University. Georgia State University. International Studies Program. Vid Adrison

Transcription

1 Universiy Inernaional Sudies Program Working Paper 0-8 Ocober 00 The Effec of Money Supply and Governmen Expendiure Shock in Indonesia: Symmeric or Asymmeric? Vid Adrison Georgia Sae Universiy Andrew Young School of Policy Sudies

2 The Effec of Money Supply and Governmen Expendiure Shock in Indonesia: Symmeric or Asymmeric? Working Paper 0-8 Vid Adrison Ocober 00 Inernaional Sudies Program Andrew Young School of Policy Sudies Georgia Sae Universiy Alana, Georgia Unied Saes of America Phone: (404) Fax: (404) Inerne: hp://isp-aysps.gsu.edu Copyrigh 00, he Andrew Young School of Policy Sudies, Georgia Sae Universiy. No par of he maerial proeced by his copyrigh noice may be reproduced or uilized in any form or by any means wihou prior wrien permission from he copyrigh owner.

3 Inernaional Sudies Program Andrew Young School of Policy Sudies The Andrew Young School of Policy Sudies was esablished a Georgia Sae Universiy wih he objecive of promoing excellence in he design, implemenaion, and evaluaion of public policy. In addiion o wo academic deparmens (economics and public adminisraion), he Andrew Young School houses seven leading research ceners and policy programs, including he Inernaional Sudies Program. The mission of he Inernaional Sudies Program is o provide academic and professional raining, applied research, and echnical assisance in suppor of sound public policy and susainable economic growh in developing and ransiional economies. The Inernaional Sudies Program a he Andrew Young School of Policy Sudies is recognized worldwide for is effors in suppor of economic and public policy reforms hrough echnical assisance and raining around he world. This repuaion has been buil serving a diverse clien base, including he World Bank, he U.S. Agency for Inernaional Developmen (USAID), he Unied Naions Developmen Programme (UNDP), finance minisries, governmen organizaions, legislaive bodies and privae secor insiuions. The success of he Inernaional Sudies Program reflecs he breadh and deph of he inhouse echnical experise ha he Inernaional Sudies Program can draw upon. The Andrew Young School's faculy are leading expers in economics and public policy and have auhored books, published in major academic and echnical journals, and have exensive experience in designing and implemening echnical assisance and raining programs. Andrew Young School faculy have been acive in policy reform in over 40counries around he world. Our echnical assisance sraegy is no o merely provide echnical prescripions for policy reform, bu o engage in a collaboraive effor wih he hos governmen and donor agency o idenify and analyze he issues a hand, arrive a policy soluions and implemen reforms. The Inernaional Sudies Program specializes in four broad policy areas: Fiscal policy, including ax reforms, public expendiure reviews, ax adminisraion reform Fiscal decenralizaion, including fiscal decenralizaion reforms, design of inergovernmenal ransfer sysems, urban governmen finance Budgeing and fiscal managemen, including local governmen budgeing, performance-based budgeing, capial budgeing, muli-year budgeing Economic analysis and revenue forecasing, including micro-simulaion, ime series forecasing, For more informaion abou our echnical assisance aciviies and raining programs, please visi our websie a hp://isp-aysps.gsu.edu or conac us by a ispaysps@gsu.edu.

4 The Effec of Money Supply and Governmen Expendiure Shock in Indonesia: Symmeric or Asymmeric? Submied in Parial Fulfillmen of he Requiremens of ECON 8000C Dr. Sally Wallace By Vid Adrison Andrew Young School of Policy Sudies Georgia Sae Universiy 00

5 The Effec of Money Supply and Governmen Expendiure Shock in Indonesia: Symmeric or Asymmeric? I. Inroducion: The economic crisis in Indonesia has aken place for more han 4 years. While oher Asian counries such as Souh Korea and Malaysia have shown significan improvemen, Indonesia seems o have he slowes economic recovery process. Table shows he relaively poor economic recovery performance of Indonesia and ha he growh rae has no been resored o he pre-crisis level. Oher indicaors such as exchange rae - which is sill very volaile - also give he same conclusion. Table : ECONOMIC GROTWH THE SEVEN INDUSTRIAL COUNTRIES AND SEVERAL ASIAN COUNTRIES (PERCENT PER ANNUM) Group of Counries Qr.3 Qr.4 Qr. Qr. Qr.3 Seven Major Indusrial Counries Unied Saes Japan (.30) (0.0) (0.70)... Germany Unied Kingdom Ialy France Canada N I Es Korea Republic of (7.00) Hongkong (5.00) Taiwan (.40)... Singapore (0.90) (5.60) A S E A N Indonesia ) (3.0) Malaysia (7.40) Thailand 5.90 (.40) (0.80) Philippine (0.60) ) Based on 993 Consan Price Source : Indonesian Economic and Financial Saisic (Several Ediions), Cenral Bank of Indonesia In order o deal wih he economic crisis, he governmen has imposed several policies, boh moneary and fiscal. Governmen spending, which is one of he fiscal policies used in he Keynesian model, has no yielded a significan improvemen in aggregae demand. As can be seen in Table.a, nominal governmen expendiure has coninuously risen by a significan amoun since he crisis hi Indonesia in 997. However, he amoun of governmen expendiure decreased in real erms due o very high inflaion (Table.b). Theoreically, using only expansionary fiscal policy (i.e., increase in governmen expendiure) o simulae he economy would creae a crowding-ou effec. The governmen also imposed an expansionary moneary policy afer he exchange rae for domesic currency sabilized in he firs semeser of 998. The money supply and ineres raes are wo indicaors ha can be used o reflec he moneary policy imposed by he cenral bank. Ineres raes have been much lower compared o he 998 level. The

