What Determines Firms Decision to Formalize?

What Determines Firms Decision to Formalize? Empirical Evidence from Rural Indonesia Neil McCulloch Günther G. Schulze Janina Voss IDS, Univ. of Sussex/ UK Univ. of Freiburg/ Germany Fourth IZA/ World Bank Conference on Employment and Development Bonn, May 5, 2009

Motivation Informal sector is large most firms, much employment, particularly of the poor We don t know much about the pattern of formalization (who is formal and who isn t) We don t know much about the determinants of formalization (costs and benefits and how they vary by firm and owner characteristics) 05.05.2009 Formalization in Rural Indonesia 2

Structure of the Talk 1. Introduction 2. Our Data: The Rural Investment Climate Survey in Indonesia 3. The Empirical Approach 4. Costs and Benefits of Formalization 5. Who Goes Formal? 6. Concluding Remarks 05.05.2009 Formalization in Rural Indonesia 3

The Rural Investment Climate Survey (RICS) Survey in the field early 2006, data refer to 2005 Household, enterprise, community questionaires 2461 micro and small firms micro (1 4 empl., 2198 firms), small (5 19 empl.) Six districts throughout rural Indonesia, 149 villages Enterprises: 1757 household, 618 standalone, 146 listed Sector: 54 % trading, 35 % services, 11% manufacturing Manager/Owner: 63% male, 37 % female 05.05.2009 Formalization in Rural Indonesia 4

Business Licensing in Indonesia Complicated, unclear process, mainly at district level Required licenses: Tax identification number (NPWP) Physical licenses: Construction/building license (IMB) Sectoral licenses: trade license (SIUP), industrial registration (TDI) Business registration (TDP) Only 2 % of firms are fully licensed, 23% have at least one license. 05.05.2009 Formalization in Rural Indonesia 5

Cost Benefit Approach Firms decide to get licensed if benefits > costs Avg. Costs: 550,000 IDR, 11 days per lic. Benefits: Taxes, bribes, sales, access to credit, government contracts? Do benefits depend on firm characteristics? Stylized facts: Licensed firms are larger and older, they pay more taxes, bribes, have better access to credit and sell more to the govnmt. Endogeneity problem! 05.05.2009 Formalization in Rural Indonesia 6

Econometric Approach Endogeneity problem e.g., do firms pay higher taxes due to the license or do high tax payers decide to get a license to reduce the tax burden? IV approach, instrument: community averages for licensing Interaction with firm characteristics to capture firm heterogeneity 05.05.2009 Formalization in Rural Indonesia 7

1. Taxes 05.05.2009 (1) OLS (2) 2SLS (3) 2SLS licensed 1.045-1.423-3.760 (0.381)*** (0.581)** (1.116)*** L*qsales_2 1.147 (0.944) L*qsales_3 2.293 (1.184)* L*qsales_4 2.684 (0.995)*** L*qsales_5 2.997 (1.445)** education 0.342 0.388 0.411 (0.060)*** (0.096)*** (0.089)*** age 0.013 0.021 0.022 (0.011) (0.012)* (0.011)** female -0.404-0.359-0.389 (0.228)* (0.213)* (0.193)** Ind_ethn -0.189-0.374-0.470 (0.270) (0.251) (0.257)* chinese 0.826 0.705 0.327 (0.636) (0.880) (0.910) islam -0.469-0.587-0.676 (0.317) (0.345)* (0.334)** employee 0.105 0.099 0.082 (0.036)*** (0.045)** (0.044)* lnsales 0.251 0.347 (0.083)*** (0.059)*** lnfasset 0.087 0.089 0.092 (0.037)** (0.031)*** (0.028)*** Firm age -0.021-0.010-0.009 (0.015) (0.015) (0.014) villtax 0.804 0.819 (0.062)*** (0.061)*** Constant -3.056-3.950-0.572 (1.196)** (1.467)*** (1.357) Observations 1901 1782 1782 R-squared 0.33 0.42 0.43 Robust standard errors in parentheses Regression contains district dummies * significant at 10%; ** significant at 5%; *** significant at 1% Add. Controls for Sectors, Rural/Urban, Owner residing in village, district dummies 8

1. Taxes Licensed firms pay less taxes! This effect is smaller for larger firms. It is stronger for rural firms (not shown). Evidence for discrimination (religion, ethnicity). Controls: Large firms pay more. Village effects 05.05.2009 Formalization in Rural Indonesia 9

