Translated Copy of Kuncoro2004
Translated Copy of Kuncoro2004
Suap di Indonesia:
beberapa bukti dari data tingkat mikro
Ari Kuncoro
a
Universitas Indonesia
Diterbitkan online: 19 Okt 2010.
To cite this article: Ari Kuncoro (2004) Bribery in Indonesia: some evidence from micro-level data, Bulletin
of Indonesian Economic Studies, 40:3, 329-354, DOI: 10.1080/0007491042000231511
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our
licensors make no representations or warranties whatsoever as to the accuracy, completeness, or
suitability for any purpose of the Content. Any opinions and views expressed in this publication are
the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis.
The accuracy of the Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any losses, actions,
claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the
use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use
can be found at http://www.tandfonline.com/page/terms-and-conditions
CBIEDec04 27/10/04 4:23 PM Page
Carfax Publishing
Taylor & Francis Group Buletin Kajian Ekonomi Indonesia, Vol. 40, No. 3, 2004: 329–54
PENYUAPAN
DI INDONESIA:
BEBERAPA BUKTI DARI DATA TINGKAT MIKRO
Ari Kuncoro*
Universitas Indonesia
[
PENDAHULUAN
S
i Tujuan utama dari makalah ini adalah untuk menyelidiki sejauh mana dan faktor-
m faktor penentu penyuapan di Indonesia dalam sistem pemerintahan yang lebih
o desentralisasi yang diperkenalkan pada tahun 2001. Undang-undang desentralisasi
n (UU 22 Tahun 1999 tentang Pemerintah Daerah dan UU 25 Tahun 1999 tentang
Alokasi Keuangan antara Pemerintah Pusat dan Daerah) dimaksudkan untuk
F membalikkan sentralisasi tingkat tinggi sebelumnya, dengan memberikan
r pengalihan beberapa wewenang dan tanggung jawab fiskal dari pusat ke
a pemerintah tingkat yang lebih rendah (sebagian ke provinsi, tetapi terutama ke
kabupaten dan kota). Salah satu dampak yang tidak terduga dari pengaturan baru
adalah perlombaan antar pemerintah daerah, terutama di tingkat kabupaten/kota,
untuk menciptakan pajak baru. Apa yang disebut 'pajak gangguan' (seperti pajak
kambing, lift, dan pompa air) telah berlipat ganda. Selain itu, telah dikemukakan,
menjamurnya pajak dan peraturan daerah baru telah disertai dengan peningkatan
korupsi.
Studi ini berfokus pada penyuapan dalam kaitannya dengan penilaian pajak dan
intervensi peraturan pemerintah daerah seperti persyaratan untuk izin usaha,
sertifikat kepatuhan terhadap peraturan lingkungan, izin bangunan dan
keselamatan kebakaran dan inspeksi kontrak kerja. Ini tidak mencakup kegiatan
pencarian rente lainnya seperti melobi pemerintah untuk proyek-proyek tertentu,
perlindungan industri dan hak monopoli eksklusif. Studi ini menggunakan
kumpulan data yang kaya yang dihasilkan oleh Survei Khusus tentang Tata Kelola
(SSG), survei lapangan yang dilakukan di tingkat pemerintah kabupaten oleh
Institut Ekonomi dan Sosial
CBIEDec04 27/10/04 4:23 PM Page
ISSN 0007-4918 print/ISSN 1472-7234 online/04/ 030329-26 © 2004 Indonesia Project ANU
DOI: 10.1080/0007491042000231511
CBIEDec04 27/10/04 4:23 PM Page
sebagai kriteria utama untuk memilih kabupaten yang akan disurvei. Karena alasan
ini, kabupaten-kabupaten terpencil dengan kegiatan ekonomi yang sangat terbatas
tidak mungkin dimasukkan dalam sampel kami; daerah-daerah seperti
Jabodetabek, Surabaya Raya, Bandung dan Medan memiliki jumlah perusahaan
perwakilan yang lebih banyak, karena terdapat konsentrasi besar perusahaan
manufaktur di wilayah ini.
Selain perusahaan manufaktur, kami juga mewawancarai perusahaan di sektor
jasa (transportasi, ritel dan hotel dan restoran, tetapi bukan bank) dan agribisnis
(perkebunan, peternakan dan peternakan ikan), asalkan mereka terdaftar di kantor
Kadinda sebagai 'perusahaan' ( perusahaan). Perusahaan sampel dipilih secara acak
dari daftar yang disediakan oleh Kadinda. Rencana awalnya adalah mewawancarai
sekitar 10% perusahaan yang terdaftar, dengan rincian 45:45:10 antara
perusahaan manufaktur, jasa dan agribisnis. Namun, karena pengurangan
D perusahaan karena berbagai alasan (seperti tanggapan yang buruk, penolakan
o untuk diwawancarai dan penutupan perusahaan), komposisi sebenarnya tidak
w selalu sesuai dengan rencana awal ini. Satu masalah adalah bahwa untuk kabupaten
n dengan beberapa perusahaan terdaftar, target 10% diterjemahkan ke dalam
l ukuran sampel yang sangat kecil. Oleh karena itu kami sedikit melakukan
o oversampling distrik kecil dalam upaya untuk memasukkan setidaknya 25
a perusahaan, tetapi untuk berbagai alasan ukuran sampel kadang-kadang masih
d kurang dari 10.
