APPLICATION OF DYNAMIC STOCHASTIC GENERAL EQUILIBRIUM MODELS TO THE CASE OF THE SERBIAN ECONOMY***

Transcription

1 ECONOMIC ANNALS, Volume LIX, No. 201 / April June 2014 UDC: 3.33 ISSN: DOI: /EKA U Branko Urošević* Nikola Grga** APPLICATION OF DYNAMIC STOCHASTIC GENERAL EQUILIBRIUM MODELS TO THE CASE OF THE SERBIAN ECONOMY*** ABSTRACT: This paper proposes a dynamic sochasic general equilibrium (DSGE) model for he Serbian economy. I is a modificaion of he exising models of Goodhar, Osorio and Tsomocos (2009) and Marinez and Tsomocos (2012). The model inroduces imporan feaures of he Serbian economy, financial dollarizaion and foreign ownership of he banking sysem, while reaining he mos imporan elemen of he reference models, financial fricion. To solve he model we use Dynare, a specialized Malab program for solving DSGE models. The model is subjec o hree differen shocks: moneary, produciviy, and regulaory, and he resuls are presened in he form of impulse response funcions. I is concluded ha he proposed plaform has good characerisics, bu is complee applicaion o he case of he Serbian economy requires furher research. KEY WORDS: DSGE model, financial fricion, financial dollarizaion, foreign ownership in banking secor JEL CLASSIFICATION: E44, E47, E51, D51 * Faculy of Economics, Universiy of Belgrade and Naional Bank of Serbia, urosevic@ekof.bg.ac.rs ** Faculy of Economics, Universiy of Belgrade, Serbia, grga986@yahoo.com *** Branko Urošević graefully acknowledges he financial suppor from he Minisry of Educaion, Science and Technology of he Republic of Serbia, Gran No

2 Economic Annals, Volume LIX, No. 201 / April June INTRODUCTION The recen economic crisis has indicaed he inabiliy of DSGE models o adequaely address curren economic issues. Cerain auhors (Quadrini 2011, Marinez and Tsomocos 2012) aribue he poor performance of he benchmark DSGE models o he absence of financial fricion in heir seings. Brunnermeier, Eisenbach, and Sannikov (2012, p. 1) have emphasized he crucial role of financial fricion in undersanding business cycle flucuaions. Tovar (2009, p. 6) explains ha exclusion of he financial markes or financial fricion prevens benchmark DSGE models from analyzing financial vulnerabiliy, illiquidiy, and procyclicaliy, and limis heir use for sress esing in financial sabiliy exercises. Alhough benchmark DSGE models have generally no included financial markes and financial fricion, here is a long radiion of dealing wih hese elemens in he macroeconomic heory. One of he earlies works in he field was by Bernanke and Geler (1989). Today, he leading examples of models wih financial fricion are Bernanke, Geler, and Gilchris (1999) and Kiyoaki and Moore (2012). Here asymmeric informaion in he form of enrepreneur moral hazard and cosly sae verificaion resuls in financial fricion. Alernaively, Goodhar, Tsomocos, and Osorio (2009) and Tsomocos and Marinez (2012) inroduce financial fricion hrough he exisence of money, defaul, and agen heerogeneiy. Since hese feaures are able o capure he firs-order effecs of financial fricion in a parsimonious way, we decided o follow he concep of Tsomocos in he paper. The aim of his paper is o propose a dynamic sochasic general equilibrium (DSGE) model for he Serbian economy. The focus will be on he inroducion of a DSGE model for a small, financially open, dollarized economy. The model proposed in he paper is based on he work of Goodhar e al (2009) and Tsomocos and Marinez (2012). These models are micro-founded models for a closed economy based on a real business cycle paradigm wih financial fricions and hey follow he framework developed in Dubey and Geanaklopos (2003) and Tsomocos (2003). The models of Goodhar e al (2009) and Tsomocos e al (2012) comprise he liquidiy issues and endogenous defaul, wo feaures presen in he Serbian economy. However, due o he specificiy of he Serbian economy i is necessary o exend and modify hese models. Given ha 36

3 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA modelling a small, financially open, dollarized economy in a general equilibrium environmen is a very complex issue, his paper should be regarded as a firs sep in researching he phenomenon. Our proposed model exends he aforemenioned models in he following ways. Firs, i inroduces financial dollarizaion ino he seing of he DSGE model. Alhough dollarizaion is presened in Casillo, Monoro, and Tuesa (2006), heir model is based on he New Keynsian paradigm, here is no endogenous defaul, and i does no consider financial dollarizaion. Second, foreign ownership in he banking secor is inroduced. Furhermore, in order o model financial dollarizaion i is necessary o inroduce he exchange rae ino he model. The exchange rae in he model is assumed o depend only on capial flow wih foreign counries. In his way he proposed model exends he work of Tsomocos and Goodhar o he field of financially open economies. Dollarizaion refers o he use of foreign currency for all or some funcions of domesic currency in he domesic economy. The focus will be only on parial financial dollarizaion, 1 where foreign currency becomes he preferred denominaion for savings and loans. The lieraure (Levy Yeyai 2006, Alvarez- Plaa and Garcia-Herrero 2008, Aleksić, Đurđević, Palić and Tasić 2008) concenraes on he negaive side of financial dollarizaion., A high level of financial dollarizaion has a negaive effec on he moneary auhoriy conducing moneary policy and on debors abiliy o mee heir obligaions when he deb is in foreign currency and when here is exchange rae depreciaion. In his paper he focus will be predominanly on hese effecs, while he mechanism of how dollarizaion affecs economic life will be explained furher in he par of he paper ha deals wih moivaion. Anoher imporan feaure ha he previously menioned models lack is ha foreign banking groups own he majoriy of he Serbian banking secor. As par of inernaional banking groups Serbian banks have access o funds wih condiions hey would no have oherwise. A similar case is described in 1 Dollarizaion can be complee, when foreign currency is a legal ender, and parial, when foreign currency is used along wih domesic currency. Parial dollarizaion can be: ransacion dollarizaion, price dollarizaion, or financial dollarizaion. For deails see Casillo, Monoro and Tuesa (2006) 37

