Truth-telling by Third-party Auditors and the Response of Polluting Firms: Experimental Evidence from India*

TL;DR: A two-year field experiment in the Indian state of Gujarat as mentioned in this paper showed that the status quo system was largely corrupted, with auditors systematically reporting plant emissions just below the standard, although true emissions were typically higher.

Abstract: In many regulated markets, private, third-party auditors are chosen and paid by the firms that they audit, potentially creating a conflict of interest. This article reports on a two-year field experiment in the Indian state of Gujarat that sought to curb such a conflict by altering the market structure for environmental audits of industrial plants to incentivize accurate reporting. There are three main results. First, the status quo system was largely corrupted, with auditors systematically reporting plant emissions just below the standard, although true emissions were typically higher. Second, the treatment caused auditors to report more truthfully and very significantly lowered the fraction of plants that were falsely reported as compliant with pollution standards. Third, treatment plants, in turn, reduced their pollution emissions. The results suggest reformed incentives for third-party auditors can improve their reporting and make regulation more effective. JEL Codes: Q56, M42, D22. Copyright 2013, Oxford University Press.

Summary

Introduction

• 1 Einleitung und Problemstellung Steigender Wettbewerbsdruck, sinkende Margen und wirtschaftliche Stagnation zwingen viele Unternehmen dazu, ihre Investitionen in Informationstechnologie (IT) zu überdenken.
• So betru- gen sie beispielsweise bei der Deutschen Bank im Jahr 2002 lediglich 27% [Lamb02].
• 73% des IT-Budgets wurden für die laufende Produktion (Betrieb, Wartung, Support) und die Weiterentwicklung existierender IT-Lösungen aufgewendet.

Tätigkeiten im Customizing

• Innerhalb kürzester Zeit hat sich dieses Buch als Standardwerk etabliert.
• Leichte Verständlichkeit, Beispielorientierung und überschaubarer Umfang der Darstellung, also known as Die Vorzüge.
• Alle Komponenten des Controlling-Moduls werden komplett und gut nachvollziehbar erläutert.
• Das klar strukturierte Konzept führt den Leser anhand einer Fallstudie auch ohne Vorkenntnisse zu einem umfassenden Verständnis.
• Die Zusammenhänge werden Schritt für Schritt erläutert: Von den Grundlagen der Kostenrechnung und des Controlling über die Funktionsweise der SAP-Software bis hin zur Anwendung im praktischen Betrieb.

Figures

TABLE II Audit Treatment Covariate Balance

Figure 2: Audit Treatment Effect in Density Bins, All Pollutants

TABLE VI Incentive Pay from Treatment Effect Over Time

Figure 3: Time Series of Audit Reports by Treatment Status, All Pollutants

TABLE I Submission of Audit Reports

TABLE V Audit Treatment Effects on Auditor Reporting

TABLE IV Compliance in Audits Relative to Backchecks by Treatment Status

Figure 4: Quantile Treatment Effects of Audit on Endline Pollution

TABLE III Compliance in Audits Relative to Backchecks, Control Group Only

Figure 5: Regulatory Actions by Degree of Violation

Figure 1: Readings for Suspended Particulate Matter (SPM, mg/Nm3), Midline

TABLE VII Endline Pollutant Concentrations on Treatment Status

Full text

Truth-telling by Third-party Auditors
and the Response of Polluting Firms:
Experimental Evidence from India
Esther Duflo, Michael Greenstone, Rohini Pande,
Nicholas Ryan
CID Working Paper No. 262
May 2013
 Copyright 2013 Duflo, Esther; Greenstone, Michael; Pande,
Rohini; Ryan, Nicholas; and the President and Fellows
of Harvard College
at Harvard University
Center for International Development
Working Papers

