Professional Documents
Culture Documents
ABSTRACT
This study investigates professional tax preparers' willingness
to include questionable client-provided data in a tax return without
verifying the information. Although a recent survey of tax
practitioners by Yetmar et al. (1998) found that the issue of reliance
on client data is a significant ethical problem in practice, a very
limited amount of previous research has investigated the factors that
influence tax practitioners' reliance decisions. In the current paper,
the issue of client reliance is viewed as a problem of trust and
suspicion, and the general model of trust and suspicion proposed by
Kee and Knox (1970) is adopted as a conceptual framework for
addressing this issue. Based on this model and previous research
findings, it was hypothesized that client reliability, the client's yearend tax payment status, the general propensity to trust others, and
tax practitioners' attitudes toward risk would influence reliance
decisions. The findings, based on a study of CPA tax practitioners,
indicate that client reliability and tax payment status each had a
significant impact on client reliance. However, contrary to
expectations, the general tendency to trust others did not influence
client reliance decisions in a tax context. The results provide mixed
support for the hypothesized effects of tax preparers' risk attitudes
on decisions. The findings of the study also raise questions regarding
the appropriateness of tax practitioners' client reliance decisions.
INTRODUCTION
Although many studies in the tax literature have investigated
professional tax preparers' willingness to advocate aggressive return
positions, very little empirical research has examined the issue of
reliance on questionable data supplied by clients. However, recent
evidence reported by Yetmar et a1. (1998) suggests that the latter
issue is one of the most significant ethical issues in tax practice.
They surveyed a random sample of AICPA Tax Division members to
determine the extent to which 54 issues were perceived as
significant ethical problems. The results indicated that two of the top
three ethical issues related to reliance on client-provided
information. The highest rated issue was "accepting a client's
ABSTRAK
Studi ini meneliti kesediaan preparers pajak profesional untuk
memasukkan data disediakan klien yang dipertanyakan dalam
pengembalian pajak tanpa memverifikasi informasi. Meskipun survei
terbaru dari praktisi pajak dengan Yetmar et al. (1998) menemukan
bahwa masalah ketergantungan pada data klien adalah masalah
etika yang signifikan dalam praktek, jumlah yang sangat terbatas
penelitian
sebelumnya
telah
meneliti
faktor-faktor
yang
mempengaruhi keputusan ketergantungan praktisi pajak. Dalam
kertas saat ini, isu ketergantungan pada klien dipandang sebagai
masalah kepercayaan dan kecurigaan, dan model umum
kepercayaan dan kecurigaan yang diusulkan oleh Kee dan Knox
(1970) diadopsi sebagai kerangka kerja konseptual untuk mengatasi
masalah ini. Berdasarkan model ini dan temuan penelitian
sebelumnya, itu hipotesis bahwa kehandalan klien, status
pembayaran akhir tahun pajak klien, kecenderungan umum untuk
mempercayai orang lain, dan sikap praktisi pajak terhadap risiko
akan
mempengaruhi
keputusan
ketergantungan.
Temuan,
berdasarkan studi dari praktisi pajak CPA, menunjukkan bahwa
kehandalan klien dan status pembayaran pajak masing-masing
memiliki dampak yang signifikan terhadap ketergantungan pada
klien. Namun, bertentangan dengan harapan, kecenderungan umum
untuk mempercayai orang lain tidak mempengaruhi keputusan
ketergantungan klien dalam konteks pajak. Hasil penelitian ini
memberikan dukungan campuran untuk efek hipotesis sikap risiko
preparers pajak pada keputusan. Temuan dari studi ini juga
menimbulkan
pertanyaan
mengenai
kesesuaian
keputusan
ketergantungan klien praktisi pajak '.
PENDAHULUAN
Meskipun banyak penelitian dalam literatur pajak telah
menyelidiki kesediaan preparers pajak profesional untuk mendukung
posisi agresif SPT, penelitian empiris sangat sedikit yang meneliti
masalah ketergantungan pada data yang dipertanyakan diberikan
oleh klien. Namun, bukti terbaru yang dilaporkan oleh Yetmar et al.
