Regulatory capture, interest group theory, and institutional mediation: The regulatory politics of financial reporting rules, 1973-1987 Stephen J. Mezias Department of Management The Stern School, New York University 40 West 4 th Street, Room 7-04 New York, NY 10012 212-998-0229 (ph) 212-995-4234 (fax) [email protected]and Seungwha Chung Department of Management Yonsei University Korea
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Regulatory capture, interest group theory, and
institutional mediation: The regulatory politics of
In response to this victory of the accounting profession, the SEC since its inception has
delegated the determination of GAAP to an external agency. Our focus will be on the current
agency, the Financial Accounting Standards Board (FASB). Describing how the SEC delegated
authority to set accounting standards to the FASB, Skousen (1991: 118) wrote: The “... SEC
stated explicitly in ASR No. 150 that it considers those accounting principles, standards, and
practices promulgated by the FASB as having considerable authoritative support. ... The SEC
thus recognizes the FASB as the primary standard-setting body.” According to March and Olsen
(1989: 97), this use of organizations “... at the margin of the state has grown. Administrative
functions have been entrusted to semi-autonomous governmental or quasi-governmental
agencies.” Although the specific arrangement between the SEC and FASB may be unique, the
practice of delegating policy authority to an autonomous, quasi-governmental agency is not.
Carmichael and Willingham (1989: 11-12) described GAAP as a hierarchy with the Statement of
Financial Accounting Standards (SFAS) issued by the FASB at the top. Consistent with neo-
institutional arguments, the process to issue these rules is characterized by legalistic procedures
and a lengthy due process (Boland, 1982; Miller, Redding, and Bahnson, 1994; Tandy and
Wilburn, 1992). The formal records produced during the comment period, which commences
when the FASB releases an Exposure Draft describing a proposed regulation, are particularly
comprehensive. Interested parties have a period of weeks or months to write letters of comment
that the FASB will consider before releasing the final standard (Mezias and Chung, 1989; 1991).
For this reason we decided to get the data for our study from the comment periods.
In deciding which SFAS to include in our study, we began with the classification of
standards into substantive and non-substantive. Tandy and Wilburn’s (1992) classification
extends from the inception of the FASB in 1973 through December 1988. Upon review of the
18
institutional environment of financial reporting regulation, however, we decided to end our
sample a year earlier in 1987. The main reason was to avoid the effects of mergers among the
eight major firms that had dominated the certified public accounting profession since the
inception of the FASB (Wootton, Wolk, and Normand, 2003). To determine which comment
periods would provide the data for this study, we randomly chose 30 SFAS issued by the FASB
between 1973 and 1987. We based the distinction between substantive and non-substantive on
the work of Tandy and Wilburn (1992). Using this categorization, the thirty SFAS we selected
included five substantive standards and twenty-five non-substantive standards. To test the
hypotheses we needed a way to distinguish the issues that were decided in each of these SFASs.
The comment period always commences with the release of an Exposure Draft; we examined
these documents to determine the issues considered by the FASB in each Exposure Draft. These
documents characterize the decision-making in terms of various financial reporting practices and
present a rationale for why each of these financial reporting alternatives should or should not be
allowed under the new SFAS. The standards included in this study discussed a total of 151
financial reporting alternatives, which is the unit of analysis for this study.
Dependent, Independent, and Control Measures
Table 2 provides an example by outlining the alternatives considered in SFAS # 40; we
discuss several aspects of this information to illustrate how we created our measures. First, it
should be noted that this was a non-substantive standard in the category of those that are narrow
in application; this SFAS would only apply to firms that report holdings of timberlands on their
financial statements. Second, we familiarized ourselves with the context of each statement by
reading the Exposure Drafts and examining other secondary records for each SFAS, e.g., minutes
of any hearings or task force meetings. The four alternatives considered in this Exposure Draft
19
were the following: (a) Allow firms to report the value of timberlands using either historical cost,
actual expenditures, or current cost, what those expenditures would cost in current dollars. (b)
Require firms to report the value of timberlands using current cost value. (c) Require firms to
report the values of timberland using fair values. (d) Exempt firms holding timberlands from any
future requirement to report the value of timberlands using current cost accounting. Third, for
this SFAS, the first alternative was adopted, and all other alternatives rejected. However, in
some instances, the decision in the final SFAS was to allow multiple alternatives to be acceptable
under GAAP, which means that more than one alternative was accepted in the final draft. Of the
151 alternatives in our sample, 61 were accepted as part of the final SFAS; 90 were rejected.
Our dependent variable is a dummy variable called Choice: coded 1 if a particular alternative
was accepted in the SFAS and 0 if it was rejected.
To test our hypotheses we examine the association between this dependent variable and
the preferences of the organized actors for particular alternatives. This required developing a
methodology for measuring each organized actor’s preference for particular alternatives. The
ability to measure the actor’s preferences using their letters is a primary justification for studying
the comment period. Thus, we began at the FASB library where all letters written during the
comment period for every SFAS are maintained. The classification of comment letters by the
FASB maps directly onto our three organized actors. For our measure of the preferences of the
profession, we used their category called the public accounting profession; for firms, we used
their category called preparers, and for regulators, we used their category of government. For the
average SFAS, almost one hundred and eleven letters were written, but this distribution is right-
skewed; a few standards with many more letters drive up the average. For the case of standards
with many more letters than average, we believed that we could measure the preferences of
20
organized actors by looking at just a subset of letters. In doing this, we followed these
procedures: For Exposure Drafts with less than 100 letters of comment, all letters were read. For
Exposure Drafts with between 100 and 200 letters of comment, every other letter was read, and
we alternated starting with the first and second letter. For Exposure Drafts with between 200 and
300 letters of comment, every third letter was read; again, we alternated our starting point.