Truth-telling by Third-party Auditors and the Response of
Polluting Firms: Experimental Evidence from India
∗
Esther Duflo
†
Michael Greenstone
‡
Rohini Pande
§
Nicholas Ryan
¶
May 2, 2013
Abstract
In many regulated markets, private, third-party auditors are chosen and paid by the firms
that they audit, potentially creating a conflict of interest. This paper reports on a two-
year field experiment in the Indian state of Gujarat that sought to curb such a conflict by
altering the market structure for environmental audits of industrial plants to incentivize
accurate reporting. There are three main results. First, the status quo system was largely
corrupted, with auditors systematically reporting plant emissions just below the standard,
although true emissions were typically higher. Second, the treatment caused auditors to
report more truthfully and very significantly lowered the fraction of plants that were falsely
reported as compliant with pollution standards. Third, treatment plants, in turn, reduced
their pollution emissions. The results suggest reformed incentives for third-party auditors
can improve their reporting and make regulation more effective.
∗
We thank Sanjiv Tyagi, R. G. Shah, and Hardik Shah for advice and support over the course of this
project. We thank Pankaj Verma, Eric Dodge, Vipin Awatramani, Logan Clark, Yuanjian Li, Sam Norris and
Nick Hagerty for excellent research assistance and numerous seminar participants for comments. We thank the
Sustainability Science Program (SSP), the Harvard Environmental Economics Program, the Center for Energy
and Environmental Policy Research (CEEPR), the International Initiative for Impact Evaluation (3ie), the
International Growth Centre (IGC) and the National Science Foundation (NSF Award #1066006) for financial
support. Ryan thanks an AMID Early Stage Researcher Fellowship for financial support. All views and errors
are solely ours.
†
MIT, eduflo@mit.edu
‡
MIT, mgreenst@mit.edu
§
Corresponding author. Harvard Kennedy School, Harvard University, Mailbox 46, 79 JFK St., Cambridge,
MA 02138, rohini_pande@harvard.edu
¶
Harvard, nickryan@fas.harvard.edu
1

I Introduction
The use of third-party auditing to monitor the compliance of firms with regulation is ubiqui-
tous. Third-party audits are the norm in financial accounting, and in many countries credit
ratings from third-party agencies serve an important regulatory role (White, 2010). Con-
sumer and commodity markets use third-party auditors to monitor standards, including those
for food safety, healthcare, flowers, timber and many durable goods (Hatanaka et al., 2005;
Raynolds et al., 2007; Dranove and Jin, 2010). With respect to environmental regulation, the
focus of this paper, several countries use third-party auditors to verify firm compliance with
national laws and regulations (Kunreuther et al., 2002; Paliwal, 2006). Third-party auditing is
also used to enforce international environmental standards, including ISO 14001 certification
and verification of carbon abatement in the carbon offset market (Potoski and Prakash, 2005;
Bhattacharyya, 2011).
These markets share a common characteristic—the auditor is chosen by, paid by and reports
to the audited firm. This feature creates a conflict of interest between reporting the truth and
reporting what is beneficial for the client. To maintain business, third-party auditors have
incentives to shade or falsify their reports, which may corrupt information provision and, in
turn, undermine regulation. Events brought to light by the recent financial crisis suggest this
is a real concern.
1
Yet, despite periodic calls for reform to increase the independence of third-
party auditors, we are unaware of a single instance of an enacted reform that fundamentally
alters the incentives of third-party auditors.
2
This paper reports on a two-year field experiment conducted in collaboration with the
environmental regulatory body in Gujarat, India. Since 1996, the state has had a third-
1
For overviews of problems in the U.S. corporate audit and credit ratings markets see Ronen (2010) and
White (2010), respectively. Biased reporting appears to be a key issue for credit rating agencies: for a single
credit agency, Griffin and Tang (2011) show higher accuracy of the internal surveillance team’s judgments
on CDO ratings than the business-oriented ratings team’s, and that the accuracy difference predicts future
downgrades. Strobl and Xia (2011) compared ratings for the same companies provided by two credit rating
agencies, where one agency uses a issuer-pay model and the other an investor-pay model. The difference in
ratings is more pronounced when the issuer-pay rating agency plausibly has more business at stake.
2
In 2002, the Sarbanes-Oxley Act made auditors of public companies subject to oversight by a private-sector,
nonprofit corporation. This corporation determines who can perform audits, conducts investigations and sets
fines. Three former SEC Chairmen testified in favor mandatory auditor rotation, which was not adopted. (The
Act also required SEC to report to Congress on credit rating agencies but did not reform this sector.) In 2003
the Securities and Exchange Commission adopted rules on auditor independence that focused on restrictions
on and disclosure of non-audit activities. In 2008 New York State Attorney General Andrew Cuomo reached
an agreement with credit rating agencies which required upfront payment for their ratings. The Dodd-Frank
financial reform bill and the Sarbanes-Oxley Act restrict the services that auditors or credit rating agencies
can offer plants that they audit.
2