6 Cenral Bank discoun rae SBI - (one monh mauriy) which achieved 70,8 percen in 998 has gradually decreased unil i reached is minimum rae in April 000 of percen, bu hen sared o rise again unil November 00 o 7.6 percen. The reacion of he economy o moneary policy is no differen from he reacion o fiscal policy. Even hough he ineres rae has been much lower han in 998, invesmen spending has no been resored. Table.a: GDP Indonesia By Expendiure a Curren Price Expendiure HH CONSUMPTION 83, , , , , , , ,997.0 GOV CONSUMPTION 9, , , ,99. 4, ,45.9 7, , INVESTMENT 86, , ,8.3 57, ,700.4, , ,95.0 CHANGE IN STOCK,908. 8, ,59.4 6,37.5 7,859.7, , ,39.0 EXPORT 85,96. 99, , , , , , ,58.90 IMPORT 78, , , ,8.0 76, , , ,07.50 GDP 39, ,9.6 35, , ,505.9,00,333.,9,44.0,90, * Firs Three Quarer Source: Cenral Bank of Indonesia Table.b: GDP INDONESIA BY EXPENDITURE AT CONSTANT 993 MARKET PRICE BILLIONS OF RUPIAH * HH CONSUMPTION 83, ,445. 5, ,79. 73, ,9.7 67, ,957.4, GOV CONSUMPTION 9, ,44.6 3, ,68.4 3, ,87.9 7,04.3 8,767.8,88.00 INVESTMENT 86, , ,0. 8, , , , , ,08.0 CHANGE IN STOCK,908. 9,6.0 3, ,79. 7, , ,8. -6,38.3-7, EXPORT 85,96. 95, ,974.8,39.4, ,707. 9,3.6 06, , IMPORT 78, , ,937.8, ,796. 3, , ,8.6 80, GDP 39, , , , , , , , , * Firs Three Quarer Source: Cenral Bank of Indonesia 3

7 Graph : Quarerly GDP of Indonesia (based on 993 Consan Price) 0, ,000.0 Billions of Rupiah 80, , , ,000.0 HH CONSUMPTION GOV CONSUMPTION INVESTMENT CHANGE IN STOCK EXPORT IMPORT GDP 0.0-0,000.0 I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III ** 00*** Year Source: Processed by auhor using he daa from Cenral Bank of Indonesia The economic crisis in Indonesia is inerwined wih a poliical crisis, which makes he recovery process more difficul. I is raional ha economic recovery is highly correlaed wih and mus be suppored by poliical sabiliy. However, his high correlaion causes mos Indonesians (including he economic observers) o insanly blame poliical facors for he negaive movemen of economic indicaors. Slow economic recovery is believed o be he resul of he poliical crisis. Looking a he slow economic recovery and ignoring he poliical crisis for a momen, we encouner a big quesion: Is here a possibiliy of an asymmeric effec of moneary and fiscal policy on economic aciviy? If i does exis in Indonesian economy, hen we canno blame he slow economic recovery only on a poliical crisis. Besides, knowing he exisence of an asymmeric effec is very imporan. From he policy poin of view, failure o allow for asymmeric effecs on oupu migh resul in he erroneous conclusion ha disinguishing beween anicipaed and unanicipaed moneary policy is unimporan (Chu, 997). There are wo objecives o be achieved in his paper. Firs, deermining he exisence of an asymmeric reacion of oupu and inflaion o governmen expendiures and he money supply. Second, finding he bes policy o deal wih he economic crisis, i.e., deermining which policy generaes a larger and quicker response o oupu. This paper is divided ino five secions. Secion I discusses he background and objecives of his paper. Secion II conains an overview of governmen expendiures and he money supply in he pre-crisis year. The heoreical background and previous sudies abou asymmery are discussed in secion III. Secion IV examines he mehodology, which 4

8 conains daa used in he esimaion process. Secion V conains he analysis of he esimaion from secion V. The conclusions and policy implicaions are included in secion VI. II. Governmen Expendiure and Money Supply in Indonesia: An Overview We define fiscal policy as used in his paper as he amoun of governmen spending by all levels of governmen, which is shown in he naional accoun from he expendiure side. The revenue side of he governmen budge is also a par of fiscal policy. The revenue colleced by he governmen depends on he ax rae and he ax base. Changing he ax rae in order o increase revenue collecion will ake considerable ime, since i mus be approved by he parliamen and requires debae beween governmen and parliamen. If he changes in ax raes are realized, people will no consider i as a shock since hey will already be prepared. Since our focus is he effec of unanicipaed policy shock, we consider governmen expendiures as represenaive of he shock insead of governmen revenue. There are several policies ha can be caegorized as insrumens in moneary policy. In his paper we use he M money supply as proxy for moneary policy. Since M is a more sable money supply measure (Case and Fair, 00) i is expeced ha using he shock of M would give us a more sensiive impac on oupu - if i does exiss han he M. II.a. Governmen Expendiure Graph oulines he conribuion of governmen spending in aggregae demand for he Indonesian economy. The share decreases over ime, which maches economic heory. Saring in 987, he share has reached a single digi (excep for 988), eiher in nominal or real erms (using 993 consan prices). The revenue and expendiures of all levels of governmen for fiscal year 994/995 o 998/999 are presened in able 3, as well as several oher raios in governmen financial operaion. We can see he considerable conribuion of cenral governmen, eiher from he revenue side or from he expendiure side. The share of cenral governmen revenue from oal consolidaed governmen revenue always exceeds 90 percen; for he expendiure side is share is consisenly above 78 percen. More han half of all governmen expendiures are allocaed o rouine expendiure. The share of rouine expendiures gradually increased and reached is highes share in fiscal year 997/998. The reason for his significan increase is local currency depreciaion, We find ha using M as indicaor of money supply would produce he same resul. Therefore in his paper we only presen he oupu using M as money supply indicaor. We canno presen all he daa in he ime range of observaion ( ) due o he daa availabiliy for local governmen financial operaion. 5