Next we developed a coding scheme to record positions taken in individual letters. Two
researchers began this task by reading fifteen letters randomly selected from the comment
periods of SFAS not among the thirty included in this study. Based on reading these letters, we
developed three straightforward coding rules. First, if a letter stated explicit support for or
opposition to a specific alternative, then it was recorded as favoring or opposing that alternative.
For example, a letter written in response to SFAS #40, which was presented in Table 2, might
state that the author opposed requiring current value accounting for timberlands. Such a letter
would count as opposing the third alternative in the table. Second, if a letter contained a clear
statement supporting the tentative conclusions of the Exposure Draft, it was coded as favoring
alternatives advocated in the Exposure Draft. Third, if a letter contained a clear statement
opposing the tentative conclusions of the Exposure Draft, it was coded as opposing alternatives
advocated by the FASB in the Exposure Draft. This coding scheme recognizes that one letter
may mention several alternatives; thus, a single letter may count as being in favor or against
multiple alternatives. Conversely, not all letters in response to an Exposure Draft mentioned all
alternatives. There were no letters for which a position coded based on one of these rules
contradicted a position coded by applying one of the others. To verify that these narrow, text-
based rules would result in measures with a high reliability, we chose an additional SFAS not
among those included in the study at random. Two researchers read all of the comment letters
21
received in response to the Exposure Draft of the chosen SFAS, which was # 22. For over 95%
of these letters, the assessment of the text was identical for the two coders. Following the
confirmation of this extremely high level of reliability and given the large number of letters, we
used single rater coding for the letters written regarding the 151 alternatives in our sample.
Because only statements that conformed directly to these three rules were counted in creating the
variables, there was a small number of letters, fewer than ten among more than one thousand
read for this study, for which we were unable to code any content. We decided to reread these
letters to see if changes to our coding scheme might account for their content. Ultimately, we
decided to make no changes to our rules, and these letters are omitted from the study as too
ambiguous to code.
A second important decision concerned how to combine the opinions coded from
multiple letters to create a measure of the overall opinion of the organized actors regarding
alternatives discussed in the Exposure Drafts. We assumed that members of the organized actors
were structurally equivalent, which is consistent with both an institutional perspective and how
these letters are handled during the due process of the FASB. By the end of the comment period,
the letters written by all professionals, firms, and regulators have been collected into binders,
sorted by category, are available for review. We considered other ways of aggregating letters.
For example, FASB personnel have espoused the view that the comment period is not an election
and argued that `high quality’ letters have much more impact (Beresford and Van Riper, 1992).
However, we believed that deviation from the assumption of structural equivalence should be
based on a clear theory and the availability of accurate measures. We lacked a compelling theory
of what constitutes a high quality letter to help us calculate a metric other than equal weighting.
Similarly, processes such as reputation or network centrality might imply deviations from equal
22
weighting of letters, but again we were unsure as to how we might proceed in determining how
to measure the effect of these processes.1 Ultimately, we concluded that for this first attempt to
study an institutional politics model of an ongoing process, we should treat entries into the public
record of the comment period from organized actors as equal; thus, all letters are counted as they
are presented in official records of the FASB: one for one.
We created a measure of support for each alternative for each organized actor,
professionals, firms, and regulators, by aggregated the coding of letters from all members of each
organized actor. Our measure calculates the number of net favorable, positive values, or net
unfavorable, negative values, as a percent of total letters using the following formula:
PERCENT =(LETTERS FOR – LETTERS AGAINST)
(LETTERS FOR + LETTERS AGAINST)
It could range from minus one, unanimous opposition, to positive one, unanimous support. This
gives us summary measures of the overall opinion of each organized actor with respect to each
alternative in the sample. Moreover, because this measure is a percentage, it is comparable
across alternatives and across organized actors. These values were calculated for all three
organized actors with respect to the 151 alternatives in our sample and are labeled Professionals,
Firms, and Regulators in our analyses. H1 through H3 are supported if these variables have a
positive, significant effect on the likelihood that an alternative was adopted by the FASB. H4
and H5 refer to the influence of Firms and Regulators over alternatives that were part of
substantive standards. To test them, we began by creating a variable called Substantive, which
was coded one if Tandy and Wilburn (1992) categorized the SFAS for which the Exposure Draft
was issued as substantive, and zero otherwise. To test H4, we examine the interaction between
this variable and the variables Firms and Regulators; these interaction terms estimate the
1 As we mention in our discussion of the results, we did rerun all of the analyses including only letters from Big 8 firms in creating the measure for the accounting profession; this made no difference to the results.
23
mediated effect of the latter two variables for instances of substantive regulations. H4 is
supported if the interaction term for Firms and Substantive, labeled SFirms, is negative and
significant. H5 is supported if the interaction term for Regulators and Substantive, labeled
SRegulators, is positive and significant. We also included the interaction term for
Professionals and Substantive, labeled SProfessionals, to see if it had a significant effect
despite our argument that it would not.
We included four variables to control for differences across alternative that we thought
might account for variation independent of the processes of theoretical interest to us. First, both
as a control for process differences and to allow for more precise estimation of the two
interaction effects that we incorporate for hypothesis tests, we include the variable Substantive.
This will control for any systematic differences in the probability that an alternative would be
adopted depending on whether it was included in the Exposure Draft for a substantive as
opposed to non-substantive standard. Second, we included two variables to account for
differences in the amount of attention that regulatory decisions attract. While we believed
including the dummy variable for substantive standards would control for some of the attention
effects, we also wanted to control for attention specific to the comment period we studied. As a
specific control for attention during the comment period, we introduced a variable called Letters
that is equal to the count of total letters submitted to the FASB in response to the Exposure Draft
for each of the rules in our sample.