(1998) menunjukkan bahwa masalah terakhir adalah salah satu dari
isu-isu etis yang paling signifikan dalam praktek pajak. Mereka yang
disurvei sampel acak dari anggota Divisi Pajak AICPA untuk
menentukan sejauh mana 54 isu yang dianggap sebagai masalah
acak dari anggota AICPA dalam praktek pajak, yang menanggapi dua
kasus yang melibatkan isu ketergantungan klien. Hasil penelitian
menunjukkan bahwa, seperti dihipotesiskan, persepsi kelayakan
kepercayaan wajib pajak berdasarkan pengalaman dengan klien
tertentu
mempengaruhi
keputusan
ketergantungan
secara
signifikan. Juga konsisten dengan hipotesis penelitian, CPA lebih
mungkin untuk memerlukan dokumentasi pendukung untuk
pengurangan diklaim oleh wajib pajak yang berada di posisi pajak
kurang bayar pada akhir tahun, yang menunjukkan bahwa praktisi
pajak mengenali efek insentif karena status pembayaran akhir tahun
klien . Bertentangan dengan harapan, ukuran kecenderungan untuk
mempercayai orang lain tidak memiliki dampak yang signifikan
terhadap keputusan ketergantungan dan dukungan campuran
diperoleh untuk efek hipotesis sikap risiko pada kepercayaan klien.
Hasil penelitian ini menunjukkan bahwa Model Kee dan Knox
(1970) dapat memberikan kerangka teoritis berguna untuk
penyelidikan kepercayaan dan kecurigaan diantara preparers pajak.
Temuan ini juga menimbulkan pertanyaan mengenai kelayakan
keputusan ketergantungan klien CPA. Bahkan dalam kasus-kasus di
mana klien telah overstate pemotongan atau undestate pendapatan
di masa lalu dan memiliki insentif saat ini untuk menipu, perkiraan
rata-rata peserta dari kemungkinan mengandalkan data yang
dipertanyakan berkisar dari sekitar 30 sampai 40%. Disarankan
bahwa penelitian masa depan harus mencoba untuk mendapatkan
pemahaman yang lebih baik dari faktor-faktor yang mempengaruhi
kepercayaan klien dalam pengaturan persiapan pajak, dan berusaha
untuk menilai kelayakan keputusan ketergantungan klien praktisi
pajak CPA '.
Bagian selanjutnya menyediakan tinjauan literatur yang
relevan dan mengembangkan hipotesis penelitian. Ini diikuti dengan
diskusi tentang metode penelitian dan temuan empiris. Bagian akhir
dari kertas berisi diskusi tentang temuan dan keterbatasan
penelitian, serta saran untuk penelitian di masa depan.
TINJAUAN PUSTAKA DAN PENGEMBANGAN HIPOTESIS
General Model cf Trust dan Kecurigaan
Masalah ketergantungan pada data klien dipertanyakan dapat
dilihat sebagai masalah kepercayaan dan kecurigaan. Rotter (1971:
444) mendefinisikan kepercayaan sebagai:
... Sebuah harapan yang dimiliki oleh individu atau kelompok
yang kata, janji, pernyataan lisan atau tertulis dari individu atau
joint choices. One of the parties (P) is confronted with the choice of
either trusting or not trusting the other (0), who in turn has a choice
of being either trustworthy or untrustworthy. Kee and Knox's basic
conceptualization of trust and suspicion is presented in Fig. 1.
The model makes an explicit distinction between subjective
and behavioral trust or suspicion. Subjective trust refers to P's
assessment of the probability that 0 will be trustworthy, while
behavioral trust refers to the manifest act of trust. According to Kee
and Knox, a person may manifest complete trust in another until the
assessed probability of trustworthiness fans below a critical
threshold, at which point the probability of trusting behavior will
decline. The level of this threshold win depend on situational
variables, such as the risks and rewards associated with trusting
behavior. Subjective trust or suspicion will be determined by P's
perception of O's motives and/or competence. The model recognizes
three types of independent variables that may influence these
perceptions: (1) structural and situational factors, (2) dispositional
factors, and (3) previous experience.