e Untuk melengkapi informasi yang diperoleh dari wawancara tingkat perusahaan,
d kami juga melakukan wawancara dengan pejabat daerah. Tidak ada pertanyaan
'korupsi' dalam kuesioner; melainkan, tujuannya adalah untuk mengevaluasi sikap
b pejabat daerah terhadap sektor swasta dan untuk membedakan retorika dari
y praktik yang sebenarnya. Kami sangat tertarik untuk mengetahui lebih banyak
tentang sistem perizinan dan peran pajak pengganggu dalam pendapatan asli
[ daerah (PAD) masing-masing. Respondennya adalah pejabat tinggi Bappenas, Dinas
S Pendapatan, Dinas Pekerjaan Umum, Sekretariat Daerah, dan Badan Koordinasi
i Penanaman Modal Kabupaten.
m
o
n HASIL SURVEI
Tanggapan Perusahaan
F
Dari 1.808 perusahaan yang diwawancarai, 1.333 responden mengatakan mereka
r
telah membayar suap pada tahun 2001, sedangkan sisanya mengaku tidak
a
melakukannya. Suap rata untuk perusahaan-perusahaan yang melaporkan
membayar suap adalah 10,8% dari biaya produksi tahunan. 2 Kami melakukan
wawancara pasca survei terpisah dengan asosiasi hotel dan restoran, asosiasi
usaha kecil dan cabang kamar dagang dan industri lokal, dan juga dengan sejumlah
kecil perusahaan yang dipilih secara acak dari luar sampel utama. di enam
kabupaten di seluruh Indonesia, untuk mengecek keakuratan hasil survei. 3 Kami
menemukan bahwa rata-rata suap yang dihitung untuk seluruh sampel berada
dalam kisaran perkiraan asosiasi industri dan perusahaan di luar sampel. Tingkat
suap tipikal yang dilaporkan oleh responden tambahan ini berkisar antara 10-15%
dari biaya produksi tahunan, meskipun beberapa responden menyarankan bahwa
mereka bisa jauh lebih tinggi.
Kami melihat beberapa karakteristik perusahaan untuk mendapatkan wawasan
awal ke dalam proses penyuapan yang mungkin memandu perumusan hipotesis
lanjutan di bagian pemodelan (tabel 1). Diamati bahwa tingkat suap tampaknya
meningkat dengan ukuran perusahaan, tetapi kemudian menurun. Tabel 1 juga
mengungkapkan bahwa perusahaan manufaktur cenderung membayar suap lebih
rendah daripada perusahaan jasa, sementara agri-
CBIEDec04 27/10/04 4:23 PM Page
Ukuran (pendapatan
penjualan tahunan) Kecil (< 1 488
Rp 1 miliar) 0,
4
Kecil-menengah (Rp 1-5 miliar) 1 322
1,
6
Lebih besar -menengah (Rp 5–10 miliar) 9,6 324
D Besar (> Rp 10 miliar) 8,2 195
o Sektor
w Manufaktur 9,3 646
n Agribisnis 1 69
0,
l 3
o Jasa 1 618
a 1,
d 3
e Lokasi
d Perkotaan (kota) 1 617
1,
8
b Non-perkotaan (kabupaten) 1 716
y 0,
3
[ Orientasi ekspor
S Eksportirb 9,7 371
i Non-eksportir 1 962
m 0,
o 4
n Kepemilikan
Asingc 1 167
2,
F 0
r Domestik 1 1.164
a 0,
2
Sumber daya minyak/gas kabupaten
Minyakd 1 162
2,
8
Nonmigas 9,8 1.171
CBIEDec04 27/10/04 4:23 PM Page
(kota) 22,3 3
Jakarta Utara 20,2 10
Banjarmasin 18,8 4
Jakarta Selatan 18.6 29
Mojokerto (kota) 18,3 6
Jambi 16,3 8
Ujung Pandang 14,0 29
D Jakarta Pusat 13,4 32
o Pontianak (kota) 13,3 12
w Palembang 13,2 12
Samarinda 13,1 9
n
Padang 12,3 19
l
Manado 11,4 9
o Jakarta Barat 11,3 11
a Medan 11,2 44
d Semarang (kota) 10,4 38
e Bandung (kota) 10.1 59
d Malang (kota) 9.8 7
Jakarta Timur 9.4 59
b Gorontalo 9.1 10
y Batam 8.7 20
Bogor (kota) 8.4 12
Pekalongan (kota) 8.0 22
[
Surabaya 7.7 69
S Denpasar 7.5 17
i Bitung 7.4 9
m Bandar Lampung 5.9 15
o Surakarta 3.9 24
n Yogyakarta 3.8 19
karena beberapa kabupaten memiliki jumlah perusahaan yang sangat kecil (kurang
dari 10) dalam sampel dan, sebagai akibatnya, rata-rata suap sangat sensitif
terhadap keberadaan perusahaan outlier. Banjarmasin, misalnya, mencatat tingkat
suap tinggi sebesar 18,8%, tetapi ukuran sampelnya hanya tujuh perusahaan—di
mana hanya empat yang melaporkan positif suap.
