BEYOND COMPLIANCE: SUSTAINABLE BUSINESS PRACTICES ...

4 downloads 17042 Views 159KB Size Report
have reduced compliance costs and a lower risk of future environmental liabilities (Konar and. Cohen 2001). But whether the line of causa-. Waugh Lecture.
BEYOND COMPLIANCE: SUSTAINABLE BUSINESS PRACTICES AND THE BOTTOM LINE DENNIS J. AIGNER, JEFF HOPKINS, AND ROBERT JOHANSSON

In recent years, there has been considerable research activity devoted to the relationship between the environmental performance and financial performance or stock returns of publicly traded companies. Also, many new “green” and “socially responsible” mutual funds have been created, as well as forprofit efforts that evaluate and rate corporate environmental performance and link it to stock returns.1 That there is a positive correlation between environmental performance and financial performance on a cross-sectional basis is clear for many industry sectors, and there is a premium observed when one compares stock returns over time for “good” versus “bad” environmental performers in these same sectors. Not surprisingly, the strength of the correlation varies across sectors, as does the observed stock returns premium attributed to good environmental performance. What seems to be happening is that the traditional notion of environmental compliance as a necessary cost of doing business is being transformed into something where pollution prevention, waste treatment, recycling, etc., and going “beyond compliance” represent either a direct cost savings or a competitive advantage (Porter and Van der Linde 1995a and Porter and Van der Linde 1995b). In addition, firms with better environmental records may have reduced compliance costs and a lower risk of future environmental liabilities (Konar and Cohen 2001). But whether the line of causa-

Waugh Lecture. Dennis J. Aigner is professor and dean at the Donald Bren School of Environmental Science & Management, University of California, Santa Barbara, CA. Jeff Hopkins and Rob Johansson are economists at the Economic Research Service of the U.S. Department of Agriculture. The views expressed in this article are those of the authors and not necessarily those of their institutions. The authors would like to thank Wade Brorsen, Nigel Key, Jorge Fernandez-Cornejo, and Robert King for their review and comments on earlier versions of this article. This article was presented as the Waugh Memorial Lecture at the Annual Meeting of the American Agricultural Economics Association in Montreal, Canada, July 30, 2003. 1 The most prominent of these are the rankings of Innovest Strategic Value Advisors and the Dow Jones Sustainability Index.

tion goes from environmental performance to financial performance or vice versa (the idea that financially successful firms can afford to be good environmental performers), or that both are the result of good management, is still an open question.2 Measuring environmental performance and management quality is still problematic, and only recently have firms begun to pay attention—publicly at least—to environmental issues as a central part of business strategy.3 Most of the literature on this subject to date falls into four categories. The first is portfolio studies wherein the stock returns of “good” environmental performers are compared to those of “bad” environmental performers. In this category are academic studies like Cohen, Fenn, and Konar and the evidence provided by the rating firms. In the Cohen, Fenn, and Konar study, industry-balanced portfolios of environmental leaders and laggards as of 1987– 1989 were constructed from the S&P 500 and the average financial performance was then tracked. In 80% of the comparisons, the lowpolluter portfolio performed better than the high-polluter portfolio. Adjusted for risk, the number was 75%. In both comparisons, however, the number of instances in which the difference was statistically significant was considerably less. Other efforts in this regard have grown out of the accelerating interest in socially responsible investing (SRI) here and abroad. In the United States, SRI has been growing twice as fast as mutual fund investing. And 79% of SRI funds screen (either in or out) on environmental performance. 2 Some recent studies focusing on disentangling the causal effects are King and Lenox, Khanna and Anton, and Molloy, Erekson, and Gorman. 3 By now, over 2,000 firms worldwide publish annual environmental reports and there is a push to make them transparent by conforming to a global reporting standard. Several organizations committed to good environmental performance also have surfaced, most notable among them the World Business Council for Sustainable Development, headquartered in Geneva, which counts among its members upwards of 170 of the world’s most prominent multinational firms.

Amer. J. Agr. Econ. 85 (Number 5, 2003): 1126–1139 Copyright 2003 American Agricultural Economics Association

Aigner, Hopkins, and Johansson

To service this growing demand for information on environmental performance, for-profit firms have sprung up. One of the first is Innovest Strategic Value Advisors, which rates public companies in a variety of industries on environmental performance and sells those ratings to the investment community. Their rating model relies on both publicly available data and qualitative assessments of future environmental and financial performance, and covers aspects of firm performance beyond just environmental. They also correlate environmental performance with stock returns. For example, in a recent analysis of the Electric Utility Industry, comparing the top 50% of the industry to the bottom 50% on the basis of environmental performance over a three-year period, the growth in average stock returns was 30 percentage points higher for the top environmental performers. Similar results are found for other industry sectors with considerable environmental exposure, some more dramatic, some less, but all in the same direction. In 1999, Dow-Jones in collaboration with Sustainable Asset Management Ltd. of Zurich launched the Dow-Jones Sustainability Group Index (DJSGI), a global effort that tracks the financial performance of the top 10% of sustainability-driven companies in sixtyone industries from twenty-seven countries. Their evaluation model is also broader than just environmental performance, covering economic and socio-cultural factors as well. A recent comparison covering a seven-year period showed that the DJSGI outperformed the Dow-Jones Global Index (DJGI) by 46 percentage points in terms of growth. The second category of research on the topic consists of various event studies that look at the impact of the environmental bad news on stock price. An early paper in this regard is White, where the impact of the Exxon Valdez oil spill on Exxon’s stock price was analyzed. White found not only an instantaneous and large impact but also one that was sustained over a long event period. Another good example is Konar and Cohen (1997), where the authors not only estimate the (negative) abnormal returns associated with the first Toxic Release Inventory (TRI) announcements in 1989 but also ask whether, as a result, firms reduced their emissions. Comparing TRI emissions data for 1992 to the 1989 benchmark, all forty of the firms with the largest negative abnormal returns in 1989 had lowered their emissions (appropriately normalized) by 1992. Third, several studies use cross-sectional data to explore the relationship between cor-