4 Economic Annals, Volume LIX, No. 201 / April June 2014 Tsenova (2012). According o Draguinović (2008, p. 18) he exisence of an alernaive source of finance makes banks less dependen on he liquidiy provided by he Cenral Bank. Therefore foreign ownership migh addiionally affec ransmission of moneary policy and is efficiency. Finally, since he Serbian economy is small compared o he size of he European Union economy, changes in Serbia in demand for funds from abroad would have a negligible effec on he financial markes of he European Union. The paper is organized as follows. The firs secion explains he moivaion of our paper. The model is defined in he second secion. The hird par conains a numerical example of he model, and he las secion concludes. 2. MOTIVATION DSGE models are a very useful ool for economic policy analysis. However, as ye no DSGE model has been developed o analyse Serbian economic policy. This paper is moivaed by he desire o sar a discussion on he opic, and o conribue o he developmen of a DSGE model for he Serbian economy and he developmen of DSGE models for developing counries wih dollarizaion issues in general. The following saisics show he exen of financial dollarizaion and foreign ownership in he Serbian banking secor, he consequences of which deermined he inroducion of hese feaures in our proposed model. On he Serbian banking secor s balance shee, loans consiue almos 75% of oal asses and securiies jus 9% 2. On he oher hand, deposis make up almos 70% of liabiliies 3. Thus loans and deposis play a dominan role in he balance shees of Serbian banks. In his paper we will only focus on he currency srucure of he loans and deposis o show he exen of he financial dollarizaion of he Serbian economy. The srucure of loans and deposis in he Serbian economy in erms of denominaion is presened in Figure 1 and Figure 2: 2 Naional Bank of Serbia (2012) BANKING SECTOR IN SERBIA: Third Quarer Repor 2012,page 7. 3 Naional Bank of Serbia (2012) BANKING SECTOR IN SERBIA: Third Quarer Repor 2012, pages

5 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Figure 1: Currency srucure of loans (in billions of RSD) Figure 2: Currency srucure of deposis (in billions of RSD) Source: Naional Bank of Serbia (2012), pp 9 and 21 Euro-denominaed loans and deposis srongly dominae loans and deposis in dinars (as well as oher foreign currencies). Furhermore, he currency srucure of deposis is harmonized wih he currency srucure of loans. Neverheless, banks are no proeced agains he negaive impac of unexpeced changes in exchange raes. Exchange rae risk is ransformed ino defaul risk: when debors revenue is in domesic currency and heir credi obligaions are in foreign currency, a change in he exchange rae will direcly affec he amoun of credi obligaion. If he exchange rae depreciaes he debor will need more domesic currency o honour he deb, which increases in erms of domesic currency. Given ha revenues say relaively consan, a change in he exchange rae affecs he debor s abiliy o honour he credi obligaion in is enirey. This phenomenon is described, e..g. in Levy Yeyao (2006) and Božović, Urošević and Živković (2009). When such a large proporion of asses and liabiliies is denominaed in a foreign currency i affecs he abiliy of he Cenral Bank o conduc moneary policy. In a regime of inflaion argeing, he ineres rae is supposed o be he mos imporan channel for ransmission of moneary policy. The ineres rae channel implies ha a change in he key rae is ransmied o oher raes, which, under cerain condiions, will laer affec aggregae demand. However, a change in he key domesic rae will only affec he raes on loans in domesic currency. However, mos of he loans are denominaed in or indexed o a foreign 39

6 Economic Annals, Volume LIX, No. 201 / April June 2014 currency, primarily he euro, and depend on he inernaional ineres rae, mainly EURIBOR, while domesic key ineres raes do no affec hem. Therefore he efficiency of he ineres rae channel in condiions of high financial dollarizaion is significanly reduced. This saemen is in accordance wih he findings of Aleksić, Đurđević, Palić and Tasić (2008, pp. 28, 29), which argue ha he effeciveness of his channel depends on he level of dollarizaion, and a he curren level of dollarizaion is effeciveness is shaded. The ownership srucure of Serbian banks is as follows: foreign-owned banks accoun for 74% of oal asses, 74% of oal capial, and 71% of employmen in he banking secor, and pos profis of RSD 17.5bn 4. The majoriy of he foreign owners of banks in he Serbian economy are from Eurozone counries 5. This fac migh have encouraged dollarizaion (euroizaion) in Serbia. According o Draguinović (2008), foreign ownership means ha he banking secor is able o ge foreign capial easily and is less sensiive o liquidiy provided by he Cenral Bank. Therefore foreign ownership affecs moneary policy and is efficiency. 3. THE MODEL The model consiss of one household, wo yeoman farmers, a commercial and a Cenral Bank. Yeoman farmer is a generic erm in macroeconomics for he producer who parially consumes he goods he produces. The cenral place in he model belongs o he commercial bank, which acs as an inermediary beween secors wih a defici and secors wih a surplus of funds. The household is he labour provider and acs as a ne deposior in he financial marke, while he yeoman farmers provide consumer goods and ac as ne borrowers. The yeoman farmers differ in heir abiliy o access differen financial markes. The Cenral Bank, on he oher hand, is a sraegic dummy (i.e., is behaviour is no condiional on he sae of he sysem), which parially provides liquidiy o he sysem. In he model he commercial bank has a paren company, which is an alernaive provider of liquidiy o he commercial bank, bu is behaviour is no he objec of ineres in his paper. I is only imporan 4 Source: Naional Bank of Serbia, BANKING SECTOR IN SERBIA: Third Quarer Repor 2012,, page 3. 5 Source: Naional Bank of Serbia, BANKING SECTOR IN SERBIA: Third Quarer Repor 2012,, page 3. 40