CBIEDec04 27/10/04 4:23 PM Page
Kabupaten Rat N
a-
Banjar 25,0 5
Cianjur 22.1 15
Tangerang (kabupaten) 17,1 16
Lombok Barat 15,5 22
Sukabumi 14,7 17
Riau Kepulauan 13,7 19
D Kutai 13,6 7
o Bandung (kabupaten) 13,4 71
w Maros 12,7 15
n Musi Banyuasin 12,0 10
l Gianyar 11,8 10
o Mojokerto 11,7 24
a Deli Serdang 10,1 24
d Karawang 10,0 2
e Sidoarjo 9,8 29
d Bekasi 9,7 33
Batang Hari 9,7 10
Jepara 9,4 55
b
Bogor (kabupaten) 9,0 49
y
Badung 8,7 11
Bantul 8,5 11
[
Pasuruan 8,0 20
S Purwakarta 8,0 15
i Lamongan 7,4 40
m Pontianak (kabupaten) 7,4 8
o Semarang (kabupaten) 7,4 17
n Magelang 7,1 6
Klaten 6,9 27
F Lampung Tengah 6,6 15
r Malang (kabupaten) 6,3 14
a Gresik 5,5 28
Pekalongan (kabupaten) 5,1 28
Serang 4,6 6
Asahan 4,5 15
Pati 2,8 22
a
Rata-rata adalah hitung ed hanya dari perusahaan yang melaporkan suap positif. Berbeda
dengan tabel 1, nilai rata-rata yang ditampilkan di sini tidak berbobot: sebuah kabupaten
dengan 10 perusahaan dihitung sama dengan 30 perusahaan lainnya.
adalah bahwa semakin menguntungkan perusahaan, semakin tinggi suap yang akan
dibayarkan (sekali lagi setelah meningkatkan laba dan suap untuk ukuran
perusahaan).
CBIEDec04 27/10/04 4:23 PM Page
benar menerima apa yang dijanjikan oleh seorang pejabat mungkin tidak ada
hubungannya dengan tindakan pejabat itu dibandingkan dengan persaingan
CBIEDec04 27/10/04 4:23 PM Page
oleh suap
orang lain dari berbagai bagian birokrasi, seperti dalam kasus perusahaan bus
tersebut di atas. Jadi, jika ada kurangnya kepercayaan di pihak firma bahwa ia akan
menerima dispensasi yang dijanjikan, ia akan membayar suap yang relatif kecil—
tetapi dalam keadaan di mana ia mengharapkan harus membayar suap ke sejumlah
orang lain. pejabat juga. Jadi mungkin tidak ada korelasi yang jelas antara secara
keseluruhan dan kemungkinan suap yang dianggap efektif.
Berbagai karakteristik perusahaan itu sendiri mungkin diharapkan juga terkait
dengan pembayaran suapnya. Perusahaan yang berhasil dalam menegosiasikan
suap yang relatif rendah karena keterampilan tawar-menawar mereka akan lebih
menguntungkan daripada perusahaan lain yang identik dalam hal lain, dan
mungkin diharapkan untuk hidup lebih lama dan tumbuh lebih cepat (mengingat
bahwa pertumbuhan sangat bergantung pada laba yang diinvestasikan kembali).
Dengan kata lain, tingkat suap cenderung berkorelasi negatif dengan usia dan
D ukuran perusahaan, hal-hal lain dianggap sama. Kemungkinan yang terkait erat
o adalah bahwa pejabat yang mengambil pandangan jangka panjang dari pendapatan
w mereka dari penyuapan akan menyadari bahwa adalah kepentingan mereka sendiri
n untuk tidak membebankan biaya berlebihan pada perusahaan di mana mereka
l menjalankan otoritas pengaturan: jika perusahaan tersebut keluar dari bisnis, atau
o pindah, sumber pendapatan akan hilang dan usaha perlu dikeluarkan untuk
a menggantinya. Hubungan yang berkembang selama periode yang panjang antara
d perusahaan yang lebih tua dan pejabat yang korup dapat dicirikan oleh tingkat
e suap yang lebih rendah.
d Perbedaan antardaerah dalam hal sumber daya dapat mempengaruhi perilaku
korupsi. Dampak pada kualitas pemerintahan oleh pejabat lokal dari pengaturan
b desentralisasi baru yang mengalihkan sebagian besar pendapatan sumber daya
y alam kembali ke kabupaten sumber masih harus dilihat, tetapi mungkin tidak jauh
berbeda dari dampak ledakan sumber daya pada pejabat pemerintah di negara lain.