Sustainable Business Practices

1127

porate environmental and financial performance. Russo and Fouts argue that a proactive environmental policy offers the firm an opportunity to redesign its production and/or delivery processes, enhance its internal processes for waste reduction, and encourage organizational learning. Using a sample of 243 firms with environmental performance measured by compliance behavior, they conclude that better environmental performance is associated with better financial performance. Hart and Ahuja looked at the relationship between emissions reductions using TRI data and return on equity one and two years hence and found a significant positive relationship especially among firms with relatively high emissions compared to industry averages. An interesting twist on the subject is provided by Dowell, Hart, and Yeung, who ask whether adherence to a stringent global environmental standard is a benefit or a liability to multinational companies that establish operations in developing countries with less stringent local environmental standards. They found a positive and significant relationship between a firm’s market value and its level of environmental standards, and that a single, stringent global environmental standard was the most common corporate practice, thus dispelling in large part the notion of a “race to the bottom” in the behavior of multinationals. King and Lenox examined whether firms that operate in “cleaner” industries are primarily responsible for the observed positive relationship between environmental and financial performance4 and find that once industry effects are accounted for there still is a negative association between emissions levels (their measure of environmental (non) performance) and market value. A fourth category of research has recently emerged, that being theoretical studies. One of the most interesting is the work of Heinkel, Kraus, and Zechner, who investigate the conditions under which exclusionary ethical investing leads to fewer investors holding the stock of nongreen firms, leading to a lower stock price and a higher cost of capital. If this higher cost exceeds the additional cost of “reforming,” then polluting firms will reform. A key determinant in their model is the share of invested funds held by green investors. They find 4 In their study and that of Konar and Cohen (2001), performance is measured by Tobin’s q, which is the ratio of a firm’s market value to the replacement cost of its tangible assets. This measure is particularly attractive for the purpose at hand.

1128

Number 5, 2003

that if this share exceeds 20%, then there will be an effect.5 Evidence is accumulating in that part of our nonfarm economy governed by publicly traded companies in environmentally sensitive industries that good environmental performance is associated with good financial performance. But why is this relevant to agriculture? First, regulatory and policy trends have increasingly sought to reverse agriculture’s exemption from environmental regulatory measures. Second, it is necessary to find productivity-enhancing technologies to meet the demands of a growing global population. Third, the attributes of agricultural products, which extend to production processes, are increasingly important for determining consumer demand. Combined, these factors have resulted in a reassessment of the traditional comparative advantage that U.S. producers have held in global agricultural markets. The drivers facing agriculture are not that different from those facing other firms trying to plot out a sustainable growth path. Intuitively, one can see how an efficiently run enterprise would minimize unnecessary production inputs, which would both enhance the bottom line as well as minimize the impact from excess discharge of environmentally harmful inputs. There is also the incentive for firms to invest in environmental management today in the expectation that future forces (including regulatory, competitive, changing levels of demand or the preferences underlying demand) will require enhanced environmental performance. Our initial application is a preliminary formulation of the relationship between economic and environmental competitiveness within agricultural firms. Correlation between these two measures is both interesting and complex. Many questions spring to mind: “How does this relationship change across industries or regions?” “How have these measures changed over time?” etc. Clouding these issues are how local, state, and federal environmental regulations and expectations of more stringent regulations in the future might induce investments in environmental management. Agriculture and the Environment The agricultural sector is integrally linked to the environment because it is the predom5

Presently, the share of total invested funds in “socially responsible” accounts is less than 10% but is growing rapidly.

Amer. J. Agr. Econ.

inant use of land across the United States. Roughly half of all land in the lower fortyeight states is farmland and about a fifth of all land is devoted to crop production. Furthermore, crop production has become more intensive over time. Since 1960, wetland areas have decreased by more than 50% and the uses of industrial fertilizers and pesticides have increased by as much as 200% (Council on Environmental Quality). Animal production has also become more intensive, becoming geographically concentrated on larger facilities (e.g., hog production—McBride and Key). Both crop and animal production generate pollutants that enter the air, surface and ground waters. The Natural Resources Conservation Service estimates that the annual loss of soil from water erosion is approximately 1.14 billion tons per year (USDA, NRCS). Much of this runoff carries with it inorganic and organic fertilizers and pesticides. The U.S. Environmental Protection Agency (EPA) estimates that pollutants originating from agriculture runoff contribute to 60% of the pollution in impaired river areas, 30% in impaired lake areas, 15% in the impaired estuarine areas, and 15% in the impaired coastal shoreline assessed (U.S. EPA 2002). However, agricultural production has been mostly immune from environmental regulation (Ruhl).6 Why? First, there are many logistical difficulties in regulating agricultural impacts on the environment. Much agricultural pollution is “nonpoint source.” That is, many agricultural pollutants arrive via disperse transport mechanisms, whether through runoff, through groundwater leaching, or through the atmosphere. Therefore, it is difficult to assign blame to an individual for such things as excessively high levels of pollutants in a stream or lake. Nevertheless, while the amounts of pollutants leaving a particular farm in a particular year may not be “excessive,” at higher levels of aggregation and over time these pollutants contribute to a significant degradation of U.S. air, water, and soil resources. While Ruhl notes that farming may have escaped regulatory attention due to its persistence as a human activity and the relatively long periods required to quantify significant environmental impacts, he concludes on page 10204 that . . . the cumulative effects of more than 450 years of crop and livestock farming in 6 Obvious exceptions are the requirements placed on large animal feeding operations by the National Pollutant Discharge Elimination System (NPDES); see U.S. EPA (2003).

Aigner, Hopkins, and Johansson

America are no longer obscure; if we continue to leave farms unregulated, it is by choice, not by ignorance. One would be hard pressed to identify another industry with as poor an environmental record and as light a regulatory burden. . .