We also wanted to control for attention that might be given to an issue prior to the
comment period. After an issue is admitted to the agenda of the FASB, decisions are made
regarding the necessity of engaging in any four possible due process steps, which generally occur
prior to issuing an Exposure Draft. Consequently, they can be seen as part of the context
24
preceding the issuance of an Exposure Draft that structures the attention from organized actors
reflected in the letters of comment. Possible due process steps include the following (Miller,
Redding, and Bahnson, 1994): (1) Forming a task force: The FASB may choose to ask persons
with various expertise on a particular issue to serve in a group that meets to consider the issue
placed on its technical agenda. (2) Issuing a discussion memorandum: The FASB may choose to
prepare a formal document outlining preliminary ideas and logic concerning the issue placed on
its technical agenda. (3) Having an invitation to comment: The FASB may publicly announce its
desire to hear from parties interested in the issue placed on its technical agenda as part of its
early deliberations on the issue. (4) Holding a public hearing: The FASB may choose to
schedule a public meeting for discussion of the issue placed on its technical agenda with
interested parties. We used a count of the total number of due process steps executed for a
particular standard as our measure of the structuring of attention; this variable is called
DueProcess. Finally, consistent with much of the literature on legal environments, Sutton and
Dobbin (1996: 799) noted that outcomes tended to be time dependent. Despite the fact that we
purposely constructed a window of time to avoid major process changes, we did examine FASB
decisions over a period of about fourteen years. To control for possible effects of the passage of
time during our study period, we introduce a control variable called Months. This variable
tracks the passage of time for alternatives in our sample with a count of the number of months
between the release date of the Exposure Draft that corresponds to a particular alternative and the
release date of the earliest Exposure Draft in our sample.
ESTIMATION AND RESULTS
To test the hypotheses, we ran maximum likelihood logistic regression models to predict
whether specific alternatives discussed in the Exposure Draft would be adopted in the final rule
25
issued by the FASB. The dependent variable was a categorical representation of the decision by
the FASB for each of the alternatives in the Exposure Drafts. This variable was coded zero if the
alternative was rejected in the final SFAS issued by the FASB; it was coded one if the alternative
was accepted. This categorical dependent variable is called Choice. The following equation
represents the basic model, which can be estimated using maximum likelihood logistic
where i designates the alternative. To test H1 through H3, we examine the significance and
direction of the coefficients for the variables named after each of the three organized actors.
These measure support (positive values) or opposition (negative values) for each alternative;
positive, significant coefficients are consistent with the hypotheses. To test H4 and H5, we
examine the coefficients of the interaction terms. If the coefficient of SFirms is negative and
significant, then H4 is supported; if the coefficient of SRegulators is positive and significant,
then H5 is supported.
Descriptive statistics for all the variables are reported in Table 3; a correlation matrix for
the variables is reported in Table 4. The values on all of the variables that measure the opinions
of the various organized actors are negative; this implies that organized actors on balance tend to
oppose alternatives more than they support alternatives. Apparently, being opposed to an
alternative is a greater motivation to writing a letter than supporting one. The mean of the
variable Substantive indicates that slightly more than one quarter of the alternatives, 38 of 151,
in the sample are from substantive standards. Not surprisingly, the scope of discussion in these
standards is broader, and more alternatives tend to be considered. The average alternative in our
26
sample was discussed in nearly one hundred and eleven letters. Across the fourteen-year period
of our study, the average alternative was included in an Exposure Draft issued 70 months, or just
under six years after the first. The alternatives included in our sample were not random across
time, but were likely to occur earlier in the time frame of our study. As we will discuss in
reviewing the results, this slowing down of standard setting by the FASB, already evident in
1987, has only become more evident in the ensuing years. The descriptive statistics also indicate
that the average alternative in our sample was derived from a due process that included only
slightly more than one of the four possible steps executed by the FASB.
Correlations among the independent variables are presented in Table 4. There are
significant and fairly large correlations among the independent variables for the several
regression equations that we need to run to test our remaining hypotheses, suggesting the
possibility of multicollinearity. To assess the potential effects on any of the coefficients, we
examined the variance inflation factors for the individual variables. While there is no formal
statistical rule for assessing the influence of multicollinearity on coefficients, the rule of thumb is
to assume that values exceeding ten indicate problems (Allison, 1999); none of the measures for
any of our variable were greater than eight, even in the full model; most were less than two.
Therefore, we conclude that despite the presence of some significant and fairly large
correlations, none of the estimates of the effects of the independent variables was affected
unduly by multicollinearity.
The correlation matrix also reveals some interesting relationships related to the
institutional politics model. First, all of the correlations among the opinions of the organized
actors are significantly greater than zero. Even for the substantive standards for which we
predicted that the influence of regulators will be enhanced and that of firms diminished, the
27
correlation between these two variables is positive. We attribute this to the fact that most of the
participants, even those acting as representatives of firms and as regulators, are certified public
accountants (Mezias and Scarselletta, 1994). The correlations also suggest that the strongest of
these positive relationships is that between the opinions expressed by firms and those expressed
by professionals. The single highest correlation is between the opinions of professionals and
firms for alternatives discussed in substantive standards; the second highest is between the same
two actors for alternatives discussed in all standards. It is also true, however, that correlations
between professionals and regulators tend to be high. The next highest correlation is between
professionals and regulators for alternatives discussed as part of substantive standards followed
by the correlation between the same two organized actors across all alternatives. As would be
expected the correlation between being part of a substantive standard and both the number of
letters written and the number of due process steps executed are positive and significant.