7 ha he paren company provides he commercial bank wih he funds i requires a an ineres rae deermined ouside of he model. Finally, he agens in he sysem are inerconneced by a sysem of markes. The srucure of he model is presened in Figure 3. Figure 3: Srucure of he Model APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Source: Auhor's illusraion Financial dollarizaion is inroduced hough parial deposi and loan dollarizaion. In oher words, a par of deposis and loans are in foreign currency. For he sake of simpliciy, insead of having deposis and loans in foreign currency, foreign-currency-indexed loans and deposis are inroduced (hereinafer euro-indexed deposis and credis). Euro-indexed credis and deposis are regular credis and deposis wih an exchange rae clause, i.e., he amoun of ousanding deb in euro-indexed deposi and credi markes changes 41

8 Economic Annals, Volume LIX, No. 201 / April June 2014 wih changes in he exchange rae. In his way he same effec is achieved, bu more simply, each ime money relaing o euro-indexed loans and deposis is changed. The exchange rae clause makes agens sensiive o changes in he exchange rae, while he quaniy of euro-indexed loans and deposis means ha he role of he exchange rae is very imporan in decision-making. The economy conains four differen ypes of marke: he financial marke, he commodiy marke, he labour marke, and he exchange rae marke. The financial markes are he inerbank marke (beween he Cenral Bank and he commercial bank), he domesic currency financial marke (beween he commercial bank and yeoman farmer γ), he euro-indexed deposi marke (beween he commercial bank and household α), and he euro-indexed loan marke (beween he commercial bank and yeoman farmer δ). Furhermore, he commodiy marke and he labour marke boh connec household α wih yeoman farmer γ and yeoman farmer δ. In he labour marke, he commodiy marke, he inerbank marke, he foreign exchange marke, and he domesic financial marke, he price or ineres rae is deermined by equaing respecive supply and demand, in he same way as in Goodhar e al (2009) and Marinez e al (2012). By conras, ineres raes in he euro-indexed loan and deposi markes are deermined ouside of he model. In he case of he Serbian economy, due o he currency srucure of deposis and loans (see Figures 1 and 2), inernaional ineres raes such as Libor and Euribor play a dominan role. To implemen his feaure of he Serbian economy in he model he following assumpion has been made: he commercial bank akes as many deposis as household α is willing o inves in he commercial bank a an ineres rae level ha is deermined ouside of he model. Furhermore, yeoman farmer δ acceps as many funds as he commercial bank is willing o inves in he euro-indexed loan marke a he rae deermined ouside of he model. Thus, ineres raes on euroindexed loans and deposis are equal o he sum of he ineres raes in he inernaional financial marke, which is represened in he model by he ineres rae on he loans from he paren company and premiums, which are all assumed o be consan in he model. 42

9 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA The agens also differ according o which financial marke hey have access o. Yeoman farmer δ can only ge a euro-indexed loan, while yeoman farmer γ can only ge a loan in domesic currency. Furhermore, household α can only inves money in euro-indexed deposis. The division of yeoman farmers is condiioned by he need o have asses in differen currencies in one model. On he oher hand, from he currency srucure of deposis (see Figure 2) i seems ha he limiaion on household invesmen does no diverge much from realiy bu makes he model more parsimonious. Financial openness is achieved because he commercial bank can borrow from abroad. Foreign ownership provides Serbian banks ha belong o inernaional banking groups wih an opion o borrow from he paren company. We rea Serbian economy as a small open economy. Thus, he ineres rae on loans from he paren company does no vary wih he amoun of funds demanded by he commercial bank. Because he commercial bank has an alernaive source of funding i is less dependen on he liquidiy ha he Cenral Bank provides. Our model exends he models of Goodhar e al (2009) and Marinez and Tsomocos (2012). We expand hese models in erms of paricipans (a paren company and anoher yeoman farmer are added), markes (foreign exchange rae marke and euro-indexed loan and deposi markes are added), and asses (euro-indexed loans and euro-indexed deposis are added). In regard o Goodhar e al (2009) he model is simplified so ha here is no bank heerogeneiy. The omission of bank heerogeneiy has lile effec on he main findings of he paper while faciliaing boh racking of shock propagaion hroughou he model and explanaion of he resuls. The model differs from Marinez and Tsomocos (2012), as here is producion and deposi of funds while he agens are no endowed wih consumer goods. 3.2 Financial Fricion In he model, financial fricion is represened by money, defaul, marke liquidiy, and agen heerogeneiy. These fricions are he key ingrediens of he models of Marinez and Tsomocos (2012) and Goodhar e al (2009), and as he proposed model is based on heir papers hese fricions are also par of our model. The following paragraphs explain he role of financial fricion in he model. 43

10 Economic Annals, Volume LIX, No. 201 / April June 2014 Defaul arises as an equilibrium phenomenon, since agens can choose he amoun of ousanding deb hey pay. Similar o in Marinez and Tsomocos (2012), he cos of defaul is given by a penaly ha reduces uiliy and which can be regarded as a repuaion sancion as he loan obligaion is no fulfilled in is enirey and which is proporional o he unpaid amoun of loan obligaion. On he oher hand, money is inroduced via cash-in-advance consrain. I is a requiremen ha every consumer or firm mus have sufficien money before hey can buy a good. In he model, commodiies and asses can only be raded for money, while all asses are only delivered in money. Since he yeoman farmers do no have he required money a he ime of he rade, hey have o borrow prior o he purchase. Some of he money already exiss in he sysem due o household deposis in he bank, bu he oher par has o come from ouside. Money eners he sysem in hree ways: hrough open marke operaions, borrowing from abroad, and small governmen subsidies. Open marke operaions (OMO) represen he Cenral Bank lending o he commercial bank via he inerbank marke (Repo Marke). Borrowing from abroad is represened by he commercial bank borrowing from is paren company. Finally, small governmen subsidy is he consan amoun of money ha boh yeoman farmers ge a he beginning of each period o reduce he level of defaul on heir loan obligaions. Marke liquidiy is represened by he proporion of proceedings from sales ha is available o he agens a he momen of purchase. In he model a very high level of illiquidiy is assumed; hence he sales proceedings are available o he agen only a he end of he period and so canno be used o make purchases in he curren period. If he agens do no have oher sources of finance hey will have o borrow from he commercial bank in order o consume. Conrary o Goodhar e al (2009) where low level of liquidiy refers only o he commodiy marke, he low level of liquidiy in his model refers o boh commodiy and labour markes. Alhough he assumpion of a represenaive agen faciliaes he process of solving he model i prevens he model from addressing some relevan issues in economic heory. Wihou agen heerogeneiy here will be no rading beween agens, and consequenly no defaul. Absence of defaul is in sark conras o 44