There are four noteworthy instances of federal regulations intended to mitigate farming’s environmental impacts (Ervin et al., p. 11). These are the Food Quality Protection Act of 1996, which enables the EPA to regulate pesticide use; the Federal Water Pollution Control Act of 1972, which requires landowners to obtain a permit from the U.S. Army Corps of Engineers before altering wetlands linked to navigable waters; the Endangered Species Act, which allows the federal government to restrict agricultural practices as part of species recovery plans; and, lastly, the Clean Water Act, which allows the federal government to control animal production practices on concentrated animal feeding operations. That said, the effectiveness and actual use of these laws to mitigate agriculture’s environmental impacts is limited to a small percentage of U.S. agriculture (Ervin et al.).7 The questions raised above regarding the links between economic performance on U.S. farms and their environmental management are, therefore, especially pertinent. We can effectively rule out the “race-to-the-bottom” argument as suggested by Palmer, Oates, and Portney, which would predict diminished economic performance of farms in heavily regulated regions, and diminished environmental performance of farms in not heavily regulated regions. We then hypothesize a frontier relationship between farms’ economic efficiency and their investments in environmental management as suggested by Schaltegger and Synnestvedt. The strength of this relationship is expected to vary across regions and farm type, or even across time. To examine the interaction between economic performance, environmental management, farm and farmer characteristics, we essentially require three things: a definition of farm economic performance, a definition of environmental management or performance, and sufficient data to test our hypotheses.

7 Farm Bill provisions such as Sodbuster and Swampbuster, which link agricultural support payments to environmental practices, have been treated by some as de facto federal environmental regulations on crop producers (Ruhl).

Sustainable Business Practices

1129

Specific Application—Corn Production Data—The Corn Sector We start with the last of these first. We choose to examine the linkages between economic efficiency and environmental management among U.S. corn producers. Corn production for grain occupied approximately 76 million acres in 2001, was responsible for approximately $19 billion in revenue, and constituted over 75% of the total grain produced in the United States (USDA, NRCS 2003). In addition to covering a large percentage of U.S. farmland (approximately 25%), corn production uses more than 40% of the total quantity of commercial fertilizer applied to crops (Christensen). Because of both the coverage and the relative intensity of production on corn acres, the environmental management practices of corn producers has a significant bearing on the overall environmental performance of U.S. agriculture. The data used comes from the fall and spring Agricultural Resource Management Survey (ARMS) of U.S. farm operators. ARMS targets the population of farms within the fortyeight contiguous states, where a “farm operation” is defined as an establishment that sold or would normally have sold at least $1000 of agricultural products in a year. ARMS is often said to follow a complex sample design (see www.ers.usda.gov/briefing/ARMS and documents referenced therein) describing both how the target population is identified and how samples are drawn from this population. This complexity impacts both how estimates are constructed and the margins of error around these estimates. Surveyed farms have unequal probabilities of being selected for ARMS, and multiple sampling frames, using stratification and clustering procedures, are used to gather sufficient sample sizes to achieve reliability of the estimates. Full consideration of the sample design of ARMS is given to the estimates included herein. All data used come from the 2001 ARMS, focusing on the environmental management decisions and economic efficiency of corn producers. The 2001 ARMS gathered detailed data on production practices for corn, including the use of management practices and detailed costs and returns of the corn operation in isolation from the rest of the farm. Our survey of producers that planted corn with the intention of harvesting it for grain includes 1,544 observations. These observations are weighted in

1130

Number 5, 2003

Table 1.

Amer. J. Agr. Econ.

Samples—2001 Corn ARMS Survey

ERS Region Heartland Northern Crescent Northern Great Plains Prairie Gateway Other Total

States a

a

a

a

a

IL, IN, IA, KY , MN, MO , NE , OH , SD MI, MNa , NY, OHa , PAa , WI COa , MNa , NEa , ND, SDa COa , KS, NEa , TXa COa , GA, KYa , MOa , NC, OHa , PAa , TXa

Full Sample

CRM

Non-CRM

896 309 53 147 139 1544

604 133 32 101 51 921

292 176 21 46 88 623

a These states straddle more than one ERS region.

such a way that they expand to represent 94% of all acres planted to corn for grain (full coverage is not possible because detailed corn data were drawn from the nineteen highest producing states, rather than the entire contiguous United States). Farm Environmental Management Rainfall and snowmelt can cause significant erosion on cornfields, which has been linked to a number of environmental problems (Ribaudo; Feather, Hellerstein, and Hansen). A suite of management technologies, often termed best management practices (BMPs), has been developed to mitigate the environmental impacts of grain production. Such practices include alternative fertilization, tillage, rotation, and pesticide regimes. Conservation tillage or crop residue management (CRM), has been identified as being a “best management practice” for corn production (Baker, Saxton, and Ritchey; Reeder) because it significantly decreases soil loss and agricultural runoff by protecting the soil surface and slowing and filtering runoff. There are many types of tillage operations that assist in crop residue management (examples include ridgetill, mulch-till, and no-till) but they all have the common feature of leaving the previous crops’ residues on the soil surface rather than removing it through tillage operations. For the purposes of our study, we use the broad categories of conventional tillage and conservation tillage, classifying corn producers as either users or nonusers of a crop residue management practice. The criterion used for this classification is an estimate of crop residue at the time of planting relative to a threshold of 15%. It is estimated that if 15% of corn crop residue is left on the field that soil erosion is decreased by one-third relative to the case of bare soil (Sandretto and Bull). Implementing greater rates of crop residue management will reduce soil erosion further still, but at a dimin-

ishing rate (Reeder). In our corn grain sample from 2001, approximately 60% of farms reported using crop residue management that retained at least 15% residue coverage on the surveyed field (table 1). We choose to look at environmental management via CRM rather than environmental performance through runoff or erosion because environmental management is more reflective of the effort put forth by the operator to manage resources. Akin to looking at firms with an endogenous supply of environmental services, regardless of whether there is a local demand for such services, we look at the endogenous profit effects of environmental management efforts. Farm Economic Efficiency In the nonfarm sector, profitability can be inferred from stock prices and dividend values assuming that stock prices reflect all information related to the profitability of a corporation including, but not limited to, financial statements prepared under current generally accepted accounting principles. Alternatively, it can be measured directly via return on investment, return on assets, or earnings per share. Because many farms are organized as sole proprietorships, or partnerships, by definition they have no outside shareholders and therefore no independent valuation of their economic performance. Therefore we construct a proxy for economic performance using the market value of their purchased inputs and outputs, and the opportunity costs associated with their factors of production. The measure we use to characterize farmer financial performance is total resource costs per dollar of corn output per farm. Resource costs include cash costs of production and noncash costs associated with corn production. The economic efficiency measure accounts for all production inputs, without regard to the ownership or equity positions of farm