Table 5 presents the results for estimation of the maximum likelihood logit models of
Choice. We begin with a model that includes only the control variables to establish a baseline
for comparison with models including variables of theoretical variables. All of the variables as
well as the model itself are not significant. In fact, no control variable has a significant effect in
any of the models, and this model does worse than chance, classifying just over fifty percent of
the alternatives correctly as either accepted or rejected. Given that 90 of the 151 alternatives are
rejected, i.e., the dependent variable equals zero, a naïve model that simply predicts zero would
achieve nearly sixty percent correct classification. We proceed by adding the variables to test the
influence of the organized actors predicted by the first three hypotheses. All are supported in
one-sided tests against the null hypothesis of no effect; the variables Professionals (p < 0.001),
Firms (p < 0.05), and Regulators (p < 0.05) all have effects significantly greater than zero.
28
Since the variables are all measured in the same units, percentage support or opposition among
all letters sent by members of the respective organized actors, the effects are directly comparable.
The coefficients indicate that the opinion of the professionals group has about twice the effect of
either firms or regulators, which are approximately equal. The model overall is significant (p <
0.001) and achieves more than eighty-six percent correct classification. The estimates reveal
that the effects of shifts in the opinions of the organized actors on the likelihood that an
alternative would be adopted are quite dramatic. Figure 1 depicts these effects, with the lines
showing the effect of professionals, firms, and regulators shifting from uniform opposition to an
alternative (-1) to uniform support for an alternative (1).2 For professionals, the organized actor
with the largest effect, total opposition implies that the likelihood that an alternative will be
adopted by the FASB is barely a fifteen percent. Conversely, total support for an alternative
increases the likelihood of adoption to more than seventy-five percent. The effects for shifts in
the opinions of firms and regulators is similar but of slightly smaller magnitude. Total
opposition from either group reduces the likelihood of adoption of an alternative to about twenty-
five percent; total support from either group raises the likelihood of adoption of an alternative to
about sixty percent.
The final model adds the interaction variables to test the last two hypotheses. The model
overall is significant (p < 0.001); comparison of the test statistics for this model and the previous
model shows that the difference in chi-square is significant, suggesting that the addition of these
variables improves the overall fit of the model. This is also reflected by an increase in the
percentage of correctly classified cases to over ninety percent. Inspection of the coefficient for
the interaction of the opinions of the representatives of firms and substantive standards reveals
support for H4: SFirms is significant (p < 0.01) and negative. The alternatives advocated by
22 Both figures present the effects for each organized actor holding all other significant variables at their mean.
29
firms for substantive standards are less likely to be adopted. Inspection of the coefficient for the
interaction of the opinions of regulators and substantive standards reveals support for H5:
SRegulators is significant (p < 0.05) and positive. The alternatives advocated by regulators for
substantive standards are more likely to be adopted. Figure 2 depicts these effects, with the lines
showing the effect of professionals, firms, and regulators shifting from uniform opposition to an
alternative (-1) to uniform support for an alternative (1). For firms, the organized actor with the
largest effect, total opposition increases the likelihood the FASB will adopt an alternative to just
under ninety-seven percent. Conversely, total support for an alternative decreases the likelihood
of adoption to nearly one in one hundred. The effect for shifts in the opinions of regulators are
the second largest effect, with total opposition by this organized actor decreasing the likelihood
of adoption to just under seven percent; total support from regulators increases the likelihood of
adoption to nearly ninety percent. The effect of professionals is not significantly different for
substantive standards, but the effect of opinions from this group increases once the interactions
are introduced. As visual inspection of the figure reveals, the effect of opinions expressed by
members of the profession is similar to the effect of regulators, but of slightly lesser magnitude.
Reviewing all of the coefficients in the model reveals a most interesting aspect of these findings:
The interaction variables to test the last two hypotheses have the largest effects of any in the
model. The single largest effect is the negative effect of firms for substantive standards; the
second largest effect is the positive effect of regulators for substantive standards. Taken
together, these two effects indicate strong support for the notion that institutional mediation is
important to understanding outcomes in the ongoing regulation of financial reporting.
The variable for the main effect of the opinions of professionals remains significant while
the interaction with the indicator variable for substantive standards is not significant. This
30
indicates that there is no significant difference in the effect of professionals for substantive and
non-substantive standards. This is consistent with claims that professionals adapt to a changing
legal environment (Sutton and Dobbin, 1996) and that the imprimatur of normative support
(DiMaggio and Powell, 1983) is necessary for the passage of both substantive and non-
substantive standards. Interestingly, comparison of the coefficient for Professionals in the
models without and with the interaction effect reveals that inclusion of the interaction effects
increases the estimate of the coefficient by an amount approximately equal to the standard error
of the coefficient. We interpret this to suggest that the true effect of the variable is larger than
suggested by the first model. There is a similar, but even larger change in the coefficient of
Firms, which more than doubles in size after the interactions are added. Lastly, the coefficient
of the variable Regulators, loses significance after the interaction effects are added, indicating
that regulators are influential for substantive standards but have no significant effect on non-
substantive standards. Apparently, the significant, positive effect for this variable in the model
without the interaction terms was driven by its strongly positive effect for only that subset of
alternatives that were part of substantive standards.
Discussion
The support for the hypotheses derived from the institutional politics model is strong and
consistent. Regulatory capture, interest group theory, and institutional mediation are all
important parts of the story; taken together, they provide a more complete understanding of the
ongoing process of regulating financial reporting standards than would anyone of them alone.
These results are also robust to alternate specifications of the model. Using probit estimation
yields the same conclusions with respect to all of our hypotheses. Further, if we treat adoption of
an alternative as an event and use event history methods to estimate the same model, the results
31
do not change from what we reported using maximum likelihood logit. That said, we believe
that the way choices were made, explicitly rejecting or accepting alternatives at fixed points in
discrete time is actually most consistent with the logit framework. We also ran a model that
included only letters from the Big 8 accounting firms in calculating the opinion of members of
the profession. While the effect of this group was weakened slightly by this result, none of the
results for any of the hypothesis tests were changed.