2. Informal payments (1) (2) (3) (4) (5) (6) No instruments License instrumented amount Did pay amount Did pay amount Did pay licensed 0.190 0.109 1.968 1.457 8.225 2.150 (0.307) (0.155) (0.958)** (0.464)*** (2.697)*** (1.445) L*lnsales 0.902 0.053 (0.234)*** (0.123) Resides in vill 0.551 0.088 1.165 0.664 1.417 0.662 (0.308)* (0.160) (0.447)*** (0.208)*** (0.415)*** (0.211)*** indethn 0.042 0.011 0.182 0.015 0.303 0.023 (0.338) (0.139) (0.308) (0.170) (0.282) (0.171) chinese 0.755 0.471 1.244 0.512 1.399 0.635 (0.621) (0.411) (0.705)* (0.529) (0.565)** (0.500) lnsales 0.070 0.174 0.170 0.130 0.394 0.111 (0.089) (0.043)*** (0.095)* (0.043)*** (0.118)*** (0.060)* lnfasset 0.045 0.017 0.079 0.011 0.080 0.010 (0.034) (0.017) (0.033)** (0.017) (0.031)*** (0.017) villcorrpt2 0.271 0.721 0.337 0.724 (0.227) (0.070)*** (0.212) (0.070)*** employee 0.065 0.073 0.071 (0.028)** (0.033)** (0.036)** Constant 3.579 1.474 2.056 0.747 0.502 1.818 (1.390)** (0.722)** (1.492) (0.783) (1.463) (0.801)** License Neighborhood effect discrimination Size effect Village effect Obs. 1901 1901 1782 1782 1782 1782 05.05.2009 10 Robust standard errors in parentheses Additional controls for sector, rural/urban, age of firm, female, islam, education, age, district dummies

2. Informal Payments Licensing reduces probability of paying bribes! It reduces amount of bribes as well! Esp. Large firms profit from reduction in corruption payments Controls: Owner residing in the village pays less bribes Larger firms pay more and are more likely to Chinese pay more, village effects 05.05.2009 Formalization in Rural Indonesia 11

3. Total Revenue Additional controls: sectors, rural/urban, female, Chinese, Islam, residing in village, district dummies (1) (2) (3) OLS 2SLS Licensed 0.424-0.379-1.226 (0.233)* (0.291) (0.349)*** L*qemployee_2 0.701 (0.594) L*qemployee_3 0.516 (0.502) L*qemployee_4 1.699 (0.465)*** edu 0.084 0.129 0.123 (0.030)*** (0.028)*** (0.027)*** age -0.009-0.005-0.005 (0.003)** (0.005) (0.005) indethn -0.294-0.089-0.108 (0.164)* (0.152) (0.161) qemployee_2 0.120 0.221 0.111 (0.178) (0.148) (0.128) qemployee_3 0.391 0.575 0.502 (0.176)** (0.158)*** (0.194)** qemployee_4 1.364 1.503 1.004 (0.204)*** (0.207)*** (0.287)*** qlnfasset_2-0.179-0.191-0.156 (0.176) (0.156) (0.155) qlnfasset_3-0.267-0.189-0.149 (0.150)* (0.137) (0.135) qlnfasset_4 0.064-0.056-0.013 (0.148) (0.153) (0.160) qlnfasset_5 0.097 0.113 0.137 (0.139) (0.130) (0.130) entagedum_2 0.445 0.436 0.438 (0.107)*** (0.107)*** (0.107)*** entagedum_3 0.434 0.475 0.482 (0.141)*** (0.172)*** (0.166)*** villsales 0.836 0.863 (0.095)*** (0.091)*** Constant 10.630 1.421 1.278 (0.441)*** (1.180) (1.111) Observations 1901 1782 1782 R-squared 0.39 0.45 0.46 Robust standard errors in parentheses Regression contains district dummies Education matters! Firm size Firm age 05.05.2009 * significant at 10%; Formalization ** significant at 5%; in *** Rural significant Indonesia at 1% 12

3. Business Expansion/ Revenue Again: Endogeneity Overall: No effect of formalization on revenue Large firms will gain from formalization, small ones will not. Controls: Factor input (labor matters) Education Firm age 05.05.2009 Formalization in Rural Indonesia 13

4. Further results 1. Access to credit largely unaffected by licenses 2. Access to government contracts: large firms will profit from licensing, no overall effect 05.05.2009 Formalization in Rural Indonesia 14