Aigner, Hopkins, and Johansson

operators. These include both variable and fixed cash expenses, except interest payments; capital replacement; and the opportunity cost of land, labor, and capital inputs used in the corn operation. Variable cash expenses are out-of-pocket costs paid in cash, and depend on production practices and on quantities and prices of inputs. Fixed cash expenses are allocated to corn based on corn’s relative value of production within the overall farming operation. Imputations are used for the opportunity costs of labor (valued at an estimate of the offfarm wages paid to farm operators working off-farm), land (cash rental value), and capital (valued at the cost of replacing the capital investment in machinery and equipment used up in the production process). Total economic costs are divided by the value of output to allow for comparisons of efficiency across different sizes of corn operations. Figure 1 shows the cumulative density functions of economic efficiency for corn producers using conservation tillage on their surveyed

Sustainable Business Practices

1131

acres alongside the cumulative density function for producers using conventional tillage. The abscissa shows the economic efficiency of production and the ordinate shows the share of all corn farms for each population subgroup. Residue management is shown to be everywhere (i.e., at every point in their respective distributions) more efficient than conventional tillage. The corresponding share of farms producing corn for grain under breakeven or better conditions (defined so as to cover all economic costs, including the opportunity costs of owned capital and operator labor) is small, less than 25% for either group. Our estimate of economic efficiency does not reflect the influence of government payments because not all payments received are linked explicitly to the corn crop. Some payments, including Production Flexibility Contract payments, are lump-sum transfers to the household and therefore do not impact commodity returns in the same way as price supports. Alternative efficiency measures,

Figure 1. Economic efficiency of crop residue management (CRM) and non-CRM corn farms, 2001

1132

Number 5, 2003

allowing just for government payments specifically attached to production to influence efficiency, raised the share of producers at breakeven by about 6% for both groups. Likewise, using only cash costs results in more farms achieving breakeven levels but does not change the ordering of groups. However, because of heterogeneity in human and owned capital across producers, efficiency orderings of individual operators within a group may change. Methodology—The Choice of Environmental Management Figure 1 could be used as evidence that in addition to the beneficial environmental effects of crop residue management, CRM technology also increases the economic efficiency of production. A number of studies have examined this issue, noting that decreased costs of labor, equipment, and fuel associated with conservation tillage generally outweigh any potential yield decrease and increase in pesticide costs (USDA, ERS, 2003b; Sandretto and Bull; Lazarus and Selley). Economic benefits to CRM appear to be greatest in regions such as the Plains States, where CRM provides beneficial moisture retention for dryland grain production (McBride). Despite the economic and environmental benefits of crop residue management, only about 60% of operators employed CRM in 2001. Many have asked why adoption is not universal, given the benefits. Randall et al. suggest that CRM may not be attractive to producers who lack the crop management skills to adopt conservation tillage or who may have recently purchased equipment for aggressive tillage; or because the farmer is concerned with uncertain yields or reduced economic returns under conservation tillage. Several studies have examined this issue in more detail in order to understand what might explain why more farmers have not adopted crop residue management (e.g., land tenure— Soule, Tegene, and Wiebe; compatibility with pesticide use—Fuglie; farm size—Westra and Olson). Our hypothesis builds on these studies, examining how farm efficiency and the decision to employ crop residue management are related. We would like to evaluate the relationship between economic efficiency and environmental management in agriculture, distinguishing the effect of economic returns

Amer. J. Agr. Econ.

on the decision to invest in environmental management from the effect of investing in environmental management on financial returns. To identify their magnitudes, we will need to control for various farm and farmer characteristics. We propose that a farmer decides to invest in conservation tillage by comparing probable economic returns under the two tillage regimes and that the decision to invest in environmental management is endogenous to the economic returns of the particular field in question. This is similar to wage determination studies that control for endogeneity between the observed wage and explanatory variables such as union membership (see, e.g., Budd and Na; Johansson and Coggins). Consider the reduced form utility decision (Vi ) of a farmer to employ (or not) crop residue management: (1)

Vi = X i a˜ + εi

where (2)

Vi ≤ 0

for those farmers that choose conventional tillage; that is, CRMi = 0 and (3)

Vi > 0

for those farmers that choose crop residue management; that is, CRMi = 1, and Xi is a vector of various farm and farmer characteristics. Splitting our sample into the group of producers who utilize CRM and those who do not, we specify the relationship between farm and farmer characteristics and economic returns, Ri , as: (4)

lnRci = c0 + X ci aˆ c + ci

and (5)

lnRni = n0 + X ni aˆ n + ni

where the subscript “c” represents the subset of farmers employing crop residue management and “n” represents those that do not. We note that (4) and (5) can be combined as (6)

lnRi = n0 + (c0 − n0 )CRMi + X i aˆ + i

when there is no relationship between the error term (i ) and the decision to employ crop residue management. In this case the effect of environmental management on economic efficiency can be determined using a dummy

Aigner, Hopkins, and Johansson

Sustainable Business Practices

variable for crop residue management, and is measured by (c0 − n0 )CRMi . Several methods have been used to address the inherent endogeneity and potential bias of regression coefficients in a model like (1)–(5). A conventional solution is to allow for nonrandom assignment into the CRM and non-CRM subsamples along the lines of previous studies (Heckman; Lee; Budd and Na; McBride and Key). Following Budd and Na, we first estimate (1)–(3) using a probit model. This generates the probability that a farmer decides to use conservation tillage or not depending on a range of explanatory variables. The resulting coefficients yield the inverse Mills ratios (IMRs) necessary to correct for endogeneity when estimating the economic returns equations. We then estimate economic returns as a function of farm and farmer characteristics, for conservation tillage adopters and nonadopters, respectively: (7)

lnRci = c0 + X ci aˆ c1  ˆ ˆ + ci + c2 [(X i a)/(X i a)]

and (8)

lnRni = n0 + X ni aˆ n1 + n2 [−(X i a)/(1 ˆ − (X i a)] ˆ + ni .