9 which led o a huge increase in ineres raes and principal deb paymens and an increase in subsidy expendiures in he cenral governmen budge. Graph : Share of Governmen Expendiure on GDP % 8.00 Real Nominal Year Source: Processed by auhor using he daa from Cenral Bureau of Saisics Table 3: Several Measures of Governmen Financial Operaion Fiscal Year 94/9595/96 96/97 97/98 98/99 Cenral Governmen Share of Revenue on Toal Consolidaed Revenue Share of Expendiure on Toal Consolidaed Expendiure Rouine Expendiure Share on Cenral Governmen Expendiure Developmen Expendiure Share on Cenral Governmen Expendiure Local and Provincial Governmen Share of Revenue on Toal Consolidaed Revenue Share of Expendiure on Toal Consolidaed Expendiure Rouine Expendiure Share on Toal Local and Provincial Expendiure Developmen Expendiure Share on Toal Local and Provincial Expendiure Source: Processed by auhor using he daa from Cenral Bureau of Saisics II.b. Money Supply Graph 3 shows he raio of he money supply (measured by M and M) o quarerly GDP. The raio of M o GDP shows a posiive rend, while he raio of M o GDP shows a relaively horizonal rend 3. The raio of M o GDP sared o increase when he governmen abolished ineres rae resricion in 983. A furher increase ook place afer 3 The raio of M o GDP exceeds uniy since we used quarerly GDP. If we use he annual GDP, he raio would fall o around

10 he governmen deregulaed he banking secor in Ocober 988, which simulaed he esablishmen of new banks..5 Graph 3: Raio of M and M o GDP M/GDP M/GDP III. Theory: Source: Processed by auhor using he daa from Cenral Bank of Indonesia and Cenral Bureau of Saisics III.a. Previous sudies invesigaing he exisence of asymmeric effec The sudy of he asymmeric effecs of a paricular variable or policy has been conduced over a long period of ime by scholars. Johnson in 96 recognized ha igh and easy moneary policy produces asymmeric effecs on economic aciviy. Sudies conduced by De Long and Summer (988), Cover (99), Morgan (993) and Karras (996) suppor he conclusion ha a negaive moneary shock (eiher a decrease in he money supply or an increase in ineres raes) reduces oupu more han moneary expansion raises i. Garibaldi (997) sudied he asymmeric effecs of moneary policy on job creaion and desrucion using cross-counry daa. He found ha he effec of an increase in ineres rae is immediaely ransferred ino job desrucion. Conversely, he effec of easing moneary policy produces a slow response in job creaion, and in paricular, does no resul in a one-ime increase in jobs as much as one-ime decrease in jobs brough abou by higher ineres rae. Kandil (000) sudied he effec of exchange rae flucuaions on oupu using crosscounry daa. Theoreically, exchange rae flucuaions deermine aggregae demand hrough inernaional rade (expors and impors) and aggregae oupu hrough he cos of impored goods. He found ou ha an unanicipaed posiive exchange rae shock (local currency devaluaion) leads o oupu conracion due o he significan increase in price 7

11 inflaion. In conras, unanicipaed appreciaion does no resore oupu. I even decreases oupu, since ne expors decrease in he demand channel, while on he oher side, people are less willing o hold domesic currency, which conribues o price inflaion. III.b. Facors ha cause he asymmeric effec o exis Mos of he microeconomic heory suggess ha he reacion of individual behavior changes are symmeric - i.e. o he change in quaniy demanded due o he change in he relaive price of a cerain good. This is no he case in macroeconomic heory. There are several facors ha conribue o he exisence of asymmeric effecs of moneary and fiscal policy. The firs source is price rigidiy. Policy has an asymmeric effec if prices are less flexible downward han upward. In macroeconomic heory, his can happen when he aggregae supply curve is convex. This asymmeric reacion can also be explained using he menu cos approach. In an economy wih a posiive rend of inflaion, a posiive shock o firms generaes a greaer adjusmen han a negaive shock of he same size. Inflaion causes he relaive prices of he firm o change. If here is a posiive shock, firms have o adjus heir relaive prices and i creaes menu coss, which do no happen in presence of a negaive shock. A shif in aggregae demand can also creae an asymmeric effec due o price rigidiy. If aggregae demand increases, oupu will no increase significanly since prices adjus quickly. Informaion can also be he source of an asymmeric effec. For insance, in a banking secor wih a relaively high ineres rae, banks would be less willing o lend money o risky borrowers. This behavior resuls in credi raioning and a fall in oupu in a way ha does no have a counerpar during a period of easy lending policy (Jackman and Suon, 98) III.c. The imporance of recognizing he asymmeric effec for Indonesian economy If he asymmeric effec does ake place in he Indonesian economy, hen he governmen has o be more careful in implemening policy. For moneary policy, a negaive shock (eiher a decrease in he money supply or an increase in ineres raes) would cause a significan drop in oupu, while a posiive shock could only resul in a sligh increase in oupu and could be ime consuming. A negaive shock in fiscal policy would reduce he oupu of he economy. Thus, he governmen would have o spend more money han he amoun i has colleced in order o resore oupu o he iniial level. This is problem since Indonesia already has a high deb burden. IV. Mehodology: The echnique ha has been developed in order o discern he presence of asymmeric effecs of a policy is a parial equaion model. The mos well-known echnique o 8