Determinants of Formality Previous results show which characteristics are associated with lower costs increased benefits. Reduced form Probit of Formality Sales, employment, assets, Sector, rural/urban Female, ethnicity, religion, Chinese 05.05.2009 Formalization in Rural Indonesia 15

Who goes formal? Results More likely: Large firms do (highest quintile in employmt, assets, sales) Chinese (by a third) Better educated owners Less likely Majority ethnicity (by 10%) Owners who live in the village (by 15%) Rural firms (by 15 20%) Cost differences (time, money) do not matter 05.05.2009 Formalization in Rural Indonesia 16

Concluding Remarks The main reason firms go formal is to reduce rent extraction! Firms that are easier targets (large, Chinese, minority ethnicity, out of village) have a bigger incentive to get a license 05.05.2009 Formalization in Rural Indonesia 17

Firms Decision to Formalize Probit: Marginal effects from probit regression on licensed (1) (2) Formalized? qlnsales_2 0.023-0.026 (0.052) (0.045) qlnsales_3-0.010-0.015 (0.051) (0.047) qlnsales_4 0.117 0.103 (0.067)* (0.063) qlnsales_5 0.221 0.161 (0.079)*** (0.074)** qemployee_2 0.048 0.050 (0.044) (0.043) qemployee_4 0.069 0.055 (0.054) (0.051) qemployee_5 0.175 0.087 (0.072)** (0.063) qlnfasset_2 0.106 0.121 (0.068) (0.069)* qlnfasset_3-0.005 0.002 (0.058) (0.061) qlnfasset_4 0.149 0.149 (0.070)** (0.068)** qlnfasset_5 0.206 0.183 (0.069)*** (0.067)*** female -0.022-0.016 (0.036) (0.036) indethn -0.102-0.091 (0.041)** (0.042)** islam -0.083-0.059 (0.084) (0.077) Chinese 0.353 0.379 (0.158)** (0.156)** rural -0.195-0.163 (0.031)*** (0.032)*** meancost 0.000 0.000 (0.000) (0.000) meantime 0.002 0.002 (0.002) (0.002) edu 0.045 edu 0.045 0.045 (0.010)*** (0.010)*** Residing in vill. -0.155-0.155 (0.064)** (0.064)** District dummies yes yes Observations 1851 1682 1682 Robust standard errors in parentheses Regression contains district dummies * significant at 10%; ** significant at 5%; *** significant at 1% Additional controls for sector and district dummies (0.010)*** 05.05.2009 Formalization in Rural Indonesia 18

ADDITIONAL MATERIAL 05.05.2009 Formalization in Rural Indonesia 19

Literature Levenson and Maloney (1998) older and larger firms invest in formality as they have proven to be successful Jäckle and Lee (2003) older and larger firms, Peru Fajnzylnber et al. (2006) quasiexperimental regres. Discont., Brazil, formalized firms have higher revenue, employm., investment McKenzie and Sakho (2006) Bolivia, IV approach formalization increases profits for small firms (2 5 wrks), micor and large firms lose 05.05.2009 Formalization in Rural Indonesia 20

RICS Geographical coverage Labuhan Batu, North Sumatra a plantation area Kutai, East Kalimantan an area rich in mineral resources Barru, South Sulawesi a forest fringe area Malang, East Java a rich agricultural area Badung, Bali a semi urban agglomeration area Sumbawa, NTB a dryland area 05.05.2009 Formalization in Rural Indonesia 21

Mean Size and Enterprise Age formal informal mean number of employees 3.93 2.23 mean of log total sales 11.03 9.92 mean enterprise age 10.77 8.54 05.05.2009 Formalization in Rural Indonesia 22

IV Approach: first stage Sargan test on overidentification passed Strong identification, high F statistic Community characteristics no strong instruments Village w/ high share of licensed firms may perform better, thus have higher benefits. Additional control for village averages for the respective benefit analyzed 05.05.2009 Formalization in Rural Indonesia 23