Here, a c1 and an1 are vectors of the estimated changes in economic returns resulting from a marginal change in the explanatory variables for the two subsamples. The parameters aˆ are the estimates from the first-stage probit and the IMRs are given by (·)/(·) for the subset of farmers employing CRM and by −(·)/(1 − (·)) for those that use conventional tillage. (·) and (·) represent the cumulative distribution function and density function of a standard normal variable. c2 is the covariance between εi and  ci ; n2 is the covariance between εi and  ni . Testing c2 = 0 and n2 = 0 indicates whether there is selection bias occurring in the subsamples. Lastly, c and n are normally distributed error terms. Graphically we can depict these relationships using the Schaltegger and Synnestvedt concept of an environmental protection and economic success frontier. For our case, corn production and conservation tillage, we depict an environmental management—economic efficiency frontier for farms participating in the 2001 ARMS survey of Corn Producers (figure 2). First, we know that there is a difference in economic efficiency for adopters relative to nonadopters (figure 1). This difference is essentially given by Ad1 − Nad0 . To assess whether the frontier for nonadopters is monotonically increasing in conservation tillage, we can multiply nonadopter explanatory variables

Economic Efficiency

Non-adopters (Nad) All Farms (AF)

Nad0 AF 0 Ad 0

Nad 1 AF 1

Adopters (Ad)

Ad 1

CRM

1133

Environmental Management

Figure 2. Environmental management—economic efficiency frontier

1134

Number 5, 2003

Amer. J. Agr. Econ.

by the regression coefficients for adopters, to get Nad1 − Nad0 . Similarly, to determine how economic returns are impacted by environmental management for adopters we can multiply explanatory variables for adopters by the regression coefficients for nonadopters, which yields Ad1 − Ad0 . Lastly, we can multiply both sets of regression results by the all-farm sample variables to determine another measurement, AF1 − AF0 . Variables In order to isolate the relationship between environmental management (CRM) and economic efficiency from that of other factors, we include observation-specific data on farmer, farm, and field characteristics that affect both measures. Variables identified as likely having effects, based on heterogeneity among farmers, include their age, whether they work in farming only part-time or otherwise have limited economic resources, their willingness to trade off risk for expected returns, and their use of contracting relative to spot marketing. Mean values are shown in table 2. Data for 2001 show that nearly 60% of producers use crop residue management and that they enjoy a $0.31 premium in economic efficiency over nonadopters (they are 15.7% more efficient). In addition, adopters were on average older than nonadopters; less likely to be part-time or limited resource farmers; less likely to sell their corn for cash on the open market without some form of price protection; and willing to give up more in terms of maximum yield from their field if they could be showed a way to minimize the variability of yield. Table 2.

Heterogeneity among farms producing corn is expected to impact both the economic efficiency of corn production and the probability of adopting CRM strategies. Variables identified include the size of the corn operation and regional variability in cropping patterns and other geographic variables. 2001 ARMS data show that CRM adopters on average plant over 100 acres more corn than nonadopters. Adopters were especially common in the Heartland and Prairie Gateway regions but not in the Northern Crescent, where it is commonly believed that tillage is a prerequisite to optimal management because it warms up and dries out soils (that are especially cold and wet) before planting. Heterogeneity among cornfields is also hypothesized to be a driver of both economic efficiency measures and the probability of adoption. This type of heterogeneity should be picked up by the actual recorded corn yield as well as the yield goal, or the ex ante quantity of corn that the field can produce. Additionally, the use of other production technologies can sometimes correct for field heterogeneity. Two important technologies include land drainage practices to remove excess water from soils, and the use of precision agricultural technologies (using an “instantaneous” yield monitor) when harvesting corn. ARMS data indicates that both actual yield and yield goals are higher for adopters than nonadopters, and that the use of drainage and yield monitors is higher for CRM adopters. Additional data reveal that CRM adopters are also more likely to operate land designated as “potentially” highly erodible in the absence of some sort of environmental management and were more likely to have

Sample Means for 2001 ARMS Data

Variable CRM user Economic efficiency Corn acres planted Operator age Limited-resource part time Received cost-share Drainage Actual yield Yield goal Owned share of total corn acres Yield certainty tradeoff Sold corn in cash market Precision agriculture user Corn acreage is highly erodible Used no-till in the past

All Corn Farms

Adopters

Nonadopters

0.596 1.78 224 52.3 0.246 0.0258 0.376 127 140 0.529 5.75 0.465 0.159 0.190 0.237

1 1.66 273 52.4 0.238 0.0351 0.407 131 143 0.535 6.17 0.455 0.196 0.242 0.332

0 1.97 151 52.1 0.259 0.0119 0.330 121 135 0.519 5.12 0.478 0.103 0.113 0.0957

Aigner, Hopkins, and Johansson

Table 3.

Sustainable Business Practices

Results of Economic Efficiency Regression Without Selection Effects, Full Sample

Explanatory Variable Corn acres planted Operator age Limited resource/ part-time farm Northern Crescent region Northern Great Plains region Prairie Gateway region Other region Received cost-share Drainage Actual yield Yield goal Owned share of total corn acres Yield certainty tradeoff Sold crop for cash Precision agriculture user CRM user Constant R2 a b

1135

Coefficient t-statistic

With Selection Effects, CRM Adopters Coefficient

−2.6E-05 0.00646 0.302

−0.476 2.50a 2.48a

0.0809 −0.177

0.700 −1.74b

−0.218 −0.123

0.0280 0.0303 −0.0857 0.0575 −0.0194 0.00609 0.0949

0.399 0.255 −0.725 0.656 −9.30a 3.14a 1.69b

−0.00154 −0.0977 −0.0306 −0.144 3.05

−0.409 −1.76b −0.559 −3.40a 13.8a

0.4038

Without Selection Effects, Non-CRM Adopters

t-statistic

Coefficient

t-statistic

−0.000210 0.00473 0.446

−1.87b 0.819 2.63a

−1.72b −0.730

−0.183 −0.267

−1.14 −1.14

0.0359 −0.376 0.221 0.129 −0.0161 0.00630 0.155

0.363 −1.44 1.54 1.24 −7.12a 3.72a 1.08

−0.0821 −0.119 −0.335 −0.0553 −0.0208 0.00699 0.0400

−0.431 −0.449 −1.78b −0.346 −6.09a 1.90b 0.271

0.00666 −0.0668 0.150 N/A 1.58

2.41a −0.478 1.71b N/A 4.68a

−0.00289 −0.161 −0.164 N/A 3.40

−0.293 −1.99a −1.70b N/A 14.0a

0.000227 0.00859 0.170

NA

a

2.51 2.21a 1.11

0.4047

Statistically significant at the 5% level. Statistically significant at the 10% level.