12 invesigae he response of he economy if here were an asymmeric impac is TARCH (Threshold Auo Regressive Condiional Heeroscedasiciy). This echnique was developed by Zaokian in 990. However, in his paper, we are going o use an approach developed by Cover (99) and refined by Kandil (000) o find he asymmeric effec of exchange rae flucuaion. Using his approach we can disinguish he policies ha have an asymmeric effec. The above models will esimae effec of he money supply, governmen expendiures, and exchange raes on wo main macro indicaors: oupu and he price level. Over ime, real oupu flucuaes around is seady sae in response o aggregae demand shocks, energy prices, he money supply and governmen expendiures. Shocks from each variable are assumed o be randomly disribued over ime. To invesigae he asymmeric effec, we assume ha each variable is symmerically disribued around is seady sae value. A posiive shock is defined when a value of a variable in a cerain ime exceeds is rend value, and conversely for he negaive shock. IV.a. Daa In his paper, quarerly daa for Indonesian from 980: 997:4 is used. The reason for choosing his period is o neuralize he poliical facors which occurred beginning in 998. We expec he reacion of he Indonesian economy o hese hree facors o reflec he reacion o he fundamenal economic variables and o no be influenced by poliical facors. The daa are colleced from he Cenral Bank of Indonesia (BI), Cenral Bureau of Saisics and Deparmen of Finance, and Minisry of Mining and Energy. The oupu daa used in his paper is based on 993 consan prices; for he consumer price index (CPI) we use he same year as he benchmark (993=00). For he price of energy, which will be used in he inflaion esimaion, we calculae he weighed average price for each ype of oil. This daa is colleced from Minisry of Mining and Energy, and he monhly consumpion daa can only be obained from 997. The daa available from 980 are only annual daa. Therefore we divided he consumpion of each fuel ype by a facor of four o esimae he quarerly consumpion. In order o ge he weighed price, we muliply he price of each fuel ype wih is share of oal fuel consumpion. IV.b. Esimaion Process The firs sep is o find he seady sae for oupu, inflaion, he money supply, ineres raes, governmen expendiures and exchange raes. This is done by using he Hodrick Presco filer echnique, a widely used smoohing parameer among macroeconomiss. The formula of Hodrick Presco Filer is as shown in he equaion below, and i chooses he smoohed series ( s ) by minimizing he variance of y around s. The penaly parameer ( λ ), which conrols he smoohness of s series, akes a value of 600 for quarerly daa. 9

13 T = T ( y s ) λ (( s s ) ( s s )) = The resuls of he smoohed series are presened in graph Afer obaining he seady sae of each variable, we hen define he posiive and negaive shocks for governmen expendiures and he money supply. The posiive shock akes he value of he difference beween acual governmen spending and is seady sae value if i is posiive; oherwise i akes he value of zero. The negaive shock akes he value of he difference beween acual governmen spending and is seady sae value if i is negaive; oherwise i akes he value of zero 5. Since we expec he sign of negaive shock o be negaive on he dependen variable, we ake is absolue value. For he exchange rae variable, posiive shock refers o he depreciaion of local currency agains US dollar, while negaive shock refers o appreciaion. For he case of moneary policy, we choose he money supply as an indicaor of moneary policy, insead of he ineres rae. The reason behind his decision is ha he ineres rae in Indonesia was pegged by he cenral bank before 983. Therefore, using he ineres rae as a policy insrumen in his research would limi he period of observaion. 4 Noe ha here is a weakness in esimaion of seady sae of GDP. Theoreically, is seady sae value should reflec he poenial oupu. Therefore, naural rae of unemploymen should be considered in he calculaion of poenial (seady sae of) GDP. However, he problem is ha here is no daa available for unemploymen for quarerly daa. 5 The formula o define posiive shock is pos = [ abs( m ) m ] neg m = ) [ abs( m m ] m and negaive shock is 0

14 Graph 4: Acual and Trend of Real GDP of Indonesia Billions of Rupiah GDP GDPTREND Graph 5: Acual and Trend of CPI of Indonesia (993=00) CPI CPITREND

15 Graph 6: Acual and Trend of Money Supply (M) of Indonesia Billions of Rupiah M MTREND Graph 7: Acual and Trend of Shor Run Ineres Raeof Indonesia 5 0 % DRTREND DR

16 Graph 8: Acual and Trend of Nominal Governmen Expendiure of Indonesia Billions of Rupiah GOVN GOVNTREND Graph 9: Acual and Trend of Nominal Exchange Rae of Indonesia (Agains US Dollar) Rupiah/US Dollar EXRATE EXRATETREND 3

17 IV.b.i. Oupu Esimaion In order o discern he oupu reacion, we conduc several regressions. There are wo general ypes of regressions conduced in his paper. Firs is he esimaion of he level form of he dependen variable. This ype of regression gives us an indicaion ha he asymmeric reacion does ake place in Indonesian economy. However, one problem wih his regression resul is auocorrelaion. Theoreically, he presence of auocorrelaion will no cause he esimaed parameer o be biased, bu we canno make an inference. The second ype of regression uses he difference form of he dependen variable. The problem of auocorrelaion is solved, bu since he dependen variable akes he difference value, while he independen variables are in he same form wih he firs ype of regression, we have o be careful in inerpreing he resul. In order o find evidence of an asymmeric effec, we firs conduc a join hypohesis for each shock (F es). Tha is, we wan o discover wheher he effec of posiive shock from he curren period o he lags included is joinly saisically significan in affecing oupu. The same procedure applies o negaive shocks. Afer conducing an F Tes for posiive and negaive shocks, we hen es wheher he magniude of posiive and negaive shocks is saisically significan. Summary of regression resuls are presened in he appendix. The equaions for each esimaion are also presened in he appendix. The summary of resul is presened in ables I and II. Running he regression from equaion.a, gives us he resul ha he effec of governmen expendiures (GOVN), he money supply (M) and exchange raes (EXRATE) are posiive o he level of oupu (which means ha he depreciaion of local currency affecs he level of oupu). However, noe ha his resul is due o he exisence of auocorrelaion problem, since he Durbin-Wason Saisic is far below he bes hreshold (.9 compared o ). Running he regression in difference form (equaion.b) changes he probabiliy level (he independen variables become insignifican affecing he dependen variable), bu on he oher hand, his reduces he auocorrelaion problem. Table III presens he esimaion resuls for oupu in he logarihm form, while he able IV presens he esimaion resul for growh of oupu. From he saisical poin of view, he resul presened in able III is much beer han in able IV. I is shown by Aikaike Informaion Crierion (AIC) ha able III is lower han he resuls presened in able IV. The only variable ha does no significanly influence oupu is governmen expendiures. According o he oher resuls, only depreciaion of exchange rae affecs he growh of oupu. Running he regression from equaion 3.a also generaes a serious auocorrelaion problem as shown by he low DW saisic value. Therefore, even hough many of he variables seem o have a probabiliy below 0 %, we canno make any inferences due o ype I errors. Running he regression from equaion 3.b gives us beer resuls, since he auocorrelaion is no so severe as he previous regression. The only significan variable affecing oupu is negaive shock of governmen expendiure (NEGGOV), bu he 4