Structural and situational variables include factors such as
incentives, power, and characteristics of the trusted party. Rational
choice models of trust recognize the influence of situational variables
on trusting behavior. For example, Hardin (1992) suggested that a
critical element of a trusting relationship is the incentive of the
trustee to honor the trust. According to Hardin, a person will trust
another only if he/she believes it will be in that person's best interest
to be trustworthy. The influence of incentives on trust also is
supported by research indicating that suspicion is often triggered by
situational cues that raise questions about the motives of the trustee
(Fein, 1996; Hilton et al., 1993). A rational choice model of trust
would also suggest that the relative power of the parties will
influence trusting behavior. For example, P will have a greater
incentive to trust o if 0 is in a position of power relative to P. A
number of studies also have found that general characteristics of a
trustee, such as their membership in various social categories,
influence trust (Kramer, 1999).
Roberts
(1998).
praktisi
pajak
harus
mempertimbangkan
ketergantungan klien ketika mengevaluasi risiko yang terlibat dalam
pelaporan pajak. Berdasarkan data wawancara, ketergantungan klien
diidentifikasi oleh Milliron (1988) sebagai salah satu dari beberapa
faktor yang mempengaruhi keputusan pelaporan pajak preparers,
tapi Helleloid (1989) adalah satu-satunya studi yang telah berusaha
untuk menilai efek dari ketergantungan klien pada penilaian
pelaporan pajak .
Helleloid (1989) meneliti efek dari ambiguitas catatan jarak
tempuh mobil klien, dioperasionalkan sebagai frekuensi yang log
jarak tempuh mobil telah diupdate, pada rekomendasi pemotongan
mileage profesional pajak. Hasil percobaan awal menunjukkan
bahwa frekuensi yang diperbarui log tidak berpengaruh signifikan
terhadap direkomendasikan pemotongan mileage. Dalam percobaan
lanjutan, frekuensi pembaruan memiliki efek yang signifikan secara
statistik, tetapi lebih kecil pada potongan direkomendasikan.
Helleloid (1989) menyimpulkan bahwa, secara keseluruhan,
subjeknya menunjukan sensitivitas yang sangat terbatas untuk
kualitas atau keandalan catatan klien.
Meskipun (1989) temuan Helleloid ini memberikan dukungan
terbatas untuk efek ketergantungan klien pada keputusan pelaporan,
studi tidak mengatasi masalah kesediaan CPA untuk mempercayai
klien ketika mereka mungkin memiliki alasan untuk menduga
ketidakjujuran klien, atau efek dari insentif yang dirasakan klien
untuk menipu pada pengembalian pajak. Berdasarkan Kee dan Knox
(1970) model, penelitian ini mengusulkan bahwa pengalaman
sebelumnya dengan klien akan mempengaruhi keputusan
ketergantungan klien selanjutnya melalui efeknya pada persepsi
kepercayaan klien. Efek dari pengalaman sebelumnya pada
kesediaan CPA mengandalkan data klien yang disediakan adalah
masalah penting yang belum ditangani oleh penelitian sebelumnya.
Meskipun isu kredibilitas klien atau kepercayaan telah
menerima sangat sedikit perhatian dalam literatur pajak, beberapa
studi telah meneliti efek dari agresivitas klien, atau preferensi risiko
klien, saran CPA (misalnya Cloyd, 1995; Cuccia et al., 1995; Schisler,
1994;. Duncan et al, 1989). Hasil studi ini umumnya telah
menunjukkan bahwa CPA mencoba untuk mengakomodasi preferensi
risiko klien mereka; yaitu mereka lebih (kurang) cenderung membuat
keputusan pelaporan agresif untuk agresif (konservatif) klien
(Roberts, 1998).
Penalaran analogi mungkin menunjukkan bahwa CPA akan
lebih cenderung mengandalkan data dipertanyakan diberikan oleh
klien lebih agresif atau tidak dapat dipercaya; Namun, ada
a client does not violate the CPA's ethical standards provided the
realistic possibility standard has been met. In contrast, if the CPA
believes that a client has supplied fraudulent data, completing the
return without making reasonable inquiries regarding the data would
be a clear violation of professional standards. Accordingly, CPAs
should be less willing to accommodate their clients' preferences if
they feel they have supplied unreliable or fraudulent data?
This reasoning is consistent with recent studies in the auditing
literature, which have shown that client preferences influence
auditor judgment for accounting issues that are in the "gray area,"
but not for issues that represent clear violations of GAAP (e.g.