party audit system for plants with high pollution potential, wherein certified auditors annually
submit pollution readings and suggested pollution control measures for the audited plants to
the Gujarat Pollution Control Board (Gujarat High Court, 1996). Although the Gujarat
High Court put in place several safeguards to limit conflicts of interest, the basic financial
arrangement underlying these audits is typical of the practice the world over—plants hire
and pay auditors directly, and the work of those auditors is subject to very little oversight. In
conversations we had before beginning this study, the regulators, auditors, and polluting plants
all agreed that the status quo audit system produced unreliable information. As evidence of
this common opinion, the reported market price for an audit was often lower than the cost of
collecting pollution readings, suggesting that at least some readings were not even taken.
Our experiment altered the market structure in several complementary ways in order to in-
centivize accurate reporting. All 473 audit-eligible plants in two populous and heavily polluted
industrial regions of Gujarat entered the experimental sample. In each region, half the plants
were randomized into a treatment with four parts. First, treatment plants were randomly
assigned an auditor that they were required to use. Second, auditors were paid from a central
pool, rather than by the plant, and their fee was set in advance at a flat rate, high enough to
cover pollution measurement and leave the auditor a modest profit margin. Third, a random
sample of each auditor’s pollution readings were verified with follow-up visits to the audited
plants by an independent technical agency that collected readings for the same pollutants at
the same places as the auditor, usually within a couple weeks of the auditor readings. We
refer to the follow-up visits as backchecks for the remainder of the paper. While the 20%
probability of a backcheck was public knowledge, actual backcheck visits were unannounced.
Fourth and finally, at the start of the second year, treatment auditors were informed that their
pay would be linked to their reporting accuracy, as measured by the backchecks. (During the
first year, we did not specify any explicit consequence of good or poor performance. Auditors,
however, may have anticipated lower chances of staying included in the scheme if found to be
systematically biased.)
We collate data from several sources. We collected all audit reports for years one and two
filed with the regulator. We directly obtained backcheck readings from the agencies conducting
backchecks. Towards the end of the second year, we hired the same technical agencies to do
identical backchecks in a random sample of control plants; these backchecks were unannounced
3

and not used to monitor or reward auditor behavior. The availability of auditor and backcheck
readings from the same plants and at nearly the same times offers a unique opportunity to
compare true pollution levels with the auditors’ reports of pollution in both the treatment and
control plants. Finally, roughly six months after the last audit visit in the experiment, we ran
an independent endline survey of pollution outcomes in all treatment and control plants.
We have three main findings. First, status quo audit reporting is corrupted, as auditors
systematically report plant pollution readings just below the regulatory standard. The average
difference between audit and backcheck pollution readings across all reported pollutants is -0.30
standard deviations in the control group. A comparison of audit and backcheck readings in
the control indicates that 29% of audit reports falsely report readings as below the relevant
regulatory standard. Further, much of this false reporting comes in the form of extra reports
just below the standard, which are presumably less likely to attract regulatory attention than
would be reports showing compliance by a wide margin.
Second, the treatment caused auditors to report more truthfully and reduced the frac-
tion of plants that were falsely reported as compliant with pollution standards. Relative to
backcheck readings, auditors for treatment plants report pollution readings that are 0.15 to
0.21 standard deviations—or 50% to 70% higher—than control auditors. This result is ro-
bust to the inclusion of auditor fixed effects, which allows us to compare the behavior of the
same auditors simultaneously working in both treatment and control plants. This, in turn,
suggests that the results are not due to a selection of different auditors in treatment versus
control plants. Further, auditors working in treatment plants are 23 percentage points or 80%
less likely to falsely report a pollution reading as in compliance with the relevant regulatory
standard.
Third, treatment plants reduced emissions, presumably because they understood that the
regulatory authority would receive more reliable audit reports. Average pollution in the treat-
ment group fell by 0.21 standard deviations, with reductions concentrated among plants with
the highest readings. We document that in practice, the regulator reserves the harshest penal-
ties for plants with readings that significantly exceed the standard, so it is not surprising that
that the dirtiest plants responded by reducing emissions the most.
The treatment, which included multiple parts, was implemented as a single package. Hence,
we cannot separately identify the effects of the treatment components—auditor assignments,
4