11 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA realiy and he need in a sae of crisis o model financial sabiliy. In he proposed model here are hree classes of agen: he producers, he consumer, and he inermediary. Since hey differ in goals and endowmen heir decisions and consequenly heir reacion o shocks will also differ. In his way he model is suiable for analysing complex economic phenomena. The exisence of financial fricions, paricularly liquidiy and defaul, allows moneary policy o be non-neural in he shor run. On he oher hand, financial fricions do no preven moneary policy from being neural in he long run. In his way he proposed model is able o overcome he classic dichoomy in he shor run. The model is in accordance wih New Keynesian lieraure, where moneary neuraliy holds only in he long run, albei shor-run neuraliy is obained hrough oher means. In addiion, i is in accordance wih RBC lieraure, bu only in he long run, since money neuraliy in RBC models holds in boh he shor and long run. 3.3 Timing Convenion Every period sars wih a realizaion of he sae of naure,defined by a shock or is absence. Afer he shock has occurred he agens decide on heir acion. In he process of decision-making he agens ake prices as given. Once hey have made heir decision he following chain of evens occurs. Firsly, he Cenral Bank lends money o he commercial bank (i.e., he domesic inerbank mees). Then he commercial bank borrows money from abroad and repays is previous-period loan obligaions o he foreign lender. Simulaneously, he foreign exchange marke mees. Furhermore, he yeoman farmers repay he previous-period loans wih revenue from commodiy sales. Due o liquidiy issues he revenue from commodiy sales and he revenue from he labour marke come o he commodiy seller or labour provider wih one period lag in regards o he period ha he commodiy or labour was purchased. Wih he money obained, he commercial bank can now repay is previousperiod loans o he household and he Cenral Bank. The deb obligaion o he foreign lender has already been honoured, since i was one of he condiions of deermining he exchange rae. Aferwards he household deposis some of he money from he commercial bank and he labour marke ino he bank and sends he res o he commodiy marke. Also, he household sends par of is 45

12 Economic Annals, Volume LIX, No. 201 / April June 2014 labour endowmen o he labour marke. The commercial bank uses money from he Cenral Bank and he household and money borrowed from abroad o lend o he yeoman farmers. The yeoman farmers use money o hire labour (i.e., he labour marke mees) and produce he commodiy. Nex, he yeoman farmers send par of he produced goods o he commodiy marke. Finally, he commodiy marke mees and agens consume he goods. We assume ha he Cenral Bank is willing o make a new loan o he commercial bank before i has repaid is previous-period loan o ha insiuion. However, his assumpion does no affec he decision-making of he commercial bank, hough i does allow he exchange rae o be deermined before oher agens repay heir previous-period loan obligaions. In oher words, he commercial bank does no use money borrowed from he Cenral Bank o repay is previous-period loan obligaion o ha insiuion. On he conrary, hese funds are aken from he Cenral Bank wih he inenion of lending hem o he yeoman farmer and hey will be used for ha purpose, bu he commercial bank emporarily uses par of hem o repay he previous-period loan obligaion o he lender abroad. Once he, yeoman farmers have repaid heir previous-period loan obligaions, he money ha should be used o repay he commercial bank loan obligaion o he foreign lender is used for lending o domesic agens, since he obligaion o he foreign lender has already been honoured. 3.4 Household α Household α is a risk-averse agen ha maximizes he discouned sum of lifeime uiliy wih decision vecor σ α = {b α, w α, d eu } =0 inf. I obains uiliy by consuming goods and by leisure ime. Household α is endowed wih some amoun of ime ha can be spen eiher in leisure or working. Work provides i wih money ha can be used o buy consumer goods. On he oher hand, leisure also has a posiive effec on uiliy. Hence, household α always faces a rade-off beween consumpion and leisure. Furhermore, since household α has access o he financial marke, i can deposi a cerain amoun of money wih he commercial bank. In his way α can smooh is consumpion, because i can carry over a par of oday s consumpion o some fuure period when economic facors migh preven household α from consuming a is desired level. The same applies o he siuaion where economic facors preven household α from 46

13 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA obaining is desired level of consumpion oday: i can lower savings and increase consumpion oday a he expense of fuure periods. Household α differs from he household in Goodhar e al (2009) because i is no he owner of he bank and i invess in euro-indexed deposis. The opimizaion problem α faces is given below: max inf b E0 { U( ) Uh ( w )} Y p 0 (1) subjec o: b d eu p l eu eu 1 w1 d1v (1 r 1 ) f f 1 (2) (i.e., good expendiure + invesmen in euro-indexed deposis a ime = labour income carried forward from period -1 + repaymen of euro-indexed deposi from period -1) where: U - uiliy funcion b α - amoun of money sen by α o purchase consumer goods a ime d eu euro-indexed deposis made by α a ime p Y price of consumer goods a ime p l - wage paid for uni of labour a ime f - exchange rae a ime h α - α's ime endowmen w α - amoun of ime α chooses o work in period (h α - w α ) amoun of ime α chooses o spend in leisure in period (b α / p Y ) consumpion of household α a ime r eu - ineres rae in euro-indexed deposi marke in period v θ deposi repaymen rae by commercial bank in period β - discoun facor 47

