Economic Effect of Infrastructure : macroeconomic effects and microeconomic effects Naoyuki Yoshino, Masaki Nakahigashi, Victor Pontines and Umid Abidhadjaev Asian Development Bank Institute ADBInstitute Tokyo, Japan, August
Economic Effect of Infrastructure Investment () Macroeconomics analysis () Micro-data approach Sources of Finance for Infrastructure Investment () by tax payers money; () use of national savings (or postal savings); Financial Inclusion Fiscal Investment and Loan Program (3) issue bond to construct infrastructures; general obligation bond, project bond (4) Public-Private-Partnership Too much borrowing from overseas might become the burden for the future. Accumulation of domestic Savings Which Method will induce better performance of infrastructure?
Map of Japan from the North to the South Hokkaido Hokuriku Tohoku South Kyushu North Kyushu Chugoku Shikoku Kinki Tokai North Kanto South Kanto Okinawa (not included) 3
Economic Effect of Public Capital Y t f ( Kp, E, Kg t t t ) Simultaneous regression of Translog Production Function and Labor Share Function (C) 4 Yoshino & Nakahigashi 4
Explanation of Direct and Indirect Effects Y t f ( Kp, E, Kg t t t ) Direct Effect (B A) Y highway construction Y Kg Indirect Effects (C B) Y New company Kg Y New employment Kg (C) 4 Yoshino & Nakahigashi 5
Economic Effect of Infrastructure Investment (Manufacturing Industry) 6
Economic Effect of Infrastructure (Services Industry) 7
Effectiveness of Public Capital Stock - Private capital/public capital ratio to Marginal productivity of Public capital - Secondary Industry (Industrial Sector).4 Southern Kanto Marginal Productivity of Public Capital.3.. Hokkaido Shikoku Tohoku Kinki Chugoku Northern Kanto Northern Kyushu Hokuriku Tokai Southern Kyushu..4.6...4.6 (C) Private 4 Yoshino Capital &/ Nakahigashi Public Capital
Marginal Productivity of Public Capital (in Japan) (C) 4 Yoshino & Nakahigashi
Thailand: Economic Effect of Infrastructure () Output Elasticity () Marginal Productivity Private capital Public capital Direct effect Indirect effect Private Public Direct Indirect effect capital capital Capital Labor effect Capital Labor Agriculture, forest, hunting and fishing Agriculture, forest, hunting and fishing 7-.7.77.6.6.74 7-.4.363.3..34 -..56.7.33.7 -.36.45.3..4 -.5..6 -.5. -..4.3 -.6.7 -.4 -.5. -.3. -. -.5.3 -.3. Manufacturing Manufacturing 7-.7.56...4 7-.343.67.7.56.5 -.63.46.63 -.4.66 -.33.4.7 -.3. -.554.4.35..3 -.3..75.76.7 -.63..73. -.35 -.64.447.5.535 -.73 Services Services 7-.74 -.3.3 -.7.45 7-.4 -.7.7 -..5 -.7 -.6. -.7.46 -.5 -.7. -.7.5 -.67 -.6 -.3 -.64. -.7 -.4 -. -.3.3 -.6 -.4 -. -.54.33 -.63 -.7 -.4 -.3.7
Micro Case Study - Philippine micro data, Evaluation of the highway effect on tax and non-tax revenues using as case study the Southern Tagalog Arterial Road (STAR) in Batangas Province, Philippines, Evaluation is carried out using a quasiexperimental approach via a difference-indifference (DiD) analysis
Case Study: Southern Tagalog Arterial Road (STAR) The Southern Tagalog Arterial Road (STAR) project in Batangas province, Philippines (south of Metro Manila) is a modified Built-Operate- Transfer (BOT) project. The 4. km STAR tollway was built to improve road linkage between Metro Manila and Batangas City, provide easy access to the Batangas International Port, and thereby accelerate industrial development in Batangas and nearby provinces.