used no-till, a form of CRM, for corn or any other crop since 1998. Results Estimates We first present our initial estimates, implementing the restrictive assumption that the sample of adopters and nonadopters are both randomly drawn from the same population of farm operators. Using the pooled sample of adopters and nonadopters, this is implemented through separate models, first with an OLS regression to estimate the effect on economic efficiency of adopting CRM, and second with a probit model of adoption to estimate the effect of CRM on economic efficiency.8 Results are found in tables 3 and 4, column 1. Because samples are chosen in multiple phases, and because the clusters and strata are not fixed across phases, Kott’s delete-a-group jackknife procedure, a replication-based method, was used to calculate the reported standard errors rather than the classical variance formula. The general idea of replication methods is to 8 Note that while equations (4)–(6) and, by implication (1)–(3), are specified in terms of returns, they are implemented in terms of our measure of economic efficiency.

draw repeated subsets of the sample, calculate the estimator for each subset, and then estimate the variance based on how much the estimates vary across the subsets. Beginning with the estimation of equation (6), the full-sample regression of economic efficiency in corn production, we find that CRM adoption increased economic performance, at a mean savings of $0.14 per dollar of output over not adopting. Important determinants of economic efficiency were age of the operator and whether they were limited-resource or part-time farmers, both of these characteristics decreasing economic efficiency in corn production. Corn production in the Northern Crescent region is also less efficient relative to Heartland region corn production. Also significant in our regression were both yield variables, with higher ex ante yield goals being negatively associated with economic performance and actual corn yield being positively associated with economic performance. The probit model shows that increased economic efficiency in corn production is associated with adoption of CRM, and that prior use of no-till and operating highly erodible land increases the probability of adopting CRM. An alternative recursive formulation, using an instrument for economic efficiency (constructed from predictions based on the previous

1136

Number 5, 2003

Table 4.

Amer. J. Agr. Econ.

Results of Probit Adoption Equation Without Selection Effects, Full Sample

Explanatory Variable Corn acres planted Operator age Limited resource/ part-time farm Northern Crescent region Northern Great Plains region Prairie Gateway region Other region Received cost-share Drainage Actual yield Yield goal Owned share of total corn acres Yield certainty tradeoff Sold crop for cash Precision agriculture user Economic efficiency Used no-till in the past Operates highly-erodible land Constant R2 a b

With Selection Effects, CRM Adopters

Without Selection Effects, Non-CRM Adopters

Coefficient

t-statistic

Coefficient

t-statistic

Coefficient

t-statistic

N/A N/A N/A

N/A N/A N/A

0.000455 0.00115 0.170

3.91a 0.305 1.10

−0.000580 0.000262 −0.0722

−3.77a 0.0611 −0.439

N/A N/A

N/A N/A

−0.232 −0.0669

−2.24a −0.389

0.508 0.0835

4.72a 0.415

N/A N/A N/A N/A N/A N/A N/A

N/A N/A N/A N/A N/A N/A N/A

0.281 −0.365 0.0571 0.0982 −0.00478 0.000677 0.233

2.44a −1.92b 0.396 1.11 −3.82a 0.382 1.99a

0.0186 0.581 −0.365 −0.0135 0.000643 −0.000920 −0.259

0.150 2.65a −1.34 −0.142 0.461 −0.541 −1.63

N/A N/A N/A −0.109 0.878 0.495

N/A N/A N/A −2.85a 8.86a 3.55a

0.00704 −0.0107 0.112 N/A 0.274 0.146

2.04a −0.0741 1.27 N/A 4.07a 1.53

−0.00995 0.0329 −0.162 N/A −0.877 −0.446

−2.35a 0.180 −1.45 N/A −9.73a −2.81a

1.57

0.169

0.663

0.240

1.03

0.176 .0836

NA

.1329

Statistically significant at the 5% level. Statistically significant at the 10% level.

regression) was also constructed to check for feedback from the efficiency equation on the adoption decision. The results, not reported, were somewhat changed: reducing costs per dollar of output was even more important (−0.166 vs. −0.109) and the standard errors were slightly smaller, with little difference in the coefficients on historical no-till use and the highly erodible variable. This promising result indicates that endogeneity of returns on the adoption decision is something to be aware of. Accordingly, we now relax the assumption that our samples of adopters and nonadopters are both random draws from the same pool of operators. Table 3, column 2 shows our results (using equations (1) and (7)) for the group of operators who have adopted CRM, using the full information maximum likelihood (FIML) method. Testing whether c2 = 0, significant positive correlation is detected, indicating that variables positively or negatively influencing CRM adoption also positively or negatively influence economic efficiency, and that we underestimate these effects when CRM adoption and economic efficiency are not jointly estimated. Comparison with the earlier, full-

sample results provides some useful insights. Variables that have changed little include the effect of operator age on efficiency and the yield goal of the operator, which continue to be negatively associated with economic efficiency. In addition, the association between higher yields and greater efficiency remains. Compared to the dummy variable method, however, the Heckman FIML technique shows that increasing planted acres was not conducive to increases in efficiency (previous results showed zero effect). Additionally, the results show that those who are willing to give up some deterministic quantity of yield in return for a more certain yield are at the same time giving up some cost efficiencies. Also relevant is that the use of precision agriculture, an informationintensive management technology, does not contribute to increased efficiency. These results lead us to believe that some strategies that are commonly advocated to improve efficiency, including planting more acres, reducing yield risk, and using precision agriculture may not be as effective for those farmers employing crop residue management. Rather, operators that sort themselves by management ability, including environmental management, may be

