured as the ratio of TC to OC, or

(3)

A firm is allocatively inefficient if it does not satisfy the cost-minimization criterion that the ratio of input prices must be equal to the marginal rate of substitution between those inputs when producing a given level of output. The lowest possible costs for a given output, allowing the input mix to adjust to input prices, is defined as FC. Allocative efficiency (AE) is measured as the ratio of FC to TC, or

(4)

Productive efficiency (PE) is defined as the product of technical and allocative efficiency (i.e., the firm equates the marginal rate of substitution between inputs and the price ratio of those inputs) or the ratio of the firm with the lowest cost, holding output levels constant, to the actual cost for the firm under consideration. This is represented as

(5)

If a firm uses the cost-minimizing input quantities to produce a given level of output, then the level of technical and allocative efficiency is equal to one. As firms begin to use more inputs or use inputs in the wrong proportions to produce the same level of output, the levels of technical and allocative efficiency, respectively, begin to decline to zero. For example, a firm with a productive efficiency measure of 0.95 could reduce costs by 5% if the firm were technically and allocatively efficient.

A measure of productive efficiency enables a firm to compare its performance against the lowest-cost (or highest-profit or highest-revenue) firm within an industry. The productive efficiency measure can be used by a manager analyzing the firm’s overall performance. If the firm is not efficient, as determined by use of this measure, then a manager can perform further diagnostics to explain the sources of inefficiency—which could be technical or allocative.

Data and Methods <top>

The data are taken from retail fertilizer plants enrolled in the annual Fertilizer Retail Efficiency Data (FRED) survey conducted by the Center for Agricultural Business at Purdue University. This survey has provided fruitful data for research. Akridge and Hertel estimated economies of scope and scale for multiproduct fertilizer plants, and reported that plants could reduce average costs by increasing output and diversifying into anhydrous ammonia. Akridge (1989) calculated efficiency measures and found that the average plant could reduce variable costs by 10% through improvements in technical efficiency.

Using 1991 survey data, Schulze reported that fertilizer managers who maintained a high level of quality with respect to agronomic services had higher profits. Holmes compared alternative financial ratios and found differences between high-profit and low-profit fertilizer and chemical firms. Preckel, Akridge, and Boland found that aggregation of inputs or outputs should be avoided when analyzing these data. A description of the data is presented in the Appendix.

The efficiency measures and financial ratios were calculated annually for each of the 24 retail fertilizer plants using 1975­82 data (Table 1). These cross-sectional, time-series data were chosen to avoid potential problems encountered when using accounting data as described by French and by Eidman. Retail fertilizer plants market the majority of their product during three months, thereby enabling the production and accounting data to match in any given year. Because the plants in this sample used consistent accounting procedures, and all plants employ the same depreciation method, accounting book value can serve as a relative indicator of plant and equipment investment. The depreciation method used by these plants is straight line depreciation with a common time horizon. The 1975­82 time period was chosen because there were no dramatic changes in technology that occurred in these plants over this eight-year span.

The estimation procedures employed by Akridge (1989) are used here to calculate the efficiency measures. Akridge minimized a variable cost function for given levels of output (outputs were dry fertilizer, fluid fertilizer, anhydrous ammonia, chemicals, services, and other farm supplies), variable input prices (input prices were labor and energy), and fixed inputs (fixed inputs were management, plant and equipment, and other fixed inputs). This frontier cost function yields the lowest cost for the outputs produced by the most productively efficient firms. Two equations for the variable input cost shares were estimated simultaneously with the variable cost function. Greene’s statistical deterministic translog cost function and share equation system was used, and the model was estimated using a maximum-likelihood procedure. This is a short-run variable cost function covering the fertilizer plant’s fiscal year.

Two hypotheses are tested in this analysis. The first hypothesis (H1) measures whether productive efficiency and the three financial ratios are associated with one another. The second hypothesis (H2) measures the ability of a firm to repeat its performance over time. The results of both hypotheses are needed when evaluating the use of these measures in benchmarking firm performance.

The null hypothesis for H1 is that there is no association between the productivity efficiency measure and the financial ratio rankings. A rejection of H1 would suggest that managers could continue to use financial ratios to benchmark their performance without the need for computationally complex efficiency measure calculations, because the two types of measures would yield the same rankings.

The motivation for this hypothesis is that each financial ratio incorporates information from other managerial decisions, such as pricing decisions or asset investment decisions, which are controlled for in a productive efficiency measure. Of the three ratios, ERR is closest to an efficiency measure. However, ERR incorporates the results of marketing decisions through its use of sales in the denominator. Productive efficiency is equal to the ratio of the lowest possible variable costs for a given output level holding inputs constant (FC) to the firm’s actual variable cost for producing a given level of output (OC), whereas ERR is the ratio of actual variable and fixed costs to sales. Thus, fixed cost efficiency is not included in the productivity efficiency measure and ERR contains no direct information about industry best practices. The results of this hypothesis provide evidence on whether variable cost efficiency is being measured through standard financial ratios.