We also estimated two alternative measures of the effects of institutional mediation. In
both cases, while we still found support for H5, indicating that the influence of regulators was
enhanced by institutional mediation, H4 was not supported. We interpret these twin results to
suggest that the support for H4 is more tentative than our other findings. The first alternative
measure of institutional mediation was the variable measuring the number of due process steps;
interactions of this variable and the influence variables leads to a failure to support H4.
Similarly, if we use the count of the number of letters rather than the indicator variable for
substantive standards, the interaction of firms and letters again is not significant, indicating a
failure to support H4. While these two results both suggest that support for H4 should be
regarded as more tentative, we also believe there are several reasons to conclude that the results
from the main model should not be rejected. First, in both cases the sign of the interaction is
negative, that is, in the correct direction. With a sample size of only 151, we may simply not
have enough power to distinguish effects that are subtler than the dichotomous distinction
between substantive and non-substantive standards.
Second, our examination of the data revealed underlying processes driving both due
process and the writing of letters that are independent of processes of institutional mediation of
direct interest to us. In the case of due process, the FASB will sometimes execute steps in
32
recognition of other factors that are at play. For example, there were three due process steps
executed for SFAS 40 and 41, which were not substantive rules, because they were linked with
the then ongoing project on current cost accounting. In essence, these hearings became a forum
to discuss the controversial issue of replacing historical cost accounting with current cost
accounting rather than being a forum for institutional mediation of influence over the content of
those rules. In the case of letters, there was also evidence of processes that might change their
quantity that were unrelated to processes of institutional mediation. For example, we discovered
that for one of the SFAS included in our study an entire undergraduate accounting class was
required to send letters of comment as an assignment by their professor. We think it is fair to
conclude that this participation did not represent facilitation of interest group influence by
processes of institutional mediation. Nonetheless, we do believe that future research should
regard the support for H4 as more tentative; we look for future designs to determine the limits of
our finding that firm influence over ongoing regulation is reduced by institutional mediation.
Future research might also address some other weaknesses in our study. A first is
highlighted by the timing argument that caused us to reject continuous time model: Groups of
our alternatives were decided together. This is related to the fact that we derived these 151
alternatives from thirty FASB rules. The first response to this potential problem is a theoretical
one; both the way the FASB discusses these alternatives as well as how letters written during the
comment period discuss them tends to focus on specific alternatives as choice points. Despite
the fact that Exposure Drafts discuss multiple alternatives, rationales are developed for or against
each alternative, which is accepted or rejected on its own merits relative to an ideal of financial
reporting practice. This is demonstrated by the fact that 61 alternatives, just above two per
regulatory decision, were accepted as part of the SFAS that we studied. Similarly, 90
33
alternatives, three per regulatory decision rule, were rejected in these same SFAS. The best
statistical method to handle this problem would be to include a rule specific fixed effect
parameter. Unfortunately, with only 151 decisions, the addition of twenty-nine fixed effects
parameters results in a failure of the model to converge, no matter which estimation
methodology we use. Given this, we used an alternative method to assess the extent to which
the observations deviate from the independent, identically distributed assumptions required by
the estimation model. We did this by putting the data matrix in order by SFAS so that
alternatives considered as part of a single rule making process were in consecutive rows. This
yielded a set of error terms where the alternatives that were part of the same SFAS are clustered
together. We then conducted two variations on the runs test (Bradley, 1968) to assess any
deviations from independence. In the first, we examined whether consecutive residuals deviated
from the equal likelihood of being positive and negative that would hold if the observations were
independent. In the second, we examined whether consecutive residuals deviated from the equal
probability of being above or below the median absolute value among all the residuals. The test
statistics for both lead to acceptance of the null hypothesis that the values of the residuals are
independent; in both sign and magnitude, the residuals from our estimation do not deviate from
what would be expected if choices among the alternatives were independent.
Another potential problem with our results is that we have examined only one stage in the
decision making process for financial reporting regulation. As Bachrach and Baratz (1962)
observed, power and influence in a decision making process may be exercised to prevent
decisions or issues from ever entering the agenda. Thus, the real power and influence in the
process for regulating financial reporting rules occurs by ensuring that issues which powerful
actors do not wish considered never enter the FASB agenda. This implies that studying the
34
formal process of FASB decision-making is to see only the tip of the iceberg. At least two
implications of this are important to interpreting our results. First, the results of Mezias and
Scarselletta (1994), who studied the agenda process of the FASB, revealed a process in which
the accounting profession and firms dominate just as we found for non-substantive standards.
These groups make decisions about the majority of financial reporting issues without invoking
the full regulatory apparatus of the FASB. Of the minority that makes it onto the formal agenda
of the FASB, which were those included in our study, the firms and the professions again
dominate over the majority. Thus, the substantive rules, for which regulators and professionals
are able to overcome firm opposition, is even smaller relative to the whole of issues than our
ratio of one in five indicates. Second, we wondered whether agenda control has the result that
meaningful and important financial reporting issues never enter the agenda of the FASB. On that
count, we think it is fairly clear that important issues do enter the agenda of the FASB. During
the time frame of our study the FASB was looking at accounting for health care costs and other
post-retirement benefits, accounting for pensions, and towards the end of our study period began
a project on accounting for stock options used in executive compensation. A brief perusal of the
business headlines twenty years’ later reveals that these are important topics.
CONCLUSIONS AND IMPLICATIONS
The strong results of this study in support of the institutional politics model as
applied to ongoing regulation suggest some conclusions and implications. In addition,
the choice of financial reporting standards as the context of the empirical study raises
issues as well. We begin with these by quoting a front-page story in the Wall Street
Journal (February 4, 2002: A1):
As Congress investigated the collapse of a high-profile energy company, it faced a daunting challenge. One senator said that to understand the huge company's
35
shocking failure, lawmakers must consider the regulatory and legal missteps that led to its downfall. How, he wondered, could Congress restore investors' confidence in the financial system? By repealing old laws? Enacting new ones? One of his colleagues answered by recounting an old joke: A man gets a message that his mother-in-law has died. “Shall we embalm, cremate or bury?” it asks. Replies the man, “Embalm, cremate and bury. Take no chances.”