Salterio & Koonce, 1997). Thus, if past experience with a client raises
credibility questions, then tax practitioners should be less willing to
rely on the data without verification, as reflected in the following
hypothesis.
Hypothesis 1. Tax practitioners will be more (less) willing to
rely on questionable data provided by a client whose information has
been reliable (unreliable) in the past.
An important situational variable recognized in the Kee and
Knox (1970) model is the effect of incentives for dishonest behavior.
Several studies in the tax literature have suggested that a client's
year-end tax payment status affects incentives to adopt aggressive
positions (e.g. Duncan et al., 1989; LaRue & Reekers, 1989; Schisler,
1994, 1995). Specifically, an underwithheld client should have a
greater incentive to engage in aggressive or fraudulent reporting.
This incentive effect may be explained by the potentially disruptive
consequences of having to make an unexpected payment, or more
formally based on Prospect Theory, which suggests that the
potential-loss framing effect for the unexpected payment would
result in risk-seeking behavior (Roberts, 1998). It has been
suggested that tax practitioners also will make more aggressive
decisions for clients who are in an underwithheld position, in an
effort to satisfy the client (LaRue & Reekers, 1989).
Based on a review of the empirical findings on this issue,
Roberts (1998) concluded that year-end payment status alone does
not appear to affect tax preparers' decisions, but payment status
does interact with other variables such as client risk preferences to
affect the likelihood of aggressive reporting. For example, Schisler
(1994, 1995) found that payment status increased the likelihood of
aggressive recommendations when the client was described as
managers who worked for the same public accounting firm; thus,
firm training and socialization may have produced a relatively
homogeneous set of subjects who were more influenced by firm
policy than by personal dispositions. Thus, in a less homogeneous
subject group, dispositional trust should be more likely to influence
judgments, as reflected in the following hypothesis.
RESEARCH METHOD
The data for this study were obtained by mailing instruments
to a random sample of AICPA members whose membership
Instrument
The research instrument included: (1) a cover letter, (2) two
tax cases, (3) the MacDonald et al. (1972) Self-Report Trust Scale,
and (4) a supplemental data sheet.' To enhance external validity, two
separate tax cases were used. Client reliability and tax payment
status were manipulated on a between-subjects basis for both cases,
which are illustrated in Appendix I.
In the Low Reliability conditions, the case indicated that in the
past the client had been unable to provide documentation relating to
certain deductions (Case I), or that the CPA had discovered several
errors in the client's records in the past (Case 2). In the High
Reliability conditions, the case indicated that all information supplied
by the client in the past had proven to be reliable." Tax payment
status was manipulated by indicating that the client either had a
significant amount of tax due, or should receive a significant tax
refund. The cases were reviewed by three CPA firm partners, who
indicated that they seemed realistic, and that they had encountered
similar situations in the past. Taxpayer aggressiveness and payment
status were crossed to create four versions of each case, and each
potential subject was randomly assigned to one of the four versions
of each case.
METODE PENELITIAN
Data untuk penelitian ini diperoleh dengan instrumen surat
untuk sampel acak dari anggota AICPA yang informasi keanggotaan
menunjukkan bahwa mereka dalam praktek akuntan publik, dan
bahwa mereka memiliki minat di bidang perpajakan. Bagian ini
menjelaskan instrumen penelitian dan peserta.
Instrumen
Instrumen penelitian meliputi: (1) cover surat, (2) dua kasus
pajak, (3) MacDonald et al. (1972) Self-Report Skala Kepercayaan,
dan (4) lembar data tambahan. " Untuk meningkatkan validitas
eksternal, dua kasus pajak yang terpisah digunakan. Kehandalan
klien dan status pembayaran pajak yang dimanipulasi di antara
subyek dasar untuk kedua kasus, yang diilustrasikan dalam
Lampiran I.