A one-tailed test is used with the alternative hypothesis stating that the association between the productive efficiency measure and financial ratios is negative or positive depending upon the ratio being tested (Mendenhall, Scheaffer, and Wackerly, pp. 648­50). Each firm was ranked in each year highest to lowest for each of the ratios and the productive efficiency measures. Spearman’s rank correlation coefficient (r_s) is used to test H1 and is calculated as

(6)

Here, d_i is the difference in rankings and is measured as

(7)

where are the rankings for the individual efficiency measure and financial ratio, respectively, and n is the number of observations.

A higher productive efficiency measure suggests that firms would have higher levels of profit and/or lower costs if operating as efficiently as the most efficient firm in the sample. Thus, the productive efficiency measure is expected to be positively correlated with ROS and ROA, and negatively correlated with ERR.  A lack of association between the productive efficiency measure and each of the three ratios (failure to reject H1) would suggest that ratios do not appear to be good substitutes for the productive efficiency measure when benchmarking short-run variable cost efficiency.

The null hypothesis for H2 is that a plant has the ability to repeat its performance over time. The motivation for this hypothesis is to determine whether plant performance as measured by the productive efficiency measure or financial ratios can be repeated from year to year. This information is useful in assessing if any plants are truly superior performers, or chronic underperformers. Differences in the two types of measures in picking up year-to-year correlations in performance would also cast doubt on the ability to substitute one measure for the other.

The firm rankings are divided into three categories (each comprised of eight firms) based on their annual performance. The null hypothesis is that the firm has an equally likely chance (i.e., the probability is 0.333) of ranking in the top, middle, or bottom category each year. If the rankings are independent, then firms would not consistently rank in a particular group over time. A rejection of H2 would suggest that a firm’s performance in one year would more than likely be repeated in the next year. A test of the probability for ρ successes in a multinomial experiment is used to test the probability frequency distributions for the rankings on the efficiency and financial ratios (Mendenhall, Scheaffer, and Wackerly, pp. 217­18). The formula for determining the multinomial probability that a firm has a certain ranking over time is written as

(8)

where z_j is the number of times a firm ranks in category j, and Prob_j is the probability that a firm ranks in a category (Prob_j = 0.333). Only eight firms are allowed to rank in each category. The observed number of firms ranking in a particular category six or more times is determined. The cumulative probability of a firm achieving this is 0.0448, which is analogous to testing at the .05 level of significance for a one-tailed test. A test statistic is computed using the multinomial probability of success, ρ_j, and the observed frequency ρ. This Z-statistic is computed as

(9)

where σ_ρ is the standard deviation and is computed as

(10)

The results of H1 and H2 jointly provide information to a manager and should be evaluated together. For example, if the three financial ratios and the productivity measure are correlated (reject H1) and both methods provide similar insight into year-to-year correlations in performance (fail to reject H2), this would suggest that both methods are measuring the same general trends in performance.

Results <top>

Considerable variation in performance exists, as measured by the financial ratios (Figure 1) and efficiency measures (Figure 2). With regard to the financial ratios, ERR was most constant across firms and increased slightly over time (and thus ROS and ROA decreased).

In general (for 17 of the 24 firms), the correlation coefficients are not significantly different from zero (Table 2), suggesting a lack of association between each of the three financial ratios and the productive efficiency measure (failure to reject H1). The correlation coefficients between the productivity measure and ERR are negative, suggesting that increased costs are recorded when productivity levels are low. In contrast (with the exception of 1981), the correlation coefficients between return on sales (ROS) and productive efficiency are positive (net income is high when productive efficiency is high), and (with the exception of 1977) the correlation coefficients between return on assets (ROA) and productive efficiency are also positive. These results suggest that rankings are different based on each benchmarking method despite the tendency of the two types of measures to be correlated in the expected direction.

For example, when analyzing individual plant productive efficiency measures, plant 19 was ranked as most efficient with an average productive efficiency measure of 0.941 (i.e., the plant could have reduced variable costs by 5.9% if the plant attained technical and allocative efficiency). This plant also had the lowest ERR, highest ROS, and highest ROA among the 24 plants in this sample. Therefore, productive efficiency and the three ratios were consistent in their measurement for this specific plant. However, this was not the case for other plants. For example, plant 7 had the highest ERR and lowest ROS. Yet its productive efficiency measure placed it as the 7th most efficient firm. Thus, its expenses (which are considered high as a percentage of gross sales) would appear to be reasonable given the firm’s level of investment and output mix. Note that ERR measures more than a manager’s ability to control costs, while productive efficiency focuses on a single dimension of managerial performance. Consequently, inconsistencies between ratio and productive efficiency rankings do not appear unreasonable.