(1) (2) (4) (5) (6) First Stage IV Estimates Marginal effects from Probit regression on licensed Vill avg licensed 0.562 0.547 0.538 0.582 0.531 (0.086)*** (0.086)*** (0.081)*** (0.084)*** (0.094)*** education 0.039 0.039 0.039 0.039 0.037 (0.007)*** (0.007)*** (0.007)*** (0.007)*** (0.008)*** age 0.003 0.003 0.003 0.003 0.003 (0.001)*** (0.001)*** (0.001)*** (0.001)*** (0.001)*** female -0.017-0.017-0.017-0.015-0.016 (0.024) (0.024) (0.023) (0.024) (0.024) Resides in vill. -0.246-0.244-0.247-0.242-0.253 (0.076)*** (0.077)*** (0.074)*** (0.074)*** (0.074)*** indethn -0.085-0.084-0.081-0.085-0.081 (0.058) (0.056) (0.054) (0.055) (0.053) Chinese 0.056 0.048 0.046 0.047 0.049 (0.105) (0.105) (0.100) (0.100) (0.101) Islam -0.043-0.047-0.044-0.049-0.054 (0.046) (0.047) (0.046) (0.047) (0.049) Empl above median 0.085 0.086 0.108 0.087 0.081 (0.046)* (0.046)* (0.041)*** (0.046)* (0.044)* Sales above median 0.044 0.041 0.039 0.044 (0.032) (0.033) (0.033) (0.033) Fixed assets a.med. 0.069 0.070 0.069 0.070 0.068 (0.018)*** (0.018)*** (0.016)*** (0.017)*** (0.018)*** Firm age above med 0.014 0.013 0.017 0.014 0.015 (0.029) (0.029) (0.028) (0.028) (0.028) rural -0.015-0.015-0.011-0.005 0.001 (0.029) (0.028) (0.027) (0.026) (0.027) manufac -0.032-0.033-0.035-0.029-0.028 (0.028) (0.028) (0.028) (0.029) (0.028) service -0.028-0.029-0.037-0.025-0.028 (0.019) (0.019) (0.019)* (0.018) (0.018) bank 0.032 villtax -0.002 05.05.2009 Formalization in Rural Indonesia (0.007) villcorrpt 0.007 (0.012) villsales 0.018 (0.018) villgovs -0.008 (0.004)** villcred 0.104 Observations 1676 1676 1715 1676 1636 Robust standard errors in parentheses Regression contains district dummies * significant at 10%; ** significant at 5%; *** significant at 1% (0.024) (0.056)* 24

(1) (2) (3) (4) (5) (6) No instruments License instrumented amount Did pay amount Did pay amount Did pay 2. Informal payments licensed 0.190 0.109 1.968 1.457 8.225 2.150 (0.307) (0.155) (0.958)** (0.464)*** (2.697)*** (1.445) L*lnsales 0.902 0.053 (0.234)*** (0.123) edu 0.014 0.097 0.146 0.008 0.150 0.013 (0.083) (0.042)** (0.091) (0.055) (0.087)* (0.055) age 0.009 0.014 0.013 0.010 0.007 0.010 (0.012) (0.006)** (0.013) (0.007) (0.013) (0.007) female 0.275 0.234 0.396 0.221 0.383 0.229 (0.307) (0.147) (0.307) (0.150) (0.277) (0.150) Resides in vill 0.551 0.088 1.165 0.664 1.417 0.662 (0.308)* (0.160) (0.447)*** (0.208)*** (0.415)*** (0.211)*** indethn 0.042 0.011 0.182 0.015 0.303 0.023 (0.338) (0.139) (0.308) (0.170) (0.282) (0.171) chinese 0.755 0.471 1.244 0.512 1.399 0.635 (0.621) (0.411) (0.705)* (0.529) (0.565)** (0.500) islam 0.291 0.250 0.069 0.235 0.123 0.196 (0.397) (0.286) (0.402) (0.319) (0.372) (0.313) lnsales 0.070 0.174 0.170 0.130 0.394 0.111 (0.089) (0.043)*** (0.095)* (0.043)*** (0.118)*** (0.060)* lnfasset 0.045 0.017 0.079 0.011 0.080 0.010 (0.034) (0.017) (0.033)** (0.017) (0.031)*** (0.017) villcorrpt2 0.271 0.721 0.337 0.724 (0.227) (0.070)*** (0.212) (0.070)*** 05.05.2009 Formalization in Rural Indonesia employee 0.065 0.073 0.071 25

Who goes formal? 1. Licenses affect firms differently The costs and benefits of formality Interacting characteristic cost or benefit sales taxes (+) other levies : outcome stage ( ) sales to government (+) fixed assets other levies : outcome stage ( ) sales to government (+) employees credit ( ) revenue (+) female taxes (+) other levies : selection stage (+) rural taxes ( ) other levies : selection stage ( ) outcome stage (+) manufacturing sector credit ( ) sales to government (+) Ind_ethnicity other levies : selection stage ( ) Chinese sales to government ( ) Islam revenue ( ) other levies: outcome stage (+) 05.05.2009 Formalization in Rural Indonesia 26