[ Ada banyak literatur yang menyatakan bahwa korupsi cenderung lebih mengakar
S di daerah yang kaya sumber daya (Gelb dkk. 1988; Karl 1999; Auty 2001; Eifert
i dkk. 2002). Secara khusus, Karl (1999) menyatakan bahwa sumber rente yang
m besar dan terkonsentrasi dalam pendapatan nasional dapat mengubah suatu
o negara menjadi 'negara renter', dan bahwa munculnya aliran pendapatan yang
n besar secara tiba-tiba dapat digunakan untuk mendukung pemerintahan
kleptokratis.
F Dalam konteks saat ini, sebagian rente dari sumber daya alam kemungkinan
r besar akan tumpah dan ditangkap oleh perusahaan-perusahaan di kabupaten yang
a bersangkutan—khususnya yang berhubungan langsung dengan perusahaan-
perusahaan yang terlibat dalam eksploitasi sumber daya alam. Dalam kasus seperti
itu, keuntungan supernormal akan dihasilkan dari kedekatan perusahaan dengan
pejabat di industri yang bersangkutan, bukan dari kemampuan wirausaha mereka
yang lebih murni, sehingga pejabat ini cenderung memiliki pengetahuan yang lebih
akurat tentang ukuran keuntungan yang tersedia. untuk pengambilalihan. Hal ini
memperkuat peran mereka dalam proses tawar-menawar, dan diharapkan dapat
menghasilkan tingkat suap yang lebih tinggi di distrik-distrik ini.
Sektor dan lokasi di mana perusahaan beroperasi dapat mempengaruhi tingkat
suap. Perwakilan Persatuan Hotel dan Restoran Indonesia (PHRI) menyarankan
agar bisnis semacam ini memperoleh sebagian besar pendapatan mereka dalam
bentuk tunai setiap hari, membuat mereka lebih mudah menjadi sasaran
pemerasan kecil-kecilan oleh pejabat daripada manufaktur dan agribisnis.
perusahaan yang menerima pembayaran untuk penjualan lebih jarang, dan
seringkali tidak dalam bentuk uang tunai. Ada juga kemungkinan bahwa asosiasi
industri dapat memberikan pengaruh untuk melindungi anggotanya dari pejabat
pemangsa, dan bahwa keefektifan asosiasi tersebut
CBIEDec04 27/10/04 4:23 PM Page
antar sektor. Data yang disajikan dalam tabel 2a dan 2b juga menunjukkan tingkat
suap yang lebih rendah di daerah non-perkotaan. Namun, kita tidak dapat melihat
alasan yang jelas mengapa tingkat suap harus berbeda menurut lokasi, dan
perbedaan yang diamati dalam tabel ini mungkin hanya menutupi pengaruh
perbedaan ukuran atau perwakilan sektor di daerah perkotaan dan non-perkotaan.
Mungkin ada perbedaan dalam jumlah suap yang dibayarkan tergantung pada
apakah perusahaan memproduksi untuk pasar ekspor atau apakah itu milik asing.
Sulit untuk berspekulasi tentang hubungan antara orientasi ekspor dan
pembayaran suap, tetapi satu kemungkinan adalah bahwa jika perusahaan
berlokasi di zona pemrosesan ekspor di mana ada lebih sedikit persyaratan formal
untuk semua jenis izin, mereka mungkin dilindungi. dari keharusan membayar
suap. Di sisi lain, dikemukakan di atas bahwa tidak perlu memiliki banyak
peraturan untuk memaksimalkan total suap yang dibayarkan oleh perusahaan, jadi
D bahkan jika perusahaan di zona pemrosesan ekspor menghadapi hambatan
o peraturan yang lebih sedikit, mungkin ada sedikit perbedaan dalam total suap.
w jumlah suap yang mereka bayarkan.
n Perusahaan asing mungkin lebih kecil kemungkinannya untuk membayar suap
l jika budaya perusahaan mereka menolak transaksi korup dengan pejabat, atau
o karena pembatasan yang diberlakukan oleh kantor pusat mereka. Misalnya, AS dan
a beberapa negara di Uni Eropa memiliki undang-undang yang secara eksplisit
d
melarang perusahaan untuk menyuap pejabat pemerintah asing. 11 Selain itu,
e
meskipun kepemilikan asing dapat membuat perusahaan lebih rentan terhadap
d
pemerasan oleh birokrasi, perusahaan asing biasanya memiliki mitra domestik
yang dipilih secara tepat karena kemampuan mereka untuk menangkal
b
pemangsaan birokrasi. Jadi beban kerentanan dapat berupa aliran dividen kepada
y
mitra domestik yang tidak memberikan kontribusi modal kepada perusahaan,
daripada suap yang dibayarkan langsung kepada pejabat yang korup.