As the failure to name the company hinted, the story is not about Enron, the huge energy
company that had collapsed the previous year; rather it dates from the 1930s and concerns the
bankruptcy of Middle Western Utilities. Scandals arouse the passion to reform and punish, as
did the financial crises of the 1930s; yet, seventy years of history suggest that reforms have not
prevented future crises. We would suggest policy choices based on what Stryker (2002: 176)
termed “… deep historical knowledge …” might be a more effective response. Our review of
the history of regulation of financial reporting in the US reveals that crises are part of a longer
process and need to be understood in that larger context. This is not to suggest that scandals are
not important turning points in regulatory processes or that they should not arouse passions for
reform or punishment. Rather, we would urge policy makers to avoid the urge to view these
crises as the unique result of a series of unusual circumstances and human error (Perrow, 1984)
and urge them to take a more systemic view of the problem (Mezias, 1994).
We interpret our results to suggest some suggestions for the design of institutions that
will mediate regulatory decisions. For example, those who would like to see a more effective
counterbalance to the powerful alliance of corporate and professional interests should try to
strengthen the links between regulators and interest groups in all regulatory decisions, including
the majority that currently fly under the radar. Our evidence demonstrates that substantive
standards present a context where the power of firms is curbed. As Briloff (1986) argued,
because the FASB tends to issue relatively few substantive standards, it does little to establish
rules for financial reporting independent of the power nexus of firms and professionals.
36
Understanding the relationship between amendments and substantive standards is even more
disheartening from the perspective of empowering those who would oppose regulatory capture
by firms. Amendments, being non-substantive standards, tend to be dominated by firms and
professionals, with little input from regulators. Thus, the amendments that interpret the
ambiguities of substantive standards and clarify implementation issues that arise from them tend
to be resolved by processes dominated by regulatory capture. Not content with dominating the
process for most SFAS, firms and professionals lobbied hard to have the voting rules changed
after the close of our study period in 1987. The effect of this change has been to slow down the
pace of new rules at the FASB. Between its founding and the close of our sample, the fourteen
years from 1973 to 1987, the FASB issued 97 SFAS. By contrast, from 1988, the year after our
sample ends through 2005, the FASB issued only 57 SFAS in an eighteen-year period. Events in
the context of financial reporting regulation, particularly the financial scandals that marked the
turn of the century, can be interpreted to suggest that the FASB has done less recently to curb the
power of the alliance of firms and professionals.
As analysts of legal environments, particularly those interested in workplace rights have
noted, the two decades beginning about 1980 were an era in which state intervention was viewed
in a particularly dim light. The regulation of financial reporting standards was not an exception
to this trend. In 1989, the voting rules for the passage of new regulations were changed; from
that year on, a vote of five to two in favor of a new rule was required for it to pass it rather than
the four to three vote required during the period of our sample. This change was enacted amid
widespread criticism that the FASB had issued far too many new rules; the intent was clearly to
slow down the pace of new regulation. Thus, one clear difference between the fifteen years after
1987 and the fifteen years prior, which were the ones in our sample, was that fewer substantive
37
rules were passed. However, opponents of financial reporting regulation were not merely
content to slow down the FASB, they went further; the example of accounting for stock options
is illustrative. In June 1993, the FASB voted to begin the process of passing a new rule on the
valuation of stock options despite intense corporate opposition. Rather than allow the process to
continue, which according to our results would likely have meant defeat for the alternatives
advocated by firms, the corporate lobby went to Capitol Hill. Hearings were held; Senator
Joseph Lieberman introduced legislation to block the proposed rule. Although this legislation
did not pass, he subsequently introduced and Congress passed a non-binding resolution that
claimed a new rule on valuing stock options would have grave consequences for business in
general and entrepreneurs in particular. Congress also threatened to undermine the authority of
the FASB by requiring that the SEC explicitly ratify each decision that it made. Threatened with
these actions, the chairman of the SEC urged the FASB to back down, which it did. Rather than
issue a new regulation that required firms to expense some duly calculated value of stock options
on their financial statements, the FASB instead called only for disclosure of a value for options,
calculated at the discretion of firms and their accountants, in a footnote to the financial
statements (Levitt, 2002).
The best course of action for reformers and regulators who hope to avoid crises and their
attendant costs by reducing firm influence over financial reporting regulation is to use the
momentum for reform that results from financial crises and scandals to create mechanisms that
empower regulators to engage in ongoing public scrutiny of financial reporting practices. As
Schneiberg and Bartley (2001: 132) suggested, a primary goal of research on public policy
should be a better understanding of how public policy can “… reduce the costs of political
organization or provide certain actors with new political advantages.” Putting this
38
recommendation for future research in the context of our results suggests the need for policies
that enhance the ability of public to participate in the ongoing setting of financial reporting
standards. Absent a fundamental redesign of the process, the major substantive changes that
have been enacted in the wake of financial crises at the turn of the current century are likely to be
undone by amendments and other non-substantive changes produced as part of the ongoing
regulatory process.
The results of this empirical study also suggest several theoretical conclusions and
implications for future research. The first conclusion is that strong support for the utility of the
institutional politics model of ongoing regulation. Future research should be aimed at replicating
these findings and extending them to other contexts to develop a deeper understanding of how
regulatory capture, interest group theory, and institutional mediation interact in dynamic
processes of ongoing regulation. A second research implication follows from the use in this
study of a pooled measure of the opinions of organized actors. It may be that influence processes
respond to the identities of individual letter writers even though our measures do not. For
example, a letter from Anheuser-Busch may have a very different effect than a letter from a local
microbrewery. Further, there may leaders in formal or informal networks operating within the
organized actors that we have studied that may have different influence than other members of
the network. We advocate research with a different design to sort out how reputation and
network effects play a role in influencing regulatory outcomes.