Dalam kondisi Keandalan rendah, kasus menunjukkan bahwa
di masa lalu klien tidak mampu menyediakan dokumentasi yang
berkaitan dengan pemotongan tertentu (Kasus I), atau bahwa BPA
telah menemukan beberapa kesalahan dalam catatan klien di masa
lalu (Kasus 2). Dalam kondisi Tinggi Keandalan, hal ini menunjukkan
bahwa semua informasi yang diberikan oleh klien di masa lalu telah
terbukti dapat diandalkan. "Status pembayaran Pajak dimanipulasi
dengan menunjukkan bahwa klien baik memiliki sejumlah besar
pajak yang harus dibayar, atau harus menerima signifikan restitusi
pajak. Kasus-kasus yang ditinjau oleh tiga mitra perusahaan CPA,
yang menunjukkan bahwa mereka tampak realistis, dan bahwa
mereka telah mengalami situasi yang sama di Agresivitas Wajib
Pajak dan status pembayaran masa lalu. disilangkan untuk membuat
empat versi dari masing-masing kasus, dan masing-masing subjek
potensial secara acak ditugaskan untuk salah satu dari empat versi
masing-masing kasus.
Untuk menilai kesediaan mereka untuk mengandalkan data-
Participants
Instruments were mailed to a random sample of 1,000 AICPA
members. After a follow up mailing, a total of 256 usable responses
were obtained, providing a response rate of approximately 25%. A
comparison of the responses to the tax case and demographic data
for the early and late respondents indicated no significant
differences. A demographic profile of participants is provided in Table
1. As indicated in the table, 81% of the respondents were male, and
approximately 98% were either partners or managers in public
accounting firms. Since most respondents were partners or
managers, it is not surprising that the sample was predominantly
male. Data reported by Doucet and Hooks (1999) indicated that in
1997, males accounted for 84% of all partners in public accounting
firms. Thus, the gender composition of the sample appears to be
approximately representative of this population. The fact that
virtually all respondents were either partners or managers suggests
that higher level employees may be more likely to respond to
surveys; however, this was not considered a cause for concern
because these are the employees who are most likely to make the
ultimate decision regarding reliance on client-provided information.
RESULTS
Responses to Tax Cases and Preliminary Analysis
A summary of responses to the tax cases is provided in Table
2. As the data indicate, participants' mean estimates of the likelihood
they would include the questionable deductions in the client's tax
return without examining supporting documentation or otherwise
verifying them ranged from approximately 42 (32) to 73 (55) percent
for Case J (2). For both cases, the lowest estimates were obtained for
the Low Reliability, Tax Due condition, and the highest estimates
were for the High Reliability, Refund condition, which is consistent
with the research hypotheses. The responses also indicate significant
variation in participants' willingness to rely on questionable client
data, based on the relatively large standard deviations of the
estimates.
Responses to the manipulation checks are summarized in Table
3. The estimated reliability of the client-provided data was used to
test the effectiveness of the reliability manipulation. One-way ANOVA
models with estimated reliability as the dependent variable and
manipulated reliability as the independent variable indicated that
the effects of the manipulation were significant at the 0.0001 level
for both cases, although the effect appears to be more pronounced
for Case 1. Participants' estimates of the strength of the hypothetical
client's incentives to overstate their reported expenses served as a
check for the payment status manipulation. One-way ANOVA models
with estimated incentives as the dependent variable and payment
status as the independent variable indicated that the effects of this
manipulation also were significant at the 0.0001 level for both cases.
To assess the realism of the cases, respondents were asked to
estimate the probability that they could encounter similar situations
in their tax practice. The mean estimates reported in Table 3 ranged
from 64 (61) to 75 (72) percent for Case I (2), which indicates that
the hypothetical scenarios were perceived as relatively realistic.
Consistent with the findings of Yetmar et al. (1998), these results
also suggest that the problem of reliance on questionable client data
is a significant ethical issue in tax practice.
Univariate ANOVA and regression models were used to test for
possible effects of various demographic factors on participants'
willingness to rely on client data. Univariate ANOVA models indicated
that neither gender, position (partner vs. manager), education level
(bachelors vs. masters), nor firm type (national/international vs.
regional/local) had a significant influence on reliance decisions.
Linear regression models also revealed that age, years of
experience, and the percentage of time spent doing tax work
generally did not affect reliance." Based on the lack of significant
influence of demographic factors on client reliance decisions, these
variables were excluded from subsequent analyses.