[ Pertimbangan ini menunjukkan hubungan negatif antara suap yang dibayarkan dan
S kepemilikan asing.
i
m
o INVESTIGASI EMPIRIS
n
Pada bagian ini kami menggunakan kumpulan data SSG untuk memeriksa tingkat
korupsi di tingkat kabupaten pada tahun 2001–02, segera setelah undang-undang
F
tentang desentralisasi mulai berlaku. Sayangnya, kami tidak memiliki data tentang
r perilaku korup birokrat lokal di bawah rezim sebelumnya, yang dapat digunakan
a untuk mengukur tanggapan perusahaan terhadap perubahan perilaku birokrat
akibat desentralisasi. Instead we attempt to obtain a snapshot of the post-
decentralisation determi- nants of bribery, controlling for firm, location and
industry characteristics.
To ascertain the determinants of bribes paid by firms, a bribe function based on
the model described above is estimated econometrically. The estimating equation
is:
B = a + bX + cY + dZ + u (1)
Government-related Variables
The first four explanatory variables—tax payments, time spent, regulatory burden
and bribe efficacy—reflect government-related aspects of the environment faced by
firms (see appendix 1 for details of the survey questions).
Tax Payments. The model predicts that bribes will be larger for firms with higher
profitability. In the absence of data on the latter, we use as a proxy infor- mation on
a firm's tax payments (as a proportion of annual production costs).12
Time Spent. The time spent variable captures the proportion of their time firms'
managers spend negotiating with officials to obtain the most advantageous net
D dispensations. Respondents were asked to state, on a scale of one to six, the pro-
o portion of productive time devoted by management to interactions with officials in
w order to 'smooth business operations' in relation to regulatory matters (includ- ing
n taxation).
l Regulatory Burden. The model suggests that each different arm of the bureau-
o cracy may introduce regulations (and taxes) as instruments for generating bribe
a income for their officials. The regulatory burden is proxied here by the number of
d operational licences required for normal business operations.
e Bribe Efficacy. Respondents were asked to assess the likelihood that payment of
d bribes would guarantee receipt of 'promised services' (ie dispensations, in the
terminology we have employed here), on a six-point scale of responses ranging
from 'never' to 'very often'.
b
y
Firm-related Variables
[ The next set of explanatory variables reflects characteristics of the firms them-
S selves.
i Usia. Age is calculated from the date the firm was founded.
Ukuran. Size is indicated here by firms' annual sales in the previous year. Three
m
dummy variables are created to represent smaller-medium, larger-medium and
o
n large categories, while the small category is used as the reference case. 13
Location. A location dummy variable is included to distinguish firms in urban
F areas (kota) from those in non-urban areas (kabupaten).
r Sektor. Industry sector dummy variables are included to distinguish manufac-
a turing and services firms, while agribusiness is the reference case.
Export Orientation. Export orientation is represented by a dummy variable,
having the value 1 for exporters and 0 otherwise. A firm is classified as an exporter
if it exports at least 5% of its output.
Foreign Ownership. Foreign ownership is represented by a dummy variable,
having the value 1 for foreign firms and 0 otherwise. A firm is classified as for- eign
owned if it has at least 5% of its shares in foreign hands.
District-related Variable
Oil-Rich Districts. We investigate the link between bribery and natural resource
endowment by focusing on Indonesia's most valuable such resources, oil and gas. 14
'Oil-rich districts' are those for which the value of oil and gas production exceeded
5% of regional GDP in 1999; the oil dummy takes the value 1 for these
CBIEDec04 27/10/04 4:23 PM Page
Bribe rate 'Grease money' paid to officials 'Grease money' paid to officials 0.67
to get things done to set up a new business
D Number of licences
Number of operational required to set up a new
o
licences required to operate a business 0.40
w Regulatory business
n burden Difficulty of obtaining business
l Time spent with officials to permits
o smooth business 0.22
a operations
d Time Firms can predict the amount
e spent Firms receive the promised of bribes paid to officials
services after bribe is paid 0.22
d
b Bribe
y efficacy
[ a
See appendix 1 for precise questions.
S
i
m districts and 0 for others. (According to this definition there are five oil-rich dis-
o tricts among the 64 surveyed: Kutai, Riau Kepulauan, Musi Banyuasin, Batang Hari
n and Jambi.)