Aigner, Hopkins, and Johansson

better off without resorting to these activities, which are shown to increase costs relative to returns. Table 4, column 2, shows the FIML results estimated for the probit equation. Again, we compare the results with the independent probit reported in column 1, and see that the Heckman correction results in several differences. Many of the trends in variables are consistent with the story told in the efficiency equation. Greater acres did not have such a strong effect on the probability of adoption, as shown in the independent probit. Two regional differences when sample correction is used stand out. First, the coefficient on adoption in the Northern Crescent region, while still indicating a lower probability of adoption, was much smaller. Second, the probability of adoption in the Prairie Gateway region was larger. The operator’s preference for less risk in exchange for less yield was still shown to be important, but smaller in magnitude. Results for selectivity among nonadopters of CRM were not supportive of joint estimation, as the errors of the two regressions showed only weak negative correlation. Accordingly, in tables 3 and 4, column 3, we report the OLS regression results and the probit, but these were not estimated independently. Note that the variables impacting economic efficiency for nonadopters were often the same as for adopters, including that higher yields improved efficiency, but high yield goals did not. Also, cash marketings improved efficiency. An independent probit estimated for the nonadopters showed that the same variables influencing adoption negatively influenced adoption among the non-CRM sample, that is, the smaller, Northern Crescent and “Other” regions relative to the Heartland were more likely to be non-CRM, and non-CRM operators were unlikely to give up much yield for increased certainty. Premiums Comparing the fitted values for economic efficiency with alternative returns estimates are illustrative for the purposes of showing the efficiency gains to environmental management. Recall from table 2 that the sample mean economic efficiency was 1.78, or $1.78 in economic costs for each $1 in output. The group of conservers was on average more efficient by $0.31 per dollar of output relative to nonadopters ($1.97–$1.66). Holding all other variables in

Sustainable Business Practices

1137

the model constant, as we did for the results in equation (6), controlled for some of the differences in the means and showed that producers adopting CRM were more efficient by $0.14 per $1 of output. Correcting for sample selection, that is, letting the characteristics of the field, farm, and farmer determine the sample of CRM adopters and nonadopters, resulted in a mean economic efficiency estimate of $1.05 for all corn producers. This represents an unmeasured premium for adoption of $0.73 (i.e., $1.78–$1.05) on average, or an estimated cost savings of 41%, much higher than what is predicted from a simple comparison of means or an independent regression prediction. Conclusion In this paper we have made the first attempt— to our knowledge—at examining the two-way relationship between environmental performance in agriculture and economic efficiency. We build on earlier analyses that have examined the determinants of crop residue management and a growing literature devoted to documenting and explaining the positive relationship between environmental performance and financial performance within and across many industrial sectors, especially those with considerable environmental exposure. Our study is in the vein of recent work by Khanna, Kumar, and Anton, who investigate the relationship between technical and environmental efficiency and environmental self-regulation, which includes the adoption of environmentally friendly management practices. The adoption of a specific environmental management system, such as the industrial standard ISO 14001, need not imply improved environmental performance, however. Primarily this is because environmental performance is multifaceted and improvement in one dimension may come with degradation in another, at least in the short run. So it is with the crop residue management practices that are the subject of this study: While soil erosion is clearly reduced by their employment, they usually require additional pesticide use. Crop yields may also be negatively impacted, which can explain why such practices are not universally adopted. In our sample, for instance, 40% of the corn farms are nonadopters. This is analogous to the reason why many U.S. industrial firms have not yet adopted ISO 14001: The costs of certification

1138

Number 5, 2003

outweigh the nominal and perceived long-term benefits.9 More generally, improved environmental performance in the most environmentally impactful industries over the past fifteen years has been the result primarily of environmental regulation and the innovation that has taken place as a result of the need to comply with those regulations. Those efforts that go beyond compliance, of which there are an accelerating number of examples, suggest that industry has the capacity to be a major force in improving environmental conditions, not just in the United States but worldwide. Agriculture will no doubt come under increasing scrutiny with regard to its environmental impacts and further regulation is inevitable unless it can be demonstrated that farmers, as prime examples of stewards of the land, are already acting in environmentally responsible ways. References Baker, C., K. Saxton, and W. Ritchie. No-Tillage Seeding: Science and Practice. Wallingford: CAB International, 1996. Budd, J.W., and I. Na. “The Union Membership Wage Premium for Employees Covered by Collective Bargaining Agreements.” Journal of Labor Economics 18(2000):783–807. Christensen, L. “Soil, Nutrient, and Water Management Systems Used in U.S. Corn Production.” Economic Research Service, Agricultural Information Bulletin No. 774, 2000. Cohen, M.A., S.A. Fenn, and S. Konar. “Environmental and Financial Performance: Are They Related?” Unpublished working paper, Owen Graduate School of Management, Vanderbilt University, 1997. Council on Environmental Quality. Environmental Quality: 25th Anniversary Report. Washington DC, 1996. Dowell, G., S. Hart, and B. Yeung. “Do Corporate Global Environmental Standards Create or Destroy Market Value?” Management Science 46(2000):1059–74. Ervin, D.E., C.F. Runge, E.A. Graffy, W.E. Anthony, S.S. Batie, P. Faeth, T. Penny, and 9 Many firms that are good environmental performers already have environmental management systems in place, so the incremental benefit of ISO 14001 certification is perceived to be very low. That ISO 14001 is not tied directly to improved environmental performance is another announced reason for its lukewarm reception in the United States. A recent study by Russo looks at the relationship between ISO 14001 adoption and toxic emissions reductions in the Electronics Industry, and finds that especially high producers of emissions are benefited by ISO 14001 adoption.

Amer. J. Agr. Econ.