18 direcion is no as we would expec. This unexpeced sign is perhaps due o he low predicabiliy of he model, shown by he low value of he adjused R-square. Table V shows he resul of oupu reacion in difference form (from equaion 4.b). Auocorrelaion is no a problem in his esimaion, shown by he DW saisics near o he benchmark (.4 compared o ). The effec of posiive shocks in governmen expendiures in he curren period (POSGOV) is saisically significan in increasing oupu. Surprisingly, he sign of negaive shock effec of governmen expendiure in he curren period (NEGGOV) is posiive for a change in oupu. Using AIC saisics, he lag srucure in equaion 4.b gives us he bes response of oupu o he shock of changes in he money supply and governmen expendiures. In he nex sep, we compare he resul of excluding exchange rae using he same lag srucure as equaion 4.b. The reason for his is ha Indonesia imposed a managed floaing regime unil 997. The regression resul is presened in able VI. This regression provides us beer saisics han able V (including he exchange rae variable). Using he join hypohesis (F-es), we find ha he shock of governmen expendiures has a posiive effec on changes in oupu. The negaive governmen expendiure shock is no saisically significan in reducing oupu (F-es for join hypohesis is 0.88). If we conduc an asymmeric es, we would conclude ha governmen expendiures in Indonesia have a posiive asymmeric effec on oupu, i.e. a posiive shock in governmen expendiure increases oupu while negaive shock would no reduce oupu. If we sum up he coefficien of posiive shock from governmen expendiures, we would ge he muliplier effec from posiive governmen expendiure shock. The value of his muliplier is IV.b.ii. CPI and Inflaion Esimaion The summary of CPI and inflaion esimaion resuls are presened able II. Table VII presens he inflaion esimaion from equaion 5.b. From a saisical poin of view, his resul is very good since i produces a low AIC (-6.8) and a very good DW saisic (.06). From his esimaion we can conclude ha he growh of governmen expendiures and he money supply do no have saisically significan effecs on he inflaion rae. The depreciaion of he exchange rae has a saisically significan effec on inflaion, as well as on he percenage change in oil price. Table VIII shows he resul of inflaion esimaion, defines shock and uses no lags. Even hough he DW saisic is close o and he AIC is low, none of he variables used in he regression are saisically significan on he inflaion rae. Table IX provides us he reacion inflaion o shocks using wo lags for each variable. From his able, only exchange rae and fuel price shock are saisically significan in affecing he inflaion rae. V. Findings 5

19 V.a. Oupu esimaion Using he Akaike Informaion Crierion o choose he bes esimae, we would conclude ha he oupu esimaion - wih respec o he shock of governmen expendiures and he money supply - presened in able VI is he bes. From ha regression, he effec of governmen expendiure is asymmeric; forunaely he effec of posiive shock is larger han ha of negaive shock. This implies ha if he governmen does no increase is expendiure in he nex period, we do no have o worry abou he oupu conracion. The effec of he money supply shock eiher posiive or negaive - is no saisically significan o oupu. V.b. CPI and Inflaion Esimaion Using he Akaike Informaion Crierion o choose he bes esimaion, we can conclude ha he resuls presened in able IX are he bes esimae, since hey give us he lowes value of AIC, which is Changes in governmen expendiures and he money supply do no conribue significanly o he inflaion level, while wo imporan variables affecing he level of inflaion are fuel price and exchange rae. The inflaion esimaion in his paper is far from perfec, bu his resul a leas gives us an iniial indicaion ha inflaion in Indonesia can be explained from a srucural approach. V.c. Fiscal Policy in Crises Year As shown in able.a, nominal governmen expendiures rose significanly afer he crisis hi Indonesia. However, he amoun of governmen expendiures in real erm (using 993 consan prices) decreased due o very high inflaion. The share of governmen expendiures in GDP in he period of ranges from 5.4 % o 0.4% in nominal erms, while in real erms ranges from 7. o 9.0 percen. The share in real erms is always bigger han in he nominal erms (excep for during 995), meaning ha inflaion for he oher componens of GDP is relaively high compared o governmen spending. The regression resuls presened in able VI conclude ha a shock in governmen expendiures has a significan effec and ha he effec is asymmeric (posiive shock generaes a larger effec on oupu han a negaive shock does). The quesion hen: Why here was no significan improvemen in oupu given ha governmen expendiure rose significanly during he crisis year? The explanaion arises by looking ino he deails of he governmen budge. If we ake a look a he governmen budge in he crisis period, here is a significan change in he spending composiion. This happens in all levels of governmen. Graphs 0.a - 0.c presen he composiion governmen spending for cenral, provincial and local governmens respecively, for he period of 994/995 o 999/000. The share of rouine expendiures has an upward rend, and consequenly he share of developmen 6

20 expendiure has a downward rend. In he local governmen, he share of rouine expendiures is almos wo hirds of is oal expendiures (graph 0.c) Why did rouine expendiures rise significanly during he crisis year? Before we answer his quesion, please noe ha due o he daa availabiliy, we can only analyze he cenral governmen budge. However, his is jusifiable since he share of cenral governmen in he oal consolidaed governmen revenue and expendiure is considerably high (as explained in par II). The mos imporan facor ha caused rouine expendiures o rise significanly during he crisis year was he exchange rae. Local currency depreciaion caused he paymen for foreign deb (principal and is ineres) and subsidy expendiures o increase significanly. Subsidy expendiures rose considerably because of increases in fuel subsidy expendiures 6. Therefore, i is no surprising ha alhough here was a huge increase in governmen expendiures during he crisis years, his increase did no have any impac on oupu, since i mainly ook place in he secor ha did no increase he producive capabiliy. 6 Fuel price in Indonesia is no based on marke mechanism. Governmen se he price of fuel and he amoun of subsidy depends on difference beween he selling price and producion cos. The selling price is in domesic currency, while 90 percen of producion is calculaed in foreign currency. 7