14 Economic Annals, Volume LIX, No. 201 / April June Yeoman Farmer γ Yeoman farmer is a generic erm in macroeconomics for he producer who parially consumes he goods he produces. Yeoman farmer γ maximizes he sum of lifeime uiliy wih decision vecor σ γ ={b γ, q γ, v γ, μ γ } =0 inf. He obains uiliy by consuming consumer goods and suffers penalies for defauling on loan obligaions. Yeoman farmer γ has he same discoun facor as household α. Yeoman farmer γ resembles he yeoman farmer in he paper by Goodhar e al (2009), bu hey differ in ha he yeoman farmer in his paper pays a nonpecuniary defaul penaly. To produce consumer goods, yeoman farmer γ uses a producion funcion which depends on echnology and labour. He hires labour on he labour marke from household α. Since γ s uiliy depends on he consumpion of he consumer good, par of he producion he holds for himself while selling he oher par on he consumer goods marke. Furhermore, he cash-in-advance consrain requires ha he agen has sufficien funds prior o making he purchase. In he case of yeoman farmer γ his means he needs o borrow from he commercial bank on he domesic currency loan marke. Laer he repays his deb wih profis accumulaed from he proceeds of he las-period consumer goods sale and a small amoun of governmen aid (subsidy). Parial or complee defaul on his loan is an opion, alhough γ suffers a defaul penaly proporional o he amoun of he loan he did no honour. Hence, he defaul penaly is non-pecuniary and i lowers his uiliy. The opimizaion problem ha yeoman farmer γ faces is given below: max inf E { UY ( q) (1 v) } (3) subjec o: b Y A ( ); (4) l p (i.e., amoun of he consumer good produced in period = oal produciviy facor * he amoun of labour γ used o produce he consumer good in period ) 48

15 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Y 1 m p 1q 1 v ; (5) (i.e., loan repaymen o he domesic financial marke a ime = subsidy + sales revenue carried forward from period -1) b 1 rsd r (i.e., payroll paymen a ime = borrowing from domesic financial marke a ime ) (6) Furhermore, muual echnological facor A and γ s defaul penaly are represened by he AR(1) process because hey will be he subjec of he exogenous shock in he simulaion secion. A A ln( A ) 0.5ln( A) 0.5ln( A 1 ) e ; e N(0, ) (7) ln( ) 0.5ln( ) 0.5ln( 1 ) e ; e N(0, ) (8) where: U - uiliy funcion γ Y - producion of consumer goods by γ in period γ q - amoun of goods offered for sale by γ in period γ (Y - q γ ) γ s consumpion of consumer goods in period Y p price of consumer goods a ime γ μ - deb acquired by γ on he domesic financial marke in period γ v - γ 's repaymen rae on his ousanding loan in period r rsd - ineres rae on loans from domesic financial marke exended in period γ b - amoun of money sen by γ o hire labour a ime p l - wage paid for uni of labour a ime l (b γ / p ) quaniy of labour ha γ hires A oal produciviy facor in period A - oal produciviy facor parameer m γ - governmen subsidy f - exchange rae a ime γ τ r - defaul penaly on ousanding deb in period - γ s defaul penaly parameer 2 A 2 49

16 Economic Annals, Volume LIX, No. 201 / April June Yeoman Farmer δ Yeoman farmer δ differs from yeoman farmer γ in wo major characerisics. Firsly, yeoman farmer δ only has access o he euro-indexed loan marke, while yeoman farmer γ only has access o he domesic currency financial marke. Secondly, yeoman farmer δ is modelled o be willing o accep as much money as he commercial bank is willing o inves in he euro-indexed loan marke. In oher words, he yeoman farmer is assumed o have perfecly elasic demand for funds on he euro-indexed loan marke a he given rae r B,eu. Therefore, he maximizes he sum of his lifeime uiliy wih respecive decision vecor σ δ = {q δ, v δ } =0 inf. max f E { UY ( q) (1 v) d (1 r ) } inf eu B, eu f 1 (9) subjec o: b Y A ( ) (10) l p (i.e., δ s producion a ime = oal produciviy facor * amoun of he labour bough on he labour marke a ime ) f v d (11) eu B, eu Y 1 ( 1 r 1 ) m p 1q 1 f 1 (i.e., loan repaymen o euro-indexed loan marke a ime = subsidy + sales revenue carried forward from period -1) b eu d ; (12) (i.e., meeing payroll a ime = borrowing from euro-indexed loan marke a ime ) where: U - uiliy funcion Y δ - producion of consumer goods by δ in period q δ - amoun of consumer goods offered for sale by δ in period v δ - δ's repaymen rae on his ousanding deb in period 50

17 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA p Y price of consumer goods a ime μ δ - deb acquired by δ on he euro-indexed loan marke in period r B,eu - ineres rae in euro-indexed loan marke in period b δ - amoun of money sen by δ o hire labour a ime A oal produciviy facor in period τ δ - δ s defaul penaly on ousanding deb m δ governmen subsidy f - exchange rae a ime p l - wage paid for uni of labour a ime 3.7 Commercial Bank θ The commercial bank acs as an inermediary ha connecs secors wih surplus and secors wih defici in funds. I akes deposis from household α and loans from he Cenral Bank and is paren company. On he oher hand, i exends credi in he domesic financial marke and euro-indexed loan marke. Deposis from euro-indexed loans are only used for euro-indexed loans, while loans from he Cenral Bank and paren company are used for loans in he domesic financial marke. The commercial bank is hedged agains a change in he exchange rae. Primarily, he oal amoun of he funds he commercial bank invess in he euro-indexed loan marke is equal o he amoun ha i gahers in he euroindexed loan marke. This loan and deposi srucure does no differ much from he realiy, presened in Figures 1 and 2. In his way he commercial bank ransfers he exchange rae risk o he oher paricipans in he model. On he oher hand, he loans he commercial bank ges from is paren company are modelled no o vary wih he change in exchange rae. Hence, he commercial bank is hedged agains exchange rae risk. However, he fac ha he commercial bank ransferred he exchange rae risk o he oher paricipans does no mean i has solved he problems relaed o dollarizaion. Given ha he agen s coss are affeced by he change in he exchange rae while revenues are no, he risk of exchange rae change is convered o defaul risk. In he model liquidiy issues cause he phenomenon menion above, albei ha in realiy here are many oher reasons ha can have same effec. 51