The second hypothesis, independence of firm rankings over time, was not rejected—implying that firm performance in one year tends to remain the same the next year for all four measures (Table 3). The observed frequencies were much higher than the expected frequency (1.08 firms). For these plants, high-performing plants remain high-performing despite external factors. Within the top and bottom categories, approximately 15% of the firms repeated their performance. This suggests that these plant managers, on average, were able to repeat their superior performance or could not correct their poor performance from year to year.

This information is important to a manager or board of directors when evaluating performance. For example, if a plant observes one "poor" year after a period of "good" years, drastic changes in operations may have to be examined carefully to ensure that the firm is not overreacting. Similarly, the ability of a plant to repeat its performance over time for all four measures may suggest that these measures are evaluating the same general industry trends.

Implications <top>

Firm performance as measured by financial ratios incorporates the results of several management decisions rather than a single decision as suggested by the productive efficiency measure. For example, financial ratios incorporate information concerning how a manager markets outputs and purchases inputs (i.e., incorporate "price effects"), while efficiency measures incorporate information on the variable inputs (i.e., remove "price effects"). Financial ratios, such as the three analyzed in this study, enable a manager to obtain an initial check on firm performance when making comparisons across time. While efficiency measures may be more mathematically complex to compute, they provide more detailed information on a firm’s use of variable and fixed inputs to produce some output level, and are appropriate for further in-depth diagnostic work. If a manager determines that the firm’s inputs are being used in an effective manner to produce some level of output, then analyzing a firm’s input procurement and marketing procedures may help identify further information on the source of lower revenues or higher costs as measured by the financial ratios.

References <top>

Akridge, J. T. "FRED User’s Manual." Bull. No. 608, Agr. Exp. Sta., Purdue University, West Lafayette, IN, 1991.

———. "Measuring Productive Efficiency in Multiple Product Agribusiness Firms: A Dual Approach." Amer. J. Agr. Econ. 71(1989):116­25.

Akridge, J. T., and T. W. Hertel. "Multiproduct Cost Relationships for Retail Fertilizer Plants." Amer. J. Agr. Econ. 68(1986):928­38.

Babb, E. M., and R. C. Keen. "Performance of Midwest Cooperative and Proprietary Grain and Farm Supply Firms." Bull. No. 366, Agr. Exp. Sta., Purdue University, West Lafayette, IN, March 1982.

Eidman, V. R. "Discussion: Methodological and Empirical Issues in Economies of Size Studies." In Economies of Size Studies, pp. 193­201. Center for Agricultural and Rural Development, Iowa State University, Ames, 1984.

Farrell, M. J. "The Measurement of Productive Efficiency." J. Royal Statis. Society, Series A (General) 120(1957): 253­81.

Featherstone, A. M., and W. H. Rahman. "Nonparametric Analysis of the Optimizing Behavior of Midwestern Cooperatives." Rev. Agr. Econ. 18(1996):265­73.

French, B. C. "The Analysis of Productive Efficiency in Agricultural Marketing: Models, Methods, and Progress." In A Survey of Agricultural Economics Literature, Vol. 1, ed., L. R. Martin, pp. 93­206. Minneapolis: University of Minnesota Press, 1977.

Ginder, R. G., and K. R. Henningsen. "Financial Standards for Iowa Agribusiness Firms, 1990­1993." Dept. of Econ., Iowa State University, Ames, 1994.

Greene, W. H. "On the Estimation of a Flexible Frontier Production Model." J. Econometrics 13(1980):101­15.

Harris, A., and M. Fulton. "Comparative Financial Performance Analysis of Canadian Co-operatives, Investor-Owned Firms, and Industry Norms." Centre for the Study of Co-operatives, University of Saskatchewan, Saskatoon, Saskatchewan, Canada, 1996.

Holmes, J. "Financial Benchmarking: An Application to the Retail Fertilizer and Chemical Industry." Unpublished M.S. thesis, Dept. of Agr. Econ., Purdue University, West Lafayette, IN, December 1994.

Kopp, R. J. "The Measurement of Productive Efficiency: A Reconsideration." Quart. J. Econ. 96(1981):477­503.

Kopp, R. J., and W. E. Diewert. "The Decomposition of Frontier Cost Deviations into Measures of Technical and Allocative Efficiency." J. Econometrics 19(1982):319­31.