Although 1,808 firms were interviewed, for nearly one-third of these there was
insufficient information to allow their inclusion in the sample, leaving a total of
1,115 usable observations; of these, 1,027 reported positive bribe rates. Table 4
presents the Tobit estimates for the larger group, based on the assumption that the
zero bribe observations are correct. Table 5 presents the results of the two- stage
Heckman procedure, assuming that the zero observations are false. 18 As explained
above, the first stage involves a Probit regression that shows the influ- ence of each
of the independent variables on the firm's decision about whether to reveal having
paid bribes. The dependent variable takes the value of 1 if the reported bribe
payment is positive, and 0 otherwise. For each model specifica- tion, the first-stage
result is accompanied by its corresponding second-stage OLS regression in the
second column. The second stage uses the information obtained in the first stage to
undertake a modified OLS regression. The lambda is calcu- lated from the Probit
D
equation, and is used as an additional independent variable in the OLS regression to
o
w account for the assumed false zero bribe observations. 19 Table 6 presents the OLS
n regressions for the smaller group that excludes firms reporting zero bribe
l payments.
o These three econometric approaches yield generally consistent results in respect
a of most of variables, for which the coefficients are similar, as is the pattern of
d statistical significance. The main exception, so far as statistically significant
e coefficients are concerned, is the time spent variable: the coefficient obtained using
d OLS (about 1.8–1.9) lies between those obtained using the other two procedures
(about 2.3 for Tobit and 1.5–1.6 for Heckman). The Tobit procedure produces a
b higher coefficient because the regression function attempts to accommodate many
y observations that cluster at a zero bribe rate. On the other hand, the coeffi- cient
derived from the Heckman procedure is relatively small because of the assumption
[ that some of the zero bribe rates reported are actually false, and are corrected to
positive values. The other exception is the coefficient on large firm size, which is 4.3
S
in the Tobit estimates, but about 4.0 in the OLS and 3.8 in the Heckman estimates.
i
m Given the likelihood that the zero bribe rate observations include both true zeros
and actual positive values, it is probable that the Heckman procedure pro- duces
o
less bias than either OLS or Tobit. Judging from the direction of bias in the time
n
spent coefficient, the Tobit procedure appears to be the least reliable. The
coefficient on the variable lambda in the Heckman model is insignificantly differ-
F
ent from zero, however, which implies that the modified OLS regression obtained
r
from this procedure is not statistically different from the unmodified OLS regres-
a
sion reported in table 6, probably because many of the zero bribe observations are
actually correct.
The first-stage Probit equation concerning the decision about whether to reveal
the payment of bribes does provide valuable information, nevertheless. Note, for
example, that firms that perceive a high degree of bribe efficacy are significantly
less likely to reveal the payment of bribes, suggesting that many of the zero bribe
observations are indeed false.20 The equation also suggests that older firms are
significantly less likely to reveal paying bribes. Those firms are survivors in part
because of their ability to negotiate more generous net dispensations with offi-
cials, but they also appear to be reluctant to reveal their ongoing relationships with
corrupt officials. The other interesting result from the first-stage Heckman
CBIEDec04 27/10/04 4:23 PM Page
Model Specification
I II III IV
Explanatory variables
Constant –0.83 –0.01 0. 1.1
9 7
9
[– 0.51] [–0.01] [0.85] [1.0
1]
Tax payment 0.33 0.34 0.34 0.3
D [8.45] [8.79] [8.76] 4
o ** ** ** [8.7
w 3]**
n Time spent with officials 2.28 2.34 2.32 2.3
[7.07] [7.34] [7.20] 1
l
** ** ** [7.1
o
7]**
a Regulatory burden 0.10 0. 0. 0.1
d 1 0 0
e 4 8
d [1.56] [2.13] [1.23] [1.4
** 7]
Bribe efficacy –0.45 –0.46 –0.47 –
b
[– [– [– 0.46
y
1.65]* 1.70]* 1.74]* [–
1.71]*
[ Firm age –0.08 –0.09 –
S [– [– 0.09
i 2.67]** 2.91]** [–
m 2.93]*
o *
n Dummy for oil-rich regions 4.35 3.61 4.17 4.0
[3.93] [3.25] [3.85] 2
F ** ** ** [3.7
r 3]**
a Dummy for manufacturing firms 1.31
[1.08]
Dummy for service firms 2.21
[1.80]
*
Dummy for smaller-medium firms 0.
7
8
[0.95]
Dummy for larger-medium firms –1.72
[–
2.05]**
Dummy for large firms –4.25
[–
3.97]**
Dummy for urban location 0.61 0.
7
8
[0.89] [1.16]
CBIEDec04 27/10/04 4:23 PM Page
1]*
Lambda –5.99 –4.61
[–0.55] [–0.42]
Wald-chi2 196.09** 201.88*
*
No. of observations 1,115 1,115
Censored observations 88 88
Uncensored observations 1,027 1,027
a
Estimates are for all usable observations (1,115), and are based on the assumption that the
zero observations are false. Figures in parentheses are t-ratios.