Method: Difference-in-Difference (DiD) Analysis Outcome = + D + D T + where: D = (Treatment group) T = Treatment period D = (Control group) = Treatment Effect Assumption: Pre- Post Equal trends between Treatment and Control groups 3
Method: Difference-in-Difference (DiD) Timeline of STAR tollway: Analysis Outcome = + D + Inclusion of leads and lags D T + t+ 6 =, elsewhere t + 7 =, elsewhere t =, elsewhere t - =, elsewhere t - =, elsewhere t - 3 =, elsewhere t 4 forward =, 3 =, elsewhere 4
Outcome variable We employ data on property tax revenues, business tax revenues, regulatory fees and user charges of the cities and municipalities comprising Batangas Province, Philippines. The tax and non-tax revenues data were obtained from the Philippine Bureau of Local Government Finance (BLGF) 5
() Property tax Treatment D.55535 (.63) Treatment D.4** Period t+ Treatment D Period t+ Treatment D Period t Treatment D Period t- Treatment D Period t- Treatment D Period t-3 Treatment D Period t-4, forward (.5).447** (.6).47*** (.).4** (.674).63* (.645).7* (.).573*** (.) Difference-in-Difference Regression: Spillover () (3) (4) (5) (6) Property Business Business Regulatory Regulatory tax tax tax fees fees.736.67.43.37.4 (.74) (.36) (.47) (.3) (.46) -.3.***.**.4*** -. (.3).574*** (.).57** (.3).37 (.7).336 (.54).45 (.57). (.75) (.3).64*** (.45).44*** (.47).56** (.57).6** (.7).75** (.) 3.4*** (.) (.45).5*** (.54).64*** (.4).77** (.47).4** (.53).7*** (.544).56*** (.35) (.4).44** (.4).64** (.3).3** (.64).4** (.634).*** (.63).*** (.563) (.4).55*** (.6).64*** (.).3* (.44).44** (.43).3*** (.36).5*** (.45) (7) User charge. (.5).4*** (.3).37** (.64).35 (.7).5 (.74).4 (.74).73*** (.5).3*** (.67) () User charge.364 (.) -. (.5).434** (.67).4 (.5).7 (.56).47 (.53).676 (.55).77 (.745) Construction.3**.577.7.4* (.7) (.6) (.55) (.) Constant 4.6*** -.4 4.***.3 3.66*** 4.57 3.*** -.6 (.4) (.3) (.) (.4) (.7) (6.566) (.64) (7.4) N 73 7 73 73 77 73 R..4.37.44.43.5.6.3 Clustered standard errors, corrected for small number of clusters; * Significant at %. ** Significant at 5%. *** Significant at %. 6
The Southern Tagalog Arterial Road (STAR) Philippines, Manila 7
Uzbekistan: Railway
Regions Out Pre Post Diffe come railway railway rence period period Nonaffected group Affected Group GDP growth rate GDP growth rate.3.5. 7..4.
Qinghai-Tibet Railway Map
Tibet Railway R =
Japanese Bullet Train 3
Japanese Bullet Train Estimation results by group of prefectures Group Group 3 Group 5 Group 7 Group Con. Million JPY 5 5-5 Difference-in-difference coefficients across periods During Construction Period [-3] During st Phase of Operation [4-] During Phase of Operation [-3] Total Tax 663 6467 6454 Personal Income Tax 573-33 435 Corporate Tax 35-477 733 Other Taxes 65 77 576 Million JPY 3 5 5 5 Difference-in-difference coefficients estimated year by year Y Y Y3 Total Tax 6644 76 53343 Personal Income Tax 755 47 634 Corporate Tax 7 763 Other Taxes 34 77 75 Note: Numbers for tax revenue amount adjusted for CPI with base year. Pre-shinkansen construction period covers years from to. Non-affected groups include rest of the prefectures Treated groups: Group : Kagoshima, Kumamoto Group 3: Kagoshima, Kumamoto, Fukuoka Group 5: Kagoshima, Kumamoto, Fukuoka, Oita, Miyazaki Group 7: Kagoshima, Kumamoto, Fukuoka, Oita, Miyazaki, Saga, Nagasaki Group Con.: Kagoshima, Kumamoto, Fukuoka, Yamaguchi, Hiroshima, Okayama, Hyogo, Osaka 4