A final research implication is suggested by the importance of the distinction between
substantive and non-substantive regulation. While we treated this distinction as exogenous in
framing our study, we also believe that one of the key advantages of an institutional framework
is that it focuses attention on the endogenous determination of sense making in regulatory
39
processes (Edelman, et al., 1999). From this perspective, understanding efforts the slow down
in financial reporting regulation observed after 1987 requires understanding the dynamics of the
logic of regulation (Bartley and Schneiberg, 2002). Our discussion of events since the close of
our study period, consistent with many other recent studies, revealed a shift in legal
environments away from more active regulation in the recent past. One avenue for
understanding this shift is to focus on how the costs and benefits of regulation are defined. In
particular, the issue of how important actors in the processes of institutional politics came to
accept that regulations were too burdensome is likely an important part of the story. This
suggests the need for longitudinal models and a better understanding of cycles of institutional
change (March and Olsen, 1989).
40
TABLE 1: ORGANIZED ACTORS
Organized Actor
Source of Power
Type of Isomorphism
Aims of Action
Professionals Certification: Control of Principles.
Normative: What is appropriate?
Control specific domain of knowledge
Firms Control of financial statements.
Mimetic: What is prevailing practice?
Regulatory capture.
Regulators Enforcement of legal requirements.
Coercive: What is required by law?
Protect the public interest.
41
TABLE 2: Alternatives in the Exposure Draft of SFAS#40: Financial Reporting and Changing Prices: Specialized
Assets, Timberlands, and Growing Timbers
ALTERNATIVE 1: EITHER HISTORICAL COST (CONSTANT DOLLAR) AMOUNTS OR CURRENT COST MEASURE SHOULD BE PERMITTED. ACCEPTED IN FINAL STATEMENT.
ALTERNATIVE 2: THE STATEMENT SHALL REQUIRE MEASUREMENTS ON A CURRENT COST BASIS. REJECTED IN FINAL STATEMENT.
ALTERNATIVE 3: THE STATEMENT SHALL REQUIRE INFORMATION ABOUT FAIR VALUES, DEFINED AS THE PRICES THAT WOULD BE ACCEPTED AS REASONABLE IN TRANSACTIONS BETWEEN A WILLING SELLER AND A WILLING BUYER. REJECTED IN FINAL STATEMENT.
ALTERNATIVE 4: THE STATEMENT SHALL EXEMPT ACTIVITIES THAT USETIMBERLANDS AND GROWING TIMBER FROM REQUIREMENTS TO PRESENT INFORMATION ON A CURRENT COST BASIS. REJECTED IN FINAL STATEMENT.
Figure 2: Influence of Organized Actors over Substantive Standards
0
0.2
0.4
0.6
0.8
1
-1-0.89-0.78-0.66-0.55-0.44-0.33-0.22-0.11
00.110.220.330.440.550.660.770.880.99
OpinionsProfessionals Sfirms Sregulators
47
REFERENCES
Abzug, R. and Mezias, S. 1993. “The fragmented state due process protections in organizations. Organizational Science, 4: 433-453.
Allison, Paul D.1999. “Comparing logit and probit coefficients across groups.” Sociological Methods and Research, 28: 186-208.
Amenta, Edwin. 1998. Bold Relief: Institutional Politics and the Making of Modern American Policy. Princeton: Princeton University Press
Amenta, Edwin and Drew Halfmann. 2000. “Wage Wars: Institutional Politics, WPA Wages and the Struggle for U.S. Social Policy.” American Sociological Review, 65(4): 506-528.
Amenta, Edwin and Jane D. Poulsen. 1996. “Social Politics in Context: The Institutional Politics Theory and Social Spending at the End of the New Deal.” Social Forces, 75(1): 33-60.
Bachrach, P. and Baratz, M. S. 1962. “Two faces of power.” American Political Science Review, 56: 946-951.Baron, J. N., Dobbin, F. R. and Jennings, P. D. 1986. “War and peace: The evolution of modern personnel administration in U. S.
industry.” American Journal of Sociology, 92: 350-383.
Bartley, T. and M. Schneiberg. 2002. “Rationality and Institutional Contingency: The Varying Politics of Economic Regulation in the Fire Insurance Industry.” Sociological Perspectives, 45(1): 47 92.
Beresford, D. R. and Van Riper, R. 1992. “The not-so-mysterious ways of the FASB.” Journal of Accountancy, 172: 79-83. Boland, R. 1982. “Myth and technology in the American accounting profession.” Journal of
Briloff, A. 1986. "Accountancy and the public interest." Advances in Public Interest Accounting, 1:1-14.
48
Carruthers, Bruce G., and Wendy Nelson Espeland. 1991. “Accounting for rationality: Double-entry bookkeeping and the rhetoric of economic rationality.” American Journal of Sociology, 97 (1): 31-69.
DiMaggio, P. J. and Powell, W. W. 1983. “The iron cage revisited: Institutional isomorphism and collective nationality in organizational fields.” American Sociological Review, 48: 147-160.1991. The New Institutionalism in Organizational Analysis. Chicago: University of Chicago Press.
Dobbin, F. 1992. “The origins of private social insurance: Public policy and fringe benefits in America, 1920-1950.” American Journal of Sociology, 97: 1416-1450.
Dobbin, F. R., Edelman, L., Meyer, J. W., Scott, W. R. and Swidler, A. 1987. “The expansion of due process in organizations, in Zucker, L. G. (Ed.).
Institutional Patterns and Culture: 71-98. Cambridge, MA: Ballinger Publishing Company.