**significant at 5%; * significant at 10%.
CBIEDec04 27/10/04 4:23 PM Page
Model Specification
I II III I
V
Explanatory variables
Constant –0.44 0.99 1.48 1.5
9
[–0.28] [1.35] [1.28] [1.3
9]*
Tax payment 0.38 0.39 0.38 0.3
D [9.66] [9.96]* [9.92]* 8
o ** * * [9.8
8]**
w
n Time spent with officials 1.79 1.86 1.84 1.8
[5.68] [5.81]* [5.84]* 3
l
** * * [5.8
o 1]**
a Regulatory burden 0.12 0.15 0.10 0.1
d 1
e [1.83] [2.40]* [1.60] [1.7
d * * 8]*
Bribe efficacy –0.10 –0.11 –0.13 –
b [–0.38] [–0.39] [–0.47] 0.12
[–
y
0.43]
Firm age –0.06 –0.06 –
[ [– [– 0.07
S 1.84]* 2.15]** [–
i 2.15]*
m *
o
Dummy for oil-rich regions 3.53 2.93 3.50 3.4
n
[3.27] [2.72]* [3.33]* 0
** * * [3.2
F 4]**
r Dummy for manufacturing firms 1.62
a [1.37]
*
Dummy for service firms 2.46
[2.19]
**
Dummy for smaller-medium firms 0.48
[0.59]
Dummy for larger-medium firms –1.62
[–
1.96]**
Dummy for large firms –3.97
[–
3.76]**
Dummy for urban location 0.17 0.35
[0.25] [0.53]
Dummy for exporting firms 0.56
[0.75]
CBIEDec04 27/10/04 4:23 PM Page
a
The constant term, bribe efficacy variable and dummy variables for smaller-medium firms,
manufacturing firms and urban locations are not significant.
b
Mean values are calculated only for the OLS sample (N = 1,027). The OLS sample is smaller
than the full sample of those reporting positive bribes (N = 1,333) because of miss- ing
values of independent variables.
D
o total (38% in model II). By contrast, the other variables make only minor contri-
w butions to the bribe rate.
n
l
o CONCLUSIONS AND POLICY IMPLICATIONS
a We have tried in this paper to shed some light on the nature of corrupt relation-
d ships between firms and government officials at the district level in the more
e decentralised system of government created by the enactment of the laws on
d regional autonomy. We developed a simple model in which firms seek to reduce
b
y
[
S
i
m
o
n
F
r
a
CBIEDec04 27/10/04 4:23 PM Page
could collect more tax revenue if they reduced other cost burdens on firms by
reducing the number of taxes, cutting back on unnecessary regulatory interven-
tions, and acting more vigorously to restrain corrupt behaviour on the part of the
bureaucracy. In turn, additional tax revenue could be used to raise civil servants'
salaries by more than these individuals' loss of income from corrupt activity: this
follows from the fact that a firm that is permitted to evade, say, Rp 1 million in tax
will pay significantly less than Rp 1 million to the official offering the dispen- sation.
The observations above about bureaucratic competition for bribe payments
suggest that this argument may be too optimistic, however. Tax offices located in
the districts and municipalities are not part of local government but, rather, part of
the central government machinery for collecting central government taxes such as
corporate and individual income tax. Thus an important implication of the
empirical analysis is that even if a district government succeeds in reducing
D corruption among its own officials (in the hope of attracting more businesses and
o generating more tax revenue at local level), central government tax officials may
w demand even larger bribes once it becomes clear that local officials are now
n extracting a smaller share of firms' supernormal profits. An important consider-
l ation here is that officials from the centre may have little concern for the longer-
o term impact of corruption on the development of the district. Like other central
a government employees, they are subject to rotation to other districts, so it is not in
d their interests during their term of duty to make their present district of resi- dence
e more competitive in the long term by reducing the overall financial burden on firms
d from illegal imposts.
Finally, in line with the large body of literature on the behaviour of govern-
b ments in oil-producing countries, our study suggests that officials in oil-rich dis-
y tricts are somewhat more corrupt than those in non-oil districts. This suggests that
there is scope for significant gains through requiring larger royalty payments for
[ the right to exploit natural resources, and through minimising the role of state-
S owned firms in this sector, since these provide an easy means of diverting natural
i resource wealth from its rightful owners—the general public. Our analy- sis does
m not enable us to conclude that oil-related corruption is a consequence of the new
o natural resource revenue sharing arrangements, despite some anecdotal evidence
n suggesting that corruption in oil-rich districts is on the increase. It will be
interesting to observe whether these districts can use their expanded revenues to
F accelerate growth and bring about improvements in social welfare, despite the
r apparently higher level of official corruption.
a
NOTES
* I am grateful to Andrew MacIntyre and two anonymous referees, and to seminar par-
ticipants at the Australian National University, for many helpful comments. All remaining
errors are mine. I would like to thank the team members of the University of Indonesia's
Special Survey on Governance, in particular, Thia Jasmina, Ari Damayanti, Isfandiari Jafar,
Isfandiarni, Suryadi and Eugenia Mardanugraha.