T. Warman. “Agriculture and the Environment: A New Strategic Vision.” Environment 40(1998):8–40. Feather, P., D. Hellerstein, and L. Hansen. Economic Valuation of Environmental Benefits and the Targeting of Conservation Programs: The Case of the CRP. Washington DC: U.S. Department of Agriculture, Economic Research Service AER-778, April 1999. Fuglie, K. “Conservation Tillage and Pesticide Use in the Cornbelt.” Journal of Agricultural and Applied Economics 31(1999):133–47. Hart, S.L., and G. Ahuja. “Does It Pay to Go Green? An Empirical Examination of the Relationship Between Emission Reduction and Firm Performance.” Business Strategy and the Environment 5(1996):30–37. Heckman, J. “The Common Structure in Statistical Models of Truncation, Sample Selection and Limited Dependent Variables and a Simple Estimator for Such Models.” Annals of Economic and Social Measurement 5(1976):475– 92. Heinkel, R., A. Kraus, and J. Zechner. “The Effect of Green Investment on Corporate Behavior.” Journal of Financial & Quantitative Analysis 36(2001):431–49. Johansson, R.C., and J.S. Coggins. “Union Density Effects in the Supermarket Industry.” Journal of Labor Research 23(2002): 673–84. Khanna, M., and W.R.Q. Anton. “Corporate Environmental Management: Regulatory and Market-Based Incentives.” Unpublished working paper, University of Illinois, ChampaignUrbana, 2001. Khanna, M., S. Kumar, and W.R.Q. Anton. “Environmental Self-Regulation: Implications for Environmental Efficiency and Profitability.” Unpublished working paper, University of Illinois, Champaign-Urbana, 2002. King, A., and M. Lenox. “Does It Really Pay to Be Green? Accounting for Strategy Selection in the Relationship between Environmental and Financial Performance.” Journal of Industrial Ecology 4(2001):105–16. Konar, S., and M.A. Cohen. “Information as Regulation: The Effect of Community Right to Know Laws on Toxic Emissions.” Journal of Environmental Economics and Management 32(1997):109–24. ——. “Does the Market Value Environmental Performance?” Review of Economics and Statistics 83(2001):281–9. Lazarus, W.F., and R. Selley. “Minnesota Farm Machinery Economic Cost Estimates for 2002.” Extension Publication No. FO-66962002), University of Minnesota.

Aigner, Hopkins, and Johansson

Lee, L.F. “Unionism and Wage Rates: A Simultaneous Equations Model with Qualitative and Limited Dependent Variables.” International Economic Review 19(1978):415–33. McBride, W. “The Economics of Crop Residue Management in U.S. Corn Production.” Paper presented at the Southern Agricultural Economics Association Meetings, Memphis TN, 30 January–3 February, 1999. McBride, W.D., and N. Key. Economic and Structural Relationships in U.S. Hog Production. Washington DC: U.S. Department of Agriculture, Economic Research Service, Agricultural Economic Report No. 818, 2003. Molloy, L., H. Erekson, and R. Gorman. “Exploring the Relationship Between Environmental and Financial Performance.” Unpublished working paper, Miami University, 2002. Palmer, K., W. Oates, and P. Portney. “Tightening Environmental Standards: The Benefit-Cost or the No-Cost Paradigm.” The Journal of Economic Perspectives 9(1995):119–32. Porter, M., and C. Van der Linde. “Green and Competitive.” Harvard Business Review (Sept.–Oct. 1995a):20–34. ——. “Towards a New Conception of the Environment-Competitiveness Relationship.” Journal of Economic Perspectives 9(1995b):97–118. Randall, G.W., N.B. Senjem, T.L. Wagner, L.M. Busman, and J.F. Moncrief. “Tillage Best Management Practices for Water Quality Protection in Southeastern Minnesota.” University of Minnesota, Extension Publication No. BU07694, 2002. Ribaudo, M. Reducing Soil Erosion: Offsite Benefits. Washington DC: U.S. Department of Agriculture, Economic Research Service, Agricultural Economic Report No. 561, September 1986. Reeder, R., ed. “Conservation Tillage Systems and Management: Crop Residue Management with No-till, Ridge-till, Mulch-till, and Strip-till.” MidWest Plan Service Publication No. MWPS45, Iowa State University, 2000. Ruhl, J.B. “The Environmental Law of Farms: 30 Years of Making a Mole Hill Out of a Mountain.” Environmental Law Reporter 31(2001): 10203–223. Russo, M.V. “Institutional Change and Theories of Organizational Strategy: ISO 14001 and Toxic Emissions in the Electronics Industry.”

Sustainable Business Practices

1139

Unpublished working paper, Department of Management, University of Oregon, 2001. Russo, M.V., and P.A. Fouts. “A Resource-Based Perspective on Corporate Environmental Performance and Profitability.” Academy of Management Journal 40(1997):534–59. Sandretto, C., and L. Bull. Crop Residue Management and Tillage System Trends. Washington DC: U.S. Department of Agriculture, Economic Research Service Publication No. SB930, August 1996. Schaltegger, S., and T. Synnestvedt. “The Link Between ‘Green’ and Economic Success: Environmental Management as the Crucial Trigger Between Environmental and Economic Performance.” Journal of Environmental Management 65(2002):339–46. Soule, M., A. Tegene, and K. Wiebe. “Land Tenure and the Adoption of Conservation Practices.” American Journal of Agricultural Economics 82(2000):993–1005. U.S. Department of Agriculture, Natural Resources Conservation Service. Cropland Acreage, Soil Erosion, and Installation of Conservation Buffer Strips: Preliminary Estimates of the 1997 National Resources Inventory. 2003. Available online at http://www.nrcs.usda.gov/technical/ land/pubs/buffer1.html. U.S. Department of Agriculture, Economic Research Service. Agricultural Resources and Environmental Indicators. ERS Agricultural Handbook No. AH722, Washington DC: 2003b. Available online at: http://www.ers.usda. gov/publications/arei/arei2001/. U.S. Environmental Protection Agency. National Pollutant Discharge Elimination System (NPDES). 2003. Available online at http:// cfpub.epa.gov/npdes/. ——. 2000 National Water Quality Inventory. Office of Water Report No. EPA-841-R-02-001, 2002. Available at: http://oaspub.epa.gov/waters/ national rept.control. Westra, J., and K. Olson. “Farmers’ Decision Processes and Adoption of Conservation Tillage.” Staff Paper No. P 97-9, Department of Applied Economics, University of Minnesota, 1997. White, M.A. “Investor Response to the Exxon Valdez Oil Spill.” Unpublished working paper, McIntire School of Commerce, University of Virginia, 1995.