18 Economic Annals, Volume LIX, No. 201 / April June 2014 The commercial bank maximizes he ne presen value of he flows of he expeced profi wih decision vecor σ θ ={Π θ, l θ, μ IB,μ IB,eu,v θ } =0 inf. I suffers a non-pecuniary repuaion penaly for defauling on is deb obligaions proporional o he size of he non-repaid amoun of he obligaion. Furhermore, i is assumed ha he commercial bank can defaul on he loan from he Cenral Bank and deposi from household α, bu canno defaul on he loan from is paren company. Besides he defaul penaly, he commercial bank suffers a penaly for borrowing money from is paren company. Loans from he paren company have he saus of an addiional source of funding ha he commercial bank uses when i canno gaher enough funds from oher sources, primarily households. In his way, anyime he commercial bank borrows from is paren company i sends a negaive message o he public ha he commercial bank is no able o arac he funds i needs, or, even more general, ha agens do no sufficienly rus he banking sysem. Therefore, he raio of he funds borrowed from he paren company o he funds gahered from households can be an indicaor of he power he commercial bank has in he economy. The decreased power of he commercial bank has a negaive effec on is image: in he same way as defaul, a bad image negaively affecs a bank s repuaion. A lower repuaion should lower he uiliy funcion of he commercial bank, because he commercial bank s uiliy funcion, loosely speaking, depends on he profi and repuaion of he bank. The bank differs from he banks in Goodhar e al (2009) and Marinez and Tsomocos (2012) since i has access o funds abroad bu suffers penalies by borrowing abroad, and i gives differen ypes of loans o differen agens. The opimizaion problem ha he commercial bank faces is given below: max f E { U( ) (1 v )[ d (1 r )] (1 ( ) )} IB, eu inf B, IB, eu,1 IB f eu eu,2 r eu 0 f 1 d (13) subjec o: f f v (14) rsd eu B, eu IB eu eu IB, eu ( 1 r 1 ) l 1 v d 1(1 r 1 ) v [ 1 d 1(1 r 1 )] f 1 1 f 1 f 1 52

19 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA (i.e., bank s profi = revenue from domesic financial and euro-indexed credi marke for loans made a ime -1 repaymens on he loans and deposis aken a ime -1) l IB (1 r IB f ) (1 r IB, eu IB, eu ) (15) (i.e., exensions o he domesic financial marke a ime = loans from he inerbank marke and from he paren company a ime ) where: U - uiliy funcion μ IB - deb ha commercial bank acquires in he inerbank marke in period μ IB,eu - commercial bank s deb o paren company (in foreign currency) in period d eu - euro-indexed deposis made by α a ime r IB,eu ineres rae on he loans from paren company a ime r IB inerbank ineres rae a ime Π θ commercial bank s profi v γ - γ 's repaymen rae on his ousanding deb in period l θ - loans o domesic financial marke v θ θ s repaymen rae on is ousanding deb in period τ θ,1 repuaion penaly for defauling on ousanding deb τ θ,2 repuaion penaly for borrowing from he paren company β B discoun facor for commercial bank 3.8 Inerbank Marke The inerbank marke clears when he amoun of money he commercial bank promises o repay in he nex period is exchanged for he credi exension he Cenral Bank offers in he curren period. IB IB 1 r (16) M 53

20 Economic Annals, Volume LIX, No. 201 / April June Domesic Financial Marke The domesic financial marke clears when he amoun of money yeoman farmer γ promises o repay in he nex period is exchanged for he credi exension in domesic currency ha he commercial bank offers in he curren period. rsd 1 r (17) l 3.10 Consumer Goods Marke In every period, he consumer goods marke clears when he amoun of money household α offered for consumer goods is exchanged for he quaniy of consumer goods yeoman farmers γ and δ offered for sale. p Y b (18) q q 3.11 Labour Marke In every period, he labour marke clears when labour expendiures by yeoman farmers γ and δ are exchanged for he amoun of hours ha household α is willing o work. l b w b p (19) 3.12 Exchange Rae Marke In every period, he foreign exchange rae equaes supply and demand for foreign exchange. The amoun of credi ha he commercial bank borrows from is paren company and he consan amoun of foreign exchange ha he paren company has o pay o he auhoriy (M eu ) form a supply of foreign exchange. On he oher hand, repaymen of he previous-period deb obligaion o he paren company by he commercial bank and he previous-period profis of he commercial bank form demand for foreign exchange. Here, i is assumed ha repaymen of he commercial bank s previous-period deb o he paren 54

21 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA company happens simulaneously wih he new loan he commercial bank ges from is paren company. f f IBeu IBeu eu M IBeu r (1 ) (20) Euro-indexed Deposi and Loan Marke The euro-indexed loan and deposi marke differs from he oher markes because ineres raes are no deermined hrough he ineracion of supply and demand bu are given exogenously. Therefore, so ha boh markes clear (clearing of all he markes is one of he condiions of equilibrium in DSGE models), i is necessary o assume ha he commercial bank is willing o accep as much money as household α is willing o deposi a a given level of he euroindexed deposi ineres rae. The commercial bank is willing o do so, since yeoman farmer δ is modelled o be willing o accep as many funds as he commercial bank offers on ha marke Moneary Policy The moneary policy follows Goodhar e al (2009) and Marinez and Tsomocos (2012). In every period he Cenral Bank injecs a cerain amoun of money ino he sysem. I does so hrough open marke operaions (OMO). When i wans o decrease he amoun of money in he sysem and increase he ineres rae on loans in domesic currency, he Cenral Bank decreases he level of OMOs. M CB _ M (23) CB CB CB CB CB 2 ) 0.5 ln( ) 0.5 ln( ) e ; e N(0, CB) ln( 1 (24) where: M amoun of money injeced ino he sysem by he Cenral Bank a ime M - amoun of he money injeced ino he sysem by he Cenral Bank in seady sae CB η - he moneary operaions gross growh a ime CB - he moneary operaions gross growh parameer 55