21 Graph 0.a: Composiion of Cenral Governmen Expendiure 994/ / % of Toal Expendiure Rouine Expendiure Capial Expendiure /95 95/96 96/97 97/98 98/99 99/00 Fiscal Year Graph 0.b: Composiion of Provincial Governmen Expendiure 994/ / % of Toal Expendiure Rouine Expendiure Capial Expendiure /95 95/96 96/97 97/98 98/99 99/00 Fiscal Year 8

22 Graph : Cenral Governmen Rouine Evpendiure by Type 994/ / % of Toal Expendiure Salary Maerial Local Subsidy Deb and Ineres Ohers /95 95/96 96/97 97/98 98/99 99/00 Fiscal Year Graph 0.c: Composiion of Local Governmen Expendiure 994/ / % of Toal Expendiure Rouine Expendiure Capial Expendiure /95 95/96 96/97 97/98 98/99 99/00 Fiscal year 9

23 VI. Conclusions and Policy Implicaion VI.a. Conclusions and Policy Implicaion From he above resul, we can conclude ha a shock in governmen expendiures has a significan effec on changes in oupu, and ha he effec is asymmeric (posiive shock has a larger impac han negaive shock). If governmen expendiure is lower han he seady sae value (in oher words, a negaive shock), i does no reduce oupu significanly. Bu if governmen expendiure is above he seady sae value (a posiive shock) i would generae a higher oupu, wih a muliplier of This is no surprising, since we only considered one side of governmen operaions 7. One possible explanaion as o why oupu did no respond o he increase in governmen expendiures is he change in he composiion of spending. More han one half of oal governmen spending was allocaed o rouine expendiures, such as fuel subsidy expendiures and paymens o foreign deb and ineres, and hese did no increase he producive capaciy. The findings above give us wo policy implicaions. Firs, he governmen mus reallocae is spending, i.e. allocae more o developmen expendiures which will increase he producive capaciy. Second, since Indonesia currenly has a huge deb burden, i is no necessary for he governmen o increase is expendiures, since he effec on oupu reducion is no saisically significan. Moreover, he increase in governmen expendiures is likely o be financed by new deb. Also, here is no guaranee ha he governmen will spend efficienly, since corrupion is sill a big problem in Indonesia. For moneary policy, we conclude ha a money supply shock does no have a saisically significan effec on oupu. This resul suggess ha moneary auhoriy should no use a shock in money supply o boos oupu. In oher words, he moneary auhoriy should focus on is funcion in delivering a low level of inflaion. In he inflaion equaion, i is imporan o sabilize he exchange rae. The effec of he exchange rae is saisically significan in affecing he inflaion rae. The benefi of mainaining a sable exchange rae comes from is effec on domesic fuel prices. Since fuel prices in Indonesia are no based on marke mechanisms and he subsidy depends on he level of exchange raes (since 90 percen of producion cos is in foreign currency), he benefi of mainaining a sable exchange rae comes from wo sources. Firs, i reduces he impac on inflaion, and second, i reduces subsidy expendiures for he cenral governmen. 7 Noe ha if we include revenue in he esimaion, we canno consider he change in ax rae as a shock, since he process ha enables people o prepare for he change in ax rae akes a long ime. 0

24 VI.b Suggesion for Furher Work For furher work we sugges wo poins. Firs, in his paper, in order o find ou he seady sae for oupu, we use he Hodrick Presco echnique. Theoreically, he seady sae value should reflec poenial oupu. In order o esimae poenial oupu, unemploymen daa is required. However, in his paper we canno obain quarerly daa for unemploymen. Therefore, we sugges ha he unemploymen daa be aken ino accoun in order o ge a beer esimaion of poenial oupu (he seady sae value). Second, since fiscal policy also includes governmen revenues, i is beer o use he ne domesic impac approach on he analysis of governmen budge in furher work.

25 Reference: Aksoy, Yunus, Paul De Grauwe and Hans Dewacher (00), Do Asymmeric Maer for European Moneary policy?, European Economic Review, Vol 46 (00), P Ball, Laurence and N Gregory Mankiw (994), Asymmeric Price Adjusmen and Economic Flucuaion, The Economic Journal, Volume 04, Issue 43 (March 994), P 47 6 Chu, Joonsuk and Ronald A Rai (997), Effecs of Unanicipaed Moneary policy on Aggregae Japanese Oupu: The Role of Posiive and Negaive Shocks, The Canadian Journal of Economics, Volume 30, Issue 3 (Augus 997), P 7-74 Cover, James P (99), Asymmeric Effec of Posiive and Negaive Money-Supply Shocks, The Quarerly Journal of Economics, Volume 07, Issue 4 (Nov 99), P 6 8 Demeriades, Panikos O (989), The Relaionship Beween he Level and Variabiliy of Inflaion: Theory and Evidence, Journal of Applied Economerics, Volume 4, Issue 3 (July Sepember 989), P De Long, J Bradford and Lawrence Summer (988), How Macroeconomic Policy Affec Oupu?, Brooking Paper on Economic Aciviy, Volume 988, Issue (988), Garibaldi, Piero (997), The Asymmeric Effec of Moneary policy in Job Creaion and Desrucion, IMF Working Paper Kandil, Magda (000), The Asymmeric Effec of Exchange Rae Flucuaion: Theory and Evidence from Developing Counries, IMF Working Paper Kuran, Timur (983), Asymmeric Price Rigidiy and Inflaionary Bias, The American Economic Review, Volume 73, Issue 3 (June 983), P Tambakis, Demoshenes N (998), Moneary policy wih Convex Philips Curve and Asymmeric Loss, IMF Working Paper