Mendenhall, W., R. L. Scheaffer, and D. D. Wackerly. Mathematical Statistics with Applications, 3rd ed. Boston: Duxbury Press, 1986.

Plumley, G. O., and R. H. Hornbaker. "Financial Management Characteristics of Successful Farm Firms." Agr. Fin. Rev. 51(1991):9­20.

Preckel, P. V., J. T. Akridge, and M. A. Boland. "Efficiency Measures for Retail Fertilizer Dealers." Agribus.: An Internat. J. 13(July/August 1997): 497­510.

Schrader, L.F., E.M. Babb, R.D. Boynton, and M.G. Lang. "Cooperative and Proprietary Agribusinesses: Comparison of Performance." Res. Bull. No. 982, Agr. Exp. Sta., Purdue University, West Lafayette, IN, April 1985.

Schulze, T. L. "The Role and Management of Agronomic Services in Retail Fertilizer and Agricultural Chemical Firms." Unpublished M.S. thesis, Dept. of Agr. Econ., Purdue University, West Lafayette, IN, May 1993.

Van Dyne, D. L., and V. J. Rhodes. "Departmental Savings and Loss Characteristics for 12 Locally Owned

Farmer Cooperatives." Pub. No. SR 359, Agr. Exp. Sta., University of Missouri, Columbia, 1987.

Whittaker, G. W., and M. J. Morehart. "Measuring the Effect of Farm Financial Structure on Cost Efficiency." Agr. Fin. Rev. 51(1991):95­105.

Appendix: Description of Data <top>

For this study, annual cross-sectional (24 plants), time-series (1975­82) accounting data were used. Firm outputs were dry fertilizer, fluid fertilizer, anhydrous ammonia, chemicals, and services. Inputs included labor, energy, other variable inputs (net of energy and labor), and fixed inputs. Data were aggregated according to the procedures described in Akridge (1989).

Total variable costs were defined as those costs that change in response to changes in the level (and mix) of outputs being produced by a plant during the plant’s fiscal year. Total variable costs included outlays for labor, utilities, fuel and oil, advertising, repairs and maintenance, and miscellaneous operating expenses. Bad debt loss, depreciation, interest expense, and losses on sales of capital equipment were not included in total variable costs. Expenses not included in the analysis as either variable costs or fixed inputs averaged 20.6% of total operating expenses. Bad debt loss represents a write-off of those accounts that management has deemed uncollectible. Based solely on available data, it is impossible to determine the appropriate annual output figure to associate with bad debt losses.

The investment in plant and equipment is considered a fixed input here. Losses on sales of capital equipment were minimal. These expenses were either not relevant to the analysis or, due to accrual accounting procedures, were impossible to associate with the appropriate output figures. Unless otherwise noted, all producer price indices were obtained from the Bureau of Labor Statistics database.

Output levels for fertilizer, fluid fertilizer, and anhydrous ammonia were measured in tons per year. A measure of chemical output was constructed by dividing chemical revenue in dollars per year by an agricultural chemical price index. The measure of service quantity (acres per year) was calculated by dividing total service revenue by a weighted average price for fertilizer custom application. The measure of service output likely underestimates the true level of services provided by plants. Some dealers provide services such as soil testing or crop scouting at no cost, while the price of some services—especially delivery and custom application—may only reflect part of the cost of service. The remaining cost is covered in the price of the product. Such practices make accurate measurement of the true service output difficult. The output measure for the "other farm supply" category—typically more than 90% hybrid seed corn and seed soybean sales—was constructed by dividing other farm supply sales by an average seed price.

Input price data were not available from the individual plants. The state average weekly wage rate for employment in the trade sector was used as a proxy for the labor price. A weighted average energy price (cents/btu) was constructed using the North Central average commercial electricity rate and bulk gasoline price. The weights were the shares of total energy expenditures for utilities and fuel and oil, respectively. Given the range of cost items integrated into the "other variable input" category, price movements for the category were assumed to follow the general price level of the economy. The implicit GNP price deflator was used as a proxy for the price of other variable inputs.

A measure of the managerial input was obtained by adding total employee bonuses and profit sharing to the plant manager’s base salary. The bonuses were added as an adjustment for management quality differentials. The book value of plant and equipment as reported on the firm’s year-end balance sheet was divided by a machinery and equipment price index to calculate the measure of plant and equipment. The residual fixed input category (other fixed inputs) included insurance, rent and lease expense, local taxes and licenses, professional services, and other fixed overhead expenses. The sum of these items was divided by the implicit GNP price deflator to calculate the level of this input.

AEM Home Site Map Contact Us Cornell

© 2002 Cornell University
Department of Applied Economics and Management