Impact of Kyushu Shinkansen Rail on CORPORATE TAX revenue during st PHASE OF OPERATION period {4-}, mln. JPY (adjusted for CPI, base ) 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 Variable Regression Regression Regression 3 Regression 4 Regression 5 Treatment -477.54 [-.] Number of tax payers 5.554* 5.5745* 5.6* 5.5355* 5.6645* [.5] [.5] [.5] [.5] [.] Treatment3-547. [-.7] Treatment5-35.4 [-.6] Treatment7-63. [-.7] TreatmentCon -3. [-.65] Constant -66567-6654 -66533-66535 -65553 [-.35] [-.35] [-.35] [-.35] [-.3] N 7 7 7 7 7 R.65.6.6.64.677 F.345.644.7454.67.4774 Note: Treatment = Time Dummy {-3} x Group. etc. t-values are in parenthesis. Legend: * p<.; ** p<.5; *** p<.. Clustering standard errors are used, allowing for heteroscedasticity and arbitrary autocorrelation within a prefecture, but treating the errors as uncorrelated across prefectures COMPOSITION OF GROUPS Group Group5 Kagoshima Kagoshima Kumamoto Kumamoto Fukuoka Group3 Oita Kagoshima Miyazaki Kumamoto Fukuoka Group7 Kagoshima Kumamoto Fukuoka Oita Miyazaki Saga Nagasaki GroupCon Kagoshima Kumamoto Fukuoka Osaka Hyogo Okayama Hiroshima Yamaguchi 5
3 Impact of Kyushu Shinkansen Rail on TOTAL TAX revenue during nd PHASE OF OPERATION period {-3}, mln. JPY (adjusted for CPI, base ) 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 Variable Regression Regression Regression 3 Regression 4 Regression 5 Treatment 6454.57*** [5.66] Number of tax payers.533***.37674***.447***.5767***.3737*** [4.5] [5.] [5.] [5.3] [6.] Treatment3 7334.*** [.77] Treatment5 36.*** [3.] Treatment7 47.6*** [3.5] TreatmentCon 4536*** [.] Constant -3534.6-336.37-336.7-3473.7-565.4** [-.5] [-.3] [-.3] [-.34] [-.3] N 6 6 6 6 6 R.33.4755.43.57375.763 F 6.4444.75 3.444 3.333.63 Note: Treatment = Time Dummy {-3} x Group. etc. t-values are in parenthesis. Legend: * p<.; ** p<.5; *** p<.. Clustering standard errors are used, allowing for heteroscedasticity and arbitrary autocorrelation within a prefecture, but treating the errors as uncorrelated across prefectures COMPOSITION OF GROUPS Group Group5 Kagoshima Kagoshima Kumamoto Kumamoto Fukuoka Group3 Oita Kagoshima Miyazaki Kumamoto Fukuoka Group7 Kagoshima Kumamoto Fukuoka Oita Miyazaki Saga Nagasaki GroupCon Kagoshima Kumamoto Fukuoka Osaka Hyogo Okayama Hiroshima Yamaguchi 6
3 Impact of Kyushu Shinkansen Rail on INCOME TAX revenue during nd PHASE OF OPERATION period {-3}, mln. JPY (adjusted for CPI, base ) 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 Variable Regression Regression Regression 3 Regression 4 Regression 5 Treatment 435.5** [.34] Number of tax payers 4.6776*** 4.6736*** 4.6634*** 4.63333*** 4.776*** [6.5] [6.6] [6.6] [6.6] [.] Treatment3 664.46** [.4] Treatment5 5675.3** [.5] Treatment7 46.336*** [3.] TreatmentCon 536.6** [.5] Constant -3766.6*** -3336.37*** -335.*** -3363.6*** -364.46*** [-3.] [-3.] [-3.] [-3.] [-4.5] N 6 6 6 6 6 R.5643367.57764.574.574.66737 F 5.7745 6.36477 6.44653 7.45 34.7555 Note: Treatment = Time Dummy {-3} x Group. etc. t-values are in parenthesis. Legend: * p<.; ** p<.5; *** p<.. Clustering standard errors are used, allowing for heteroscedasticity and arbitrary autocorrelation within a prefecture, but treating the errors as uncorrelated across prefectures COMPOSITION OF GROUPS Group Group5 Kagoshima Kagoshima Kumamoto Kumamoto Fukuoka Group3 Oita Kagoshima Miyazaki Kumamoto Fukuoka Group7 Kagoshima Kumamoto Fukuoka Oita Miyazaki Saga Nagasaki GroupCon Kagoshima Kumamoto Fukuoka Osaka Hyogo Okayama Hiroshima Yamaguchi 7