26 3 Appendix I. Esimae Equaions.a EXRATE M GOVN GDP ε α = 3 0.b EXRATE M GOVN GDP ε α = 3 0.a EXRATE M GOVN GDP ε α = ) log( ) log( ) log( ) log( 3 0.b EXRATE M GOVN GDP ε α = ) log( ) log( ) log( ) log( a NEGEXR POSEXR NEGM POSM NEGGOV POSGOV GDP ε α = b NEGEXR POSEXR NEGM POSM NEGGOV POSGOV GDP ε α = a = NEGGOV NEGGOV NEGGOV POSGOV POSGOV POSGOV GDP α NEGM NEGM NEGGOV POSM POSM POSM NEGEXR NEGEXR NEGEXR POSEXR POSEXR POSEXR ε b = NEGGOV NEGGOV NEGGOV POSGOV POSGOV POSGOV GDP α NEGM NEGM NEGGOV POSM POSM POSM NEGEXR NEGEXR NEGEXR POSEXR POSEXR POSEXR ε a WOILP EXRATE M GOVN CPI ε α = 5.b WOILP EXRATE M GOVN CPI ε α log log log log log = 6.a WOILP NEGEXR POSEXR NEGM POSM NEGGOV POSGOV CPI ε α = b WOILP NEGEXR POSEXR NEGM POSM NEGGOV POSGOV CPI ε α = log 7.a = NEGGOV NEGGOV NEGGOV POSGOV POSGOV POSGOV CPI α NEGM NEGM NEGGOV POSM POSM POSM NEGEXR NEGEXR NEGEXR POSEXR POSEXR POSEXR WIOILP WOILP WOILP ε b

27 log CPI = α POSGOV POSGOV POSGOV NEGGOV NEGGOV NEGGOV POSM 8POSM 9 POSM 0 NEGGOV NEGM NEGM 3POSEXR 4 POSEXR 5POSEXR 6 NEGEXR 7 NEGEXR 8 NEGEXR 9 WOILP 0WOILP WIOILP ε Definiion: GDP: Gross Domesic Produc (Billion of Rupiah) GOVN: Nominal Governmen Expendiure (Billion of Rupiah) M: Money Supply (Billion of Rupiah) EXRRATE: Exchange Rae Rupiah Agains US Dollar (Rupiah / US Dollar) WOILP: Weighed Oil Price (Rupiah) CPI: Consumer Price Index POS GOV: Posiive Shock of Governmen Expendiure (Billion of Rupiah) NEGGOV: Negaive Shock of Governmen Expendiure (Billion of Rupiah) POS M: Posiive Shock of Money Supply (Billion of Rupiah) NEGGOV: Negaive Shock of Money Supply (Billion of Rupiah) POS EXR: Posiive Shock of exchange rae / Depreciaion (Rupiah NEGEXR: Negaive Shock of Exchange rae / Appreciaion (Rupiah) : Difference Operaor 4

28 Table I: Summary of Oupu Esimaion LEVEL OF OUTPUT CHANGE OF OUTPUT GDP = f (GOVN, M, EXRATE) Adj R GOVN**(), M***(), D(GDP) = f(govn, M, Adj R EXRATE***() EXRATE) DW DW.6485 LOG(GDP) = f [LOG(GOVN), LOG (M), LOG(EXRATE)] AIC AIC Adj R LOG(M)***(), Adj R LOG(EXRATE)***() D(LOG(GDP),) = f [D(LOG(GOVN),), DW D(LOG(M),), D(LOG(EXRATE),)] DW D(LOG(EXRATE),)**() GDP = f (POSGOV, NEGGOV, POSM, NEGM, POSEXR, NEGEXR) AIC AIC Adj R POSGOV***(), D(GDP,) = f (POSGOV, NEGGOV, NEGGOV***(), POSM, NEGM, POSEXR, NEGEXR) Adj R DW POSM***(), NEGM***(), DW.6489 AIC.6739 POSEXR**(-), NEGEXR***(-) AIC NEGGOV***() GDP = f [POSGOV, POSGOV(-), POSGOV(-), NEGGOV, NEGGOV(-), NEGGOV(-), POSM, POSM(-), POSM(-), NEGM, NEGM(-), NEGM(-), POSEXR, POSEXR(-), POSEXR(-), NEGEXR, NEGEXR(-), NEGEXR(-)] Adj R POSGOV***(), D(GDP,) = POSGOV(-)***(), f [POSGOV, POSGOV(-), POSGOV(-), Adj R DW POSGOV(-)***(), NEGGOV, NEGGOV(-), NEGGOV(-), NEGGOV***(), POSM, POSM(-), POSM(-), DW.460 AIC NEGGOV(-)**,(), NEGM, NEGM(-), NEGM(-), AIC NEGGOV(-)***(), POSEXR, POSEXR(-), POSEXR(-), POSM***(), NEGEXR, NEGEXR(-), NEGEXR(-)] POSM(-)***(-) D(GDP,) = f [POSGOV, POSGOV(-), POSGOV(-), NEGGOV, NEGGOV(-), NEGGOV(-), POSM, POSM(-), POSM(-), NEGM, NEGM(-), NEGM(-)] Adj R DW AIC POSGOV***(), NEGGOV**(), NEGGOV(-)*(-) Governmen Symmeric POSGOV***() POSGOV(-) **() NEGGOV**() Governmen Asymmeric Expendiure: Expendiure: *** indicae he variable is significan a % ** indicae he variable is significan a 5% * indicae he variable is significan a 0% (/-) indicae he effec on dependen variable 5