22 Economic Annals, Volume LIX, No. 201 / April June Equilibrium According o he definiion of DSGE models, a sysem is always in equilibrium. Equilibrium can come in wo differen modaliies, long-run equilibrium and shor-run equilibrium. Long-run equilibrium is a saic equilibrium and is referred o as a seady sae. Shor-run equilibrium is achieved if: 1. agens opimize heir uiliy funcions given heir budge se; 2. all markes clear; 3. expecaions are raional. The sysem is always in shor-run equilibrium. However, in order o have longrun equilibrium as well, wo addiional condiions are required. The sysem is in long-run equilibrium (seady sae) if: 1. agens opimize heir uiliy funcions given heir budge se; 2. all markes clear; 3. expecaions are raional; 4. all variables do no grow; 5. he economy is no subjec o any curren or expeced shock. 4. NUMERICAL EXAMPLE The numerical example aims o give a basic inuiion of how he model works. However, he calibraion ha will allow deeper analysis of he Serbian economy is a ask for furher developmen of he model. The ables below show he parameer values and seady-sae values of endogenous values used in simulaions. 56

23 Table 1: Parameers Table 2: Endogenous variables Parameers Value Endo. var. Value A 2.10 p Y h α 1.00 p l r IB,eu f r RSD r IB , M , A 2.10 m γ η CB 1.00 m δ d eu β 1/ b α β B 1/1.02 η α _ M w α M eu e-004 μ γ r B,eu b γ r eu γ η CB 1.00 γ η Source: Auhor's calculaion q γ v γ τ γ η δ q δ v δ Π θ μ IB μ IB,eu θ η θ η l θ v θ Simulaions APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Source: Auhor's calculaion The purpose of his secion is o describe he change of he endogenous variables due o he se of exogenous shocks. Simulaions are run for hree differen shocks: moneary shock, produciviy shock, and regulaory shock. All 57

24 Economic Annals, Volume LIX, No. 201 / April June 2014 of he simulaions are run for a small shock, i.e., 1% change in he respecive seady-sae value. In all cases, shock unexpecedly his he sysem a beginning of he firs period and provokes he reacion of he agens, who adjus heir behaviour accordingly. In simulaions, all he agens have logarihmic uiliy funcions. The model is simulaed using Dynare sofware Moneary Shock Moneary shock sands for a change in he OMO growh rae ( ) in he iniial period. From equaion (23) i is clear ha a change in he OMO deermines he amoun of liquidiy he Cenral Bank injecs ino he sysem. Thus, he following simulaion shows he reacion of he paricipans in he model o he unexpeced 1% decrease in he amoun of liquidiy he Cenral Bank injecs ino he sysem. Graphs 5.1 o 5.18 summarize he effecs of he moneary shock a he beginning of he firs period. As migh be expeced, a negaive moneary shock increases domesic currency ineres raes. However, he impulse response funcion of he ineres raes on domesic currency loans (r RSD ) differs from ha of he ineres rae on he inerbank marke (r IB ), because he commercial bank has he possibiliy o borrow money from is paren company. In his way he possibiliy of borrowing from abroad negaively affecs he ransmission of moneary policy and he usefulness of he ineres rae channel. Alhough he sysem is modelled in such a way ha a change in domesic moneary policy does no change ineres raes on euro-indexed loans and deposis (r B,eu,r eu ), moneary policy does change he level of defaul in he economy. Since he commercial bank can borrow from abroad if he Cenral Bank runs a resricive moneary policy, a negaive moneary shock implicily leads o an increased supply of foreign currency and decreases he foreign exchange rae. Because yeoman farmer δ s deb obligaion depends on he exchange rae, a decrease in he exchange rae means lower deb obligaion for yeoman farmer δ and consequenly a lower level of defaul. Given ha he negaive moneary shock has almos no effec on he defaul of yeoman farmer CB 6 In he paper, firs-order approximaion is used. For deails, see Julliard (2001) and Adjemian e al. (2011). 58

25 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA γ, he decreased defaul of yeoman farmer δ also affecs he defaul rae of he commercial bank. Negaive moneary shock has a negaive effec on he price of consumer goods and labour. Due o he lower level of funds available o he yeoman farmers for hiring labour, wages decrease and household α offers less unis of labour on he labour marke. Since he commercial bank does no compensae for he negaive moneary shock enirely and household α reduces invesmen in deposis, he yeoman farmers have less funds available for hiring labour. The decrease in he exchange rae reduces he reurn of household α on is deposi invesmen prior o he occurrence of he shock and he amoun of funds i has a is disposal, affecing negaively is deposi invesmens in he curren period. Furhermore, he smaller amoun of labour available on he labour marke reduces overall producion and consumpion. This is proof ha moneary policy has real effecs, since he change in moneary policy leads o a change in agens producion and consumpion. On he oher hand he price of consumer goods also decreases. I is reduced because household α spends less in he consumer marke due o is reduced budge. Afer he iniial period he effec of he shock sars o vanish and he variables slowly reurn o a seady sae. Some of he variables diverge from heir seadysae values for a few periods afer he iniial period, bu afer ha hey sar o converge back o he seady sae. This is because borrowing from he paren company decreases slower han he borrowing from he Cenral Bank recovers. Finally, his simulaion proves he imporance of he exchange rae o he ransmission of moneary policy. This is because mos of he changes in he sysem resul from he change in he exchange rae. On he oher hand, he ineres rae channel has relaively low impac in he given seing. Furhermore, i is proved ha he abiliy of he commercial bank o borrow from abroad prevens he Cenral Bank from conrolling he amoun of liquidiy in he sysem. Defaul also follows he expeced paern. In he end, he dominan effec of defaul on euro-indexed loans on defaul in he whole sysem migh be expeced, bearing in mind he currency srucure in he model. 59