1 By mid 2004 the number of districts was closer to 400.
2 Firms that did not report paying bribes may in fact have paid them, of course, and this
has important implications for the analysis. These are discussed in detail later.
CBIEDec04 27/10/04 4:23 PM Page
rate is not caused by multi-colinearity with time spent, since the coefficient correlation is
only 0.12. Furthermore, the coefficient of time spent is statistically still strong.
23 The weakness of this result is not a multi-colinearity problem with either the regulatory
burden or the time spent variable, since their correlation coefficients are only 0.026 and
0.016 respectively.
24 There is a strong correlation between firm age and firm size, so age is excluded from the
model specification that includes size.
REFERENCES
Auty, Richard (2001), 'The Political State and the Management of Mineral Rents in Capital-
Surplus Economies: Botswana and Saudi Arabia', Resources Policy 27: 77–86.
Eifert, Benn, Alan Gelb and Nils Borje Tallroth (2002), 'The Political Economy of Fiscal Pol-
D icy and Economic Management in Oil Exporting Countries', World Bank Policy Research
Working Paper No. 2899, Washington DC, October.
o
Gelb, Alan, and associates (1988), Oil Windfalls: Blessing or Curse?, Oxford University Press
w for the World Bank, New York NY.
n Greene, WH (1993), Econometric Analysis, 2nd edition, Macmillan Publishing Company, New
l York NY.
o Heckman, JJ (1979), 'Sample Selection Bias as a Specification Error', Econometrica 47 (1):
a 53–161.
d Karl, Terry Lynn (1999), 'The Perils of the Petro-State: Reflections on the Paradox of Plenty',
e Journal of International Affairs 53 (1): 31–48.
d Lambsdorff, JG (2003), 'Background Paper to the 2003 Corruption Index', Transparency
International, accessed from <www.transparency.org/cpi/2003/cpi2003_faq.en.html>. Lewis,
Blane D. (2003), 'Tax and Charge Creation by Regional Governments under Fiscal
b
Decentralisation: Estimates and Explanations', Bulletin of Indonesian Economic Studies
y 39 (2): 177–92.
[
S
i
m
o
n
F
r
a
CBIEDec04 27/10/04 4:23 PM Page
I General information
Province: Kabupaten/Kota:
Year started:
Sector: Manufacturing Services Agribusiness
Foreign ownership: Yes/No If Yes, % of ownership
Export orientation: Yes/No If Yes, % of total sales
Sales last year:
1 = < Rp 1 billion 2 = Rp 1–5 billion 3 = Rp 5–10 billion 4 = > Rp 10 billion
Tax payments as a percentage of total annual costs:
D
o II Questions on key bribery variables
w Bribe rate
n In one year, how much 'grease money' do you pay to bureaucrats to get things done in
relation to licensing, taxation, police matters and so on, as a percentage of total production
l
costs?
o
a Time spent with officials
How much of its productive time does management spend with district officials to smooth
d
business operations: eg to negotiate regulations, licensing and other red tape?
e
1 = < 5% 2 = 6–15% 3 = 16–25% 4 = 26–50% 5 = 51–75% 6 = > 75%
d
Regulatory burden
How many operational licences are needed in your current business?
b
y Bribe efficacy
Evaluate the following statement:
[ 'A firm that has paid bribes will receive the promised services.'
S 1 = Never 2 = Very rarely 3 = Rarely 4 = Occasionally 5 = Often 6 = Very often
i
m III Supplementary questions to check consistency of responses on key
o variables indicated
Tax payments
n If there is a way to curb bribes, what is the maximum amount of tax you would be willing to
pay, as a percentage of annual costs, in exchange for no bribe-taking behaviour?
F
Bribe rate
r
How much 'grease money' do you pay to bureaucrats as a percentage of total initial invest-
a ment costs to set up a new business?
Time spent with officials
Evaluate the following statement:
'In the last two years, business permits have become harder to obtain.'
1 = Strongly disagree 2 = Disagree 3 = Tend to disagree 4 = Tend to
agree 5 = Agree 6 = Strongly agree
Regulatory burden
How many licences are required to open a new business?
Bribe efficacy
In relation to matters involving government offices in your region, to what extent are you
able to estimate in advance the amount of 'grease money' that will be paid to officials?
1 = Never 2 = Very rarely 3 = Rarely 4 = Occasionally 5 = Often 6 = Very often