Dobbin, F. and J. Sutton. 1998. "The Strength of a Weak State: The Employment Rights Revolution and the Rise of Human Resources Management Divisions." American Journal of Sociology. 104:441-476.
Edelman, L. B. 1990. “ Legal environments and organizational governance: The expansion of due process in the American workplace.” American Journal of Sociology, 95: 1401 -1440.1992. “Legal ambiguity and symbolic structures: Organizational mediation of civil rights law.” American Journal of Sociology, 97: 1531-1576.
Edelman, Lauren B., Christopher Uggen, and Howard S. Erlanger.1999. “The Endogeneity of Legal Regulation: Grievance Procedures as Rational Myth.” American Journal of Sociology, 105: 406-454.
Edelman, Lauren B., Sally Riggs Fuller, and Iona Mara-Drita. 2001. “Diversity Rhetoric and the Managerialization of Law.” American Journal of Sociology, 106:1589-1641.
Fuller, Sally Riggs, Lauren B. Edelman, and Sharon Matusik. 2000. “Legal Readings: Employee Interpretation and Enactment of Civil Rights Law.” Academy of Management Review, 25:200-216.
Hofstadter, Richard. 1955. The Age of Reform: From Bryan to F.D.R. New York: Vintage Books.
49
Ingram, Paul and Hayagreeva Rao. 2004. "Store wars: The enactment and repeal of anti-chain legislation in the United States." American Journal of Sociology, 110: 446-487.
Jenkins, J. Craig and Barbara Brents.1989 "Social Protest, Hegemonic Competition and Social Reform: The Political Origins of the American Welfare State." American Sociological Review, 54: 891-909.
Kelly, E. and F. Dobbin. 1999. "Civil Rights Law at Work: Sex Discrimination and the Rise of Maternity Leave Policies." American Journal of Sociology, 105:455-492.
Laffont, Jean-Jacques & Tirole, Jean1991. "Privatization and Incentives." Journal of Law, Economics and Organization, 7: 84-105.
Levitt, G. 2002. Assessing the Ecological Sustainability of the Tasmanian Giant Crab Fishery, Prepared by DPIWE for Environment Australia, DPIWE, Hobart.
Mahoney, Paul G. 2001. “The Political Economy of the Securities Act of 1933.” Journal of Legal Studies, 30:1-31.2003. “The Origins of the Blue-Sky Laws: A Test of Competing Hypotheses.” Journal of Law and Economics, 46:229-251.
March, J. G. and Olsen, J. P. 1989. Rediscovering Institutions. New York: The Free Press.
Merrill, Francis E., and Melchior Palyi. 1938. “The Stock Exchange and Social Control.” The American Journal of Sociology, 43: 560-577.
Meyer, J. W. and Rowan, B. 1977. Institutionalized organizations: Formal structure as myth and ceremony. Reprinted in Meyer, J. W. and Scott, R. (Eds.). Organizational Environments: 45 67. Beverly Hills, CA: SAGE Publications.
Mezias, Stephen J. 1990. “An institutional model of organizational practice: Financial reporting at the Fortune 200.” Administrative Science Quarterly, 35: 431-457.1994. “Financial Meltdown as Normal Accident: The Case of the American Savings and Loan Industry.” Accounting, Organizations, & Society, 18: 181-192.
Mezias, S. J. and Chung, S.
50
1989. Due Process and Participation at the FASB. Morristown, NJ: The Financial Eecutives Institute.
1991. Longitudinal change in institutional environments: The case of generally accepted ccounting principles. Proceedings of the Third Interdisciplinary Perspectives on ccounting Conference, 2.9.1-2.9.20.
Mezias, S. J. and Scarselletta, M. 1992. Due Process and Participation: Agenda Setting at the FASB. Morristown, NJ: The Financial Executives Institute.1994. “Resolving financial reporting problems: An institutional analysis of the process.” Administrative Science Quarterly, 39: 654-678.
Miller, Paul W., Rodney J. Redding, and Paul R. Bahnson. 1994. The FASB: The People, the Process, and the Politics. Burr Ridge, IL: Richard D. Irwin.
Montagna, Paul. 1986. “Accounting Rationality and Financial Legitimation.”Theory and Society, 15: 103 138.
Oliver, C. 1991. “Strategic response to institutional processes.” Academy of Management Review, 16: 145-179.
Perrow, Charles. 1984. “Comments on Langton’s Ecological Theory of Organization.” Administrativ Science Quarterly, 30:278-283.
Schneiberg, Marc, and Tim Bartley. 2001. “Regulating American industries: Markets, politics, and the institutional determinants of fire insurance regulation.” American Journal of Sociology, 107 (1), 101-146.
Skousen, K. F. 1991. An Introduction to the SEC. Cincinnati: Southwestern Publishing.
Tolbert, P. S. and Zucker, L. G. 1983. Institutional sources of change in the formal structure of organizations: The
diffusion of civil service reform, 1880-1935. Administrative Science Quarterly, 28: 22-39.
Sutton, John R. and Frank Dobbin. 1996. “The Two Faces of Governance: Responses to Legal Uncertainty in American firms, 1955-1985.” American Sociological Review 61: 794-811.
Stryker, Robin.
51
2002 “ A Political Approach to Organizations and Institutions.” In Lounsbury, Michael, and Marc J. Ventresca (Eds.), Social Structure and Organizations Revisited. San Diego: Elsevier Science Ltd.
Tandy, P. R. and N. L. Wilburn.1992. “Constituent Participation in Standard-Setting: The FASB’s First 100
Statements.”U, 6(2):47-58.
Wootton, Charles W., Carel M. Wolk, and Carol Normand. 2003. “An historical perspective on mergers and acquisitions by major US accounting firms.” Accounting History, 8: 25-60.