26 Economic Annals, Volume LIX, No. 201 / April June 2014 Graph 5.1 change of p Y due o he change of M Graph change of p l due o he change of M Graph change of r IB due o he change of M Graph change of r RSD due o he change of M Graph change of f due o he change of M Graph change of b α due o he change of M Graph change of d EU due o he change of M Graph change of w due o he change of M Graph change of μ γ due o he change of M Source: Auhor's calculaion 60

27 Graph 5.10 change of b γ due o he change of M Graph change of v γ due o he change of M Graph change of q γ due o he change of M Graph change of v δ due o he change of M Graph change of q δ due o he change of M Graph change of l θ due o he change of M Graph change of v θ due o he change of M Graph change of μ IB,eu due o he change of M Graph change of μ IB due o he change of M APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Source: Auhor's calculaion 4.4 Produciviy Shock Produciviy shock sands for a change in he echnological facor (A ) in he iniial period. Shock is supposed o affec he producion funcions of boh yeoman farmers, because hey do no differ in erms of producion echnology. Thus, he following simulaion shows he reacion of he paricipans in he model o he unexpeced 1% increase of produciviy of echnology in he iniial period. Graphs 6.1 o 6.3 summarize he effecs of he produciviy shock a he beginning of he firs period. 61

28 Economic Annals, Volume LIX, No. 201 / April June 2014 As migh be expeced, increase in produciviy increases overall producion. I also has a posiive effec on he amoun of consumer goods farmers supply o he consumer goods marke. Due o increased supply he price of consumer goods decreases. Finally, he produciviy shock affecs he oher variables such as ineres raes and exchange rae, bu he effec is so small ha i can be ignored. The produciviy shock reveals he imporance of ineres raes and especially exchange rae o he duraion of he effec of he shock. Since i has very lile effec on hese variables he shock affecs he sysem for a very shor period. Graph 6.1 change of p Y due o he change of A Graph change of q γ due o he change of A Graph change of q δ due o he change of A Source: Auhor's calculaion 4.5 Regulaory Shock The following simulaion shows he effec of an unexpeced 1% increase of he defaul penaly o yeoman farmer γ (τ γ ) in he iniial period. Graphs 7.1 o 7.18 summarize he effecs of he regulaory shock a he beginning of he firs period. As migh be expeced, an increase in he defaul penaly enhances he deb repaymen rae of yeoman farmer γ. However, in he iniial period he repaymen rae says a he seady-sae level and increases only afer he iniial period. This is because yeoman farmer γ does no have sraegic space o increase his repaymen rae, due o he decisions he made in he previous period. Only afer he occurrence of he shock can he change his previous decision and decrease he level of defaul. This is he reason yeoman farmer γ s defaul sars o decrease afer he firs period. 62

29 APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Due o he prospecive increase in γ s repaymen rae, he commercial bank is willing o borrow more from he Cenral Bank and paren company. Hence, i provokes an increase in he inerbank marke ineres rae and a decrease in he exchange rae. I also has a negaive effec on he domesic financial marke ineres rae, because he amoun of funds available for lending increases. Similarly o he case of negaive moneary shock, exchange rae decrease drives down wages, he amoun of labour household α sends o he labour marke, and overall producion and consumpion. The reduced budge of household α has a negaive effec on he price of consumer goods. Furhermore, he lower exchange rae increases he repaymen rae of yeoman farmer δ and he commercial bank. Since he amoun of euro-indexed loans is much greaer han he amoun of loans in domesic currency, he repaymen rae of he commercial bank follows he paern of yeoman farmer δ s repaymen rae. Finally, afer he iniial shock he variables converge slowly o a seady sae. However, some of he variables do no sar o converge immediaely afer he firs period. Because he foreign exchange rae increases faser han he amoun of money ha he commercial bank borrows from is paren company decreases, he amoun of money he commercial bank offers on he domesic financial marke increases. This is he reason why some of he variables, e.g., r RSD and b γ, diverge from he seady sae afer he iniial shock, and hen sar o converge. Regulaory shock proves he findings from he moneary shock. The imporance of he exchange rae is shown once again, since he level of producion decreases despie he decrease in he ineres rae in he domesic financial marke. Also, i is shown ha defaul in he whole sysem srongly depends on he level of defaul on he euro-indexed loans. 63

30 Economic Annals, Volume LIX, No. 201 / April June 2014 Graph 7.1 change of p Y due o he change of τ γ Graph change of p l due o he change of τ γ Graph change of r IB due o he change of τ γ Graph change of r RSD due o he change of τ γ Graph change of f due o he change of τ γ Graph change of b α due o he change of τ γ Graph change of d EU due o he change of τ γ Graph change of w due o he change of τ γ Graph change of μ γ due o he change of τ γ Source: Auhor's calculaion 64

31 Graph 7.10 change of b γ due o he change of τ γ Graph change of v γ due o he change of τ γ Graph change of q γ due o he change of τ γ Graph change of v δ due o he change of τ γ Graph change of q δ due o he change of τ γ Graph change of l θ due o he change of τ γ; Graph change of v θ due o he change of τ γ Graph change of μ IB,eu due o he change of τ γ Graph change of μ IB due o he change of τ γ APPLICATION OF DSGE MODELS TO THE CASE OF SERBIA Source: Auhor's calculaion 5. CONCLUSION In his paper he inabiliy of exising DSGE models o depic he Serbian economy is emphasized. This inabiliy is aribued o a lack of specific feaures ha significanly impac he operaion of he Serbian economy, such as financial dollarizaion and foreign ownership of he banks. Here hese feaures are applied, producing a model ha can be applied o he Serbian economy. The proposed model shows ha he ineres rae channel has relaively low impac on economic aciviy, firsly because mos of he asses are denominaed 65