29 Table II: Summary of CPI and Inflaion Esimaion CPI CPI = f (GOVN, M, EXRATE, WOILP) INFLATION Adj R GOVN***(), M***(), D(LOG(CPI),) = f[dlog(govn),), Adj R EXRATE***(), WOILP***() D(LOG(M)), DLOG(EXRATE),), DW D(LOG(WOILP),)] DW AIC AIC D(LOG(EXRATE),)***() D(LOG(WOILP),)***() Adj R NEGGOV*(), D(LOG(CPI),) = CPI = f (POSGOV, NEGGOV, POSM***(), f (POSGOV, NEGGOV, Adj R POSM, NEGM, POSEXR, NEGEXR, DW POSEXR**(-), POSM, NEGM, POSEXR, NEGEXR, DW WOILP) AIC NEGEXR***(-) WOILP***() WOILP) AIC CPI = f [POSGOV, POSGOV(-), POSGOV(-), NEGGOV, NEGGOV(-), NEGGOV(-), POSM, POSM(-), POSM(-), NEGM, NEGM(-), NEGM(-), POSEXR, POSEXR(-), POSEXR(-), NEGEXR, NEGEXR(-), NEGEXR(-), WOILP, WOILP(-), WOILP(-)] Adj R NEGGOV***(), D(LOG(CPI),) = NEGGOV(-)**(), f [POSGOV, POSGOV(-), POSGOV(-), Adj R DW POSM***() NEGGOV, NEGGOV(-), NEGGOV(-), POSM(-)**(-) DW.7466 AIC NEGM(-)**() AIC WOILP**() POSM, POSM(-), POSM(-), NEGM, NEGM(-), NEGM(-), POSEXR, POSEXR(-), POSEXR(-), NEGEXR, NEGEXR(-), NEGEXR(-), WOILP, WOILP(-), WOILP(-)] NEGEXR**(-), NEGEXR(-)**(), WOILP***() WOILP(-)***(-) *** indicae he variable is significan a % ** indicae he variable is significan a 5% * indicae he variable is significan a 0% (/-) indicae he effec on dependen variable 6

30 Table III: Oupu Esimaion: Wihou Defining Posiive and Negaive Shock (in Log Form) Dependen Variable: LOG(GDP) Mehod: Leas Squares Dae: 03/5/0 Time: 09:3 Sample: 980: 997: Included observaions: 70 Variable Coefficien Sd. Error -Saisic Prob. C LOG(GOVN) LOG(M) LOG(EXRATE) R-squared Mean dependen var.08 Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood F-saisic 4.65 Durbin-Wason sa Prob(F-saisic) Table IV: Oupu Esimaion: Wihou Defining Posiive and Negaive Shock (in Difference of log form / growh rae) Dependen Variable: D(LOG(GDP),) Mehod: Leas Squares Dae: 03/5/0 Time: 09:4 Sample(adjused): 980: 997: Included observaions: 69 afer adjusing endpoins Variable Coefficien Sd. Error -Saisic Prob. C D(LOG(GOVN),) D(LOG(M),) D(LOG(EXRATE),) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood F-saisic Durbin-Wason sa Prob(F-saisic)

31 Table V: Reacion of Oupu o Posiive and Negaive Shock Wih Lags Dependen Variable: D(GDP,) Mehod: Leas Squares Dae: 03/5/0 Time: 09:3 Sample(adjused): 980:3 997: Included observaions: 68 afer adjusing endpoins Variable Coefficien Sd. Error -Saisic Prob. C POSGOV*** POSGOV(-) POSGOV(-) NEGGOV** NEGGOV(-) NEGGOV(-)* POSM POSM(-) POSM(-) NEGM NEGM(-) NEGM(-) POSEXR POSEXR(-) POSEXR(-) NEGEXR NEGEXR(-) NEGEXR(-) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid.57e08 Schwarz crierion Log likelihood F-saisic Durbin-Wason sa.460 Prob(F-saisic)

32 Asymmeric Tes Wald Tes: Posiive Governmen Expendiure Shock Equaion: Null Hypohesis: C()=0 C(3)=0 C(4)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Negaive Governmen Expendiure Shock Equaion: Null Hypohesis: C(5)=0 C(6)=0 C(7)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Posiive Governmen Expendiure Shock Equaion: Null Hypohesis: C()C(3)C(4)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Negaive Governmen Expendiure Shock Equaion: Null Hypohesis: C(5)C(6)C(7)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Asymmeric Tes of Governmen Expendiure Equaion: Null Hypohesis: C()C(3)C(4)=-(C(5)C(6)C(7)) F-saisic.985 Probabiliy Chi-square.985 Probabiliy

33 Table VI: Reacion of Oupu o Posiive and Negaive Shock Wih Lags (Excluding Exchange Rae Variable) Dependen Variable: D(GDP,) Mehod: Leas Squares Dae: 04/08/0 Time: :07 Sample(adjused): 980:3 997: Included observaions: 68 afer adjusing endpoins Variable Coefficien Sd. Error -Saisic Prob. C POSGOV POSGOV(-) POSGOV(-) NEGGOV NEGGOV(-) NEGGOV(-) POSM POSM(-) POSM(-) NEGM NEGM(-) NEGM(-) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid.74e08 Schwarz crierion Log likelihood F-saisic Durbin-Wason sa.6409 Prob(F-saisic)

34 Asymmeric Tes Wald Tes: Posiive Governmen Expendiure Shock Equaion Null Hypohesis: C()=C(4) C(3)=C(4) F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Negaive Governmen Expendiure Shock Equaion: Null Hypohesis: C(5)=C(7) C(6)=C(7) F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Posiive Governmen Expendiure Shock Equaion: Null Hypohesis: C()C(3)C(4)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Negaive Governmen Expendiure Shock Equaion: Null Hypohesis: C(5)C(6)C(7)=0 F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Asymmeric Tes For Governmen Expendiure Equaion: EQ0 Null Hypohesis: C()C(3)C(4)=-(C(5)C(6)C(7)) F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Posiive Money Supply Shock Equaion: Null Hypohesis: C(8)=C(0) C(9)=C(0) F-saisic Probabiliy Chi-square Probabiliy Wald Tes: Negaive Money Supply Shock Equaion: Null Hypohesis: C()=C(3) C()=C(3) F-saisic Probabiliy Chi-square Probabiliy

35 Table VII: Inflaion Rae Esimaion: Wihou Defining Posiive and Negaive Shock (Log Form) Dependen Variable: D(LOG(CPI),) Mehod: Leas Squares Dae: 03/5/0 Time: 09:4 Sample(adjused): 980: 997: Included observaions: 69 afer adjusing endpoins Variable Coefficien Sd. Error -Saisic Prob. C D(LOG(GOVN),) D(LOG(M),) D(LOG(EXRATE),) D(LOG(WOILP),) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood.7075 F-saisic Durbin-Wason sa Prob(F-saisic)