Impact of Kyushu Shinkansen Rail on CORPORATE TAX revenue during nd PHASE OF OPERATION period {-3}, mln. JPY (adjusted for CPI, base ) 3 4 5 6 7 3 4 5 6 7 3 4 5 6 7 3 Variable Regression Regression Regression 3 Regression 4 Regression 5 Treatment 733.** [.] Number of tax payers 5.57756*** 5.55543*** 5.5563*** 5.576545*** 5.647*** [3.3] [3.4] [3.4] [3.4] [3.7] Treatment3 4664.34* [] Treatment5 7.673** [.] Treatment7.365** [.34] TreatmentCon 763 [.5] Constant -5633.** -573747.** -57445.7** -57667.56** -643.7** [-.7] [-.] [-.] [-.] [-.] N 6 6 6 6 6 R.35653.355.3544.3574.364 F 5.65 5.467 5.357 5.43 6.555 Note: Treatment = Time Dummy {-3} x Group. etc. t-values are in parenthesis. Legend: * p<.; ** p<.5; *** p<.. Clustering standard errors are used, allowing for heteroscedasticity and arbitrary autocorrelation within a prefecture, but treating the errors as uncorrelated across prefectures COMPOSITION OF GROUPS Group Group5 Kagoshima Kagoshima Kumamoto Kumamoto Fukuoka Group3 Oita Kagoshima Miyazaki Kumamoto Fukuoka Group7 Kagoshima Kumamoto Fukuoka Oita Miyazaki Saga Nagasaki GroupCon Kagoshima Kumamoto Fukuoka Osaka Hyogo Okayama Hiroshima Yamaguchi
No Efforts Efforts to improve No Efforts (5, r) Operating Company Investors 度 (, r) Operating company Investors Efforts to improve (5, αr) Operating Investors company (, αr) Operating Investors Company
Private Financing for Infrastructure, Financial Inclusion Increase Domestic Savings Sell private financial products though post office Long term Savings: Insurance and Pension Funds, Too much reliance on overseas money will lead to debt overhang 3, Too much reliance on general budget will lead to budget deficits 3
Use of Pension Funds Pension Funds Ministry Of Finance Infrastructure Housing SMES Insurance Pension Funds Government Bond Infrastructure bonds Infrastructure Government Expenditures Pension Funds Infrastructure
Return From Highway Private Finance 7% Dividend /7 times Public Finance 3% Private Sector Investment (Pension) (Insurance) Money From Public Sector
Ratios of Pension Assets, Asian Development Outlook 5 33
Community Infrastructure Wind power Generator Funds Agricultural Farmer s Trust Fund Start-up business finance Local airport SME Hometown Investment Trust Fund Large Projects (highways, ports) Pension Funds, Insurance Funds Infrastructure Bond 34
Hometown Investment Trust Funds A Stable Way to Supply Risk Capital (i.e. knowledge base companies) Naoyuki YOSHINO Sahoko KAJI (ed.) 35
36
Bank-based SME Financing and Regional Financing to Riskier Borrowers () Bank Loans to relatively safer borrower () Hometown Investment Trust Funds E-Finance Start-up business, SME Safer borrowers Riskier Borrowers Banking Account Hometown Trust Funds Depositors Investors 37
Public Private Partnership (PPP) () Risk sharing between private and public sector () Incentive cut costs and to increase revenue Avoid political intervention Bonus payment for employees who run infrastructure (3) Many projects could be started by PPP Utilize domestic savings life insurance and Pension funds (long term) (4) Indirect Effects are important (tourism, manufacturing, agriculture, services) 3
Risks Associated with Infrastructure Risk sharing between private and public too much reliance on overseas money future burden for the country 3 Loans vs Investment 4 bankable projects or not? 5 Various Risks (political risk, operational risk, demand risk, ex post risk, maintenance risk, earthquakes, natural disaster risk) 3
Reference Yoshino, Naoyuki, Takanobu Nakajima and Masaki Nakahigashi () Productivity Effect of Public Capital", Yoshino, Naoyuki and Takanobu Nakajima (ed.) Economic Effect of Public Investment, Part, pp.3. (in Japanese) Yoshino, Naoyuki and Masaki Nakahigashi () Economic Effects of Infrastructure: Japan s Experience after World War II, JBIC Review, 3, pp.3. Yoshino, Naoyuki and Masaki Nakahigashi (4) The Role of Infrastructure in Economic Development, ICFAI Journal of Managerial Economics,, pp.7 4. 4
4