Hypothesis Tests
The hypotheses were tested using analysis of covariance
(ANCOVA) models. The results of the models for both cases are
included in Table 4. Both models include client reliability and yearend tax payment status as between subjects variables, and selfreported trust and risk attitude as covariates. The model for Case 1
indicates that the effects of both client reliability and payment status
HASIL
Tanggapan untuk Kasus Pajak dan Analisis Awal
Ringkasan tanggapan terhadap kasus pajak diberikan dalam
Tabel 2. Sebagai data menunjukkan, perkiraan rata-rata peserta dari
kemungkinan mereka akan mencakup pemotongan dipertanyakan
dalam SPT klien tanpa memeriksa dokumentasi pendukung atau
memverifikasi mereka berkisar dari sekitar 42 ( 32) ke 73 (55)
persen untuk Kasus 1 (2). Untuk kedua kasus, perkiraan terendah
diperoleh untuk Keandalan Rendah, kondisi Karena Pajak, dan
perkiraan tertinggi adalah untuk Keandalan tinggi, kondisi Refund,
yang konsisten dengan hipotesis penelitian. Tanggapan juga
menunjukkan variasi yang signifikan dalam kesediaan peserta
mengandalkan data klien dipertanyakan, berdasarkan standar
deviasi relatif besar dari perkiraan.
Tanggapan untuk pemeriksaan manipulasi dirangkum dalam
Tabel 3. keandalan Estimasi data-klien yang disediakan digunakan
untuk menguji efektivitas manipulasi kehandalan. Satu arah model
ANOVA dengan perkiraan keandalan sebagai variabel dependen dan
kehandalan dimanipulasi sebagai variabel independen menunjukkan
bahwa efek dari manipulasi yang signifikan pada tingkat 0,0001
untuk kedua kasus, meskipun efeknya tampaknya lebih jelas untuk
Kasus 1. Peserta ' perkiraan kekuatan insentif klien hipotetis untuk
melebih-lebihkan biaya mereka melaporkan menjabat sebagai cek
manipulasi status pembayaran. Satu arah model ANOVA dengan
insentif estimasi sebagai variabel dependen dan status pembayaran
sebagai variabel independen menunjukkan bahwa efek dari
manipulasi ini juga signifikan pada tingkat 0,0001 untuk kedua
kasus. Untuk menilai realisme kasus, responden diminta untuk
memperkirakan probabilitas bahwa mereka bisa menghadapi situasi
yang sama dalam praktek pajak mereka. Perkiraan rata-rata
dilaporkan dalam Tabel 3 berkisar dari 64 (61) 75 (72) persen untuk
Kasus I (2), yang menunjukkan bahwa skenario hipotetis yang
dianggap relatif realistis. Konsisten dengan temuan Yetmar et al.
(1998), hasil ini juga menunjukkan bahwa masalah ketergantungan
pada data klien dipertanyakan adalah masalah etika yang signifikan
dalam praktek pajak.
Univariat ANOVA dan regresi model yang digunakan untuk
menguji kemungkinan efek dari berbagai faktor demografi pada
kesediaan peserta mengandalkan data klien. Model univariat ANOVA
menunjukkan bahwa baik jenis kelamin, posisi (mitra vs manager),
tingkat pendidikan (bujangan vs master), atau jenis perusahaan
The results for Case 2 also indicate that the effects of reliability
and payment status were significant, although to a lesser degree
than for Case 1. These findings also support Hypotheses 1 and 2. As
in Case 1, the interaction of reliability and payment status did not
approach significance. For Case 2 neither of the covariates were
significant. Thus, the results for Case 2 failed to support Hypotheses
3 and 4.
data, and the factors that influence such reliance decisions. The
findings of the study are subject to a number of limitations and
should be interpreted with caution. For example, most respondents
were employed by local CPA firms. If systematic differences exist
between the judgments of tax practitioners employed in large and
small firms due to differences in such things as selection, training or
socialization, then the findings of the current study may not be
generalizable to employees of national or inter national firms. Also,
although the response rate was comparable to that typically
obtained in this type of study (e.g. Bandy et al., 1994; Schisler,
1994), there is a chance that the results were affected by
nonresponse bias. The generalizability of the findings to practice
settings also is an open question. An attempt was made to capture
some of the key contextual influences on CPAs' reliance decisions in
the hypothetical cases; however, client reliance decisions in practice
may be affected by factors not addressed in the current study, such
as economic incentives.