Frequently asked questions

Q1. What are the contributions in "Truth-telling by third-party auditors and the response of polluting firms: experimental evidence from india" ?

This paper reports on a twoyear field experiment in the Indian state of Gujarat that sought to curb such a conflict by altering the market structure for environmental audits of industrial plants to incentivize accurate reporting. First, the status quo system was largely corrupted, with auditors systematically reporting plant emissions just below the standard, although true emissions were typically higher. Second, the treatment caused auditors to report more truthfully and very significantly lowered the fraction of plants that were falsely reported as compliant with pollution standards. The results suggest reformed incentives for third-party auditors can improve their reporting and make regulation more effective. ∗We thank Sanjiv Tyagi, R. G. Shah, and Hardik Shah for advice and support over the course of this project. 

Q2. What are the future works mentioned in the paper "Truth-telling by third-party auditors and the response of polluting firms: experimental evidence from india" ?

First, where the authors have some variation, the non-experimental evidence suggests that incentive 28Compared to the results in Panel A of Table VII, the effect of the treatment on pollution emissions is smaller and insignificant in the selected sample of plants that submitted audits. The authors estimate gross costs to be around USD 1,300 per plant and, tentatively, gross benefits to be USD 7,300 per plant, for a net social gain of USD 6,000 per treatment plant. These estimates suggest that auditor-level effects, such as income or Hawthorne effects, do not drive the treatment impact. Third, economic intuition and existing evidence from the corruption and monitoring literature suggest that higher auditor pay without the other treatment components would not have changed auditor behavior much, given the absence of monitoring in the status quo. 

Q3. How many audit teams can audit a plant?

Audit teams can audit at most 15 plants per year, and an audit firm, which may employ several teams, can audit a plant at most three years in a row. 

Q4. How much did auditors and plants pay for an audit?

In interviews conducted prior to the experiment, both auditors and plants claimed that an audit report could be purchased for as low as INR 10,000-15,000 (roughly $200-$300). 

Q5. What is the polluted river in India?

Gujarat contains the two most polluted industrial clusters in India, and three of India’s five most polluted rivers (Central Pollution Control Board, 2007, 2009b). 

Q6. What is the penalty for failure to submit an audit?

On the other side of the market, for an eligible plant, failure to submit an audit is punishable by closure and disconnection of water and electricity. 

Q7. What is the effect of backchecks on auditors?

the introduction of backchecks increased auditor monitoring and likely raised the perceived likelihood that, although not explicitly part of the experiment, the regulator would disbar auditors who submitted false reports (or at least not assign them to treatment plants in year two). 

Q8. What was the effect of the treatment on auditors’ incentives to report?

the higher payments, relative to the control market pay, may have interacted with increased monitoring to further incentivize accurate reporting in the treatment, if treatment auditors decided to report more accurately because they had more to lose if disbarred from auditing. 

Q9. How many plants were assigned to the audit treatment?

Treatment plants were assigned to the audit treatment once, in 2009, for the audit years 2009 (hereafter year one) and 2010 (year two). 

Q10. What is the legal definition of a plant’s environmental consent?

A report showing noncompliance with the terms of a plant’s environmental consent can also be punished by closure or fine (Gujarat High Court, 1996). 

Q11. What did the treatment increase auditors’ incentives to file accurate reports?

assignment to a plant by an external authority and a fixed pay structure meant that auditor compensation was independent of what it reported. 

Q12. What is the way to close the loop?

And to close the loop, random assignment alone is insufficient if the price is not regulated to ensure that it covers the costs of performing an audit.