Macroeconomic and Monetary Determinants in Analyzing Recurrence Time for Inflation and Deflation Cycles

Transcription

1 Macroeconomc and Monetary Determnants n Analyzng Recurrence Tme for Inflaton and Deflaton Cycles Do Th Van Trang Department of Fnance, Bankng Academy, Hano, Vetnam tranhabach@yahoo.com Abtracts: Ths study apples the parametrc shared-fralty model to explore the lkelhood of nflaton (deflaton) recurrence tme n twenty-sx countres. Ths emprcal model concentrates on macroeconomc and monetary determnants affectng on the probablty of exstng tme to nflaton (deflaton) perod. Affectng the nflaton exstng tme s the nfluence of seven determnants as stock returns, real exchange rate, ndustral producton ndex, GDP per capta, government expendture, money supply and reserve assets. Whereas deflaton reoccurrence tme s mpacted by stock returns, GDP per capta and reserve assets. By a means of the parametrc shared-fralty model, t creates an extra econometrc tool for researchers and polcymakers to make plans and strateges n response to mplement nflaton polcy. Keywords: Inflaton, Deflaton, The Parametrc Shared-Fralty Model. 1

2 1. Introducton In macroeconomc polces of a country, controlled nflaton at low level s always consdered carefully and become an mportant goal. The stablty n prce s a man factor n determnng the growth rate n economy; therefore fscal polcy and monetary polcy wll be consdered strctly to mantan nflaton at an expected rate. Fredman (1977) mentoned on the changng n nflaton leadng to unstable responses by monetary makers, whch mght nduce unpredctable about the future nflaton. It s very normal n many countres that f nflaton presents hgh and varable rates, the economy wll have to bear a lot of rsk and take a long tme to recovery. Guerrero and Parker (2006) showed the bdrectonal causalty between nflaton and recesson when examne ths relatonshp among 94 countres. The deflaton rate s at hgh level whch caused the decrease n growth rate although t s not usually stmulate recesson. Fsher (1993) ndcated that nflaton decreased economc growth through declnng captal n nvestment, productvty growth and budget defcts. Thus, they concluded that nflaton at low level and small defcts were not need for hgh growth even examnng n long perods. Moreover, Khan and Senhadj (2001) revealed that nflaton n developng countres has been found a threshold of 11% of nflaton rates. If the nflaton rate was hgher than ths threshold, t would have negatve mpact on economc growth. The study of Bck (2010) also confrmed for ths theory by usng a generalzed panel threshold model. Therefore, nflaton controllng goal s always consdered by the central banks when mplementng monetary and fscal polces, especally for developng countres. The queston s rased up how long an economy can be stayed n nflaton or deflaton perod and whch determnants affect the tme of nflaton or deflaton perod. Inflaton and deflaton ssues have attracted the attentons from many researchers n lterature for a long tme. A huge number of studes have analyzed the volatlty of nflaton among many groups of countres all over the world such as G7, emergng countres or ASEAN countres (Bkker and Kennedy, 1999; Choudhry, 2001; Koman and Tmothy, 2010; Narayana and Narayan, 2013). Samuelson and Solow (1960) analyzed the ssues of archvng and mantanng the stable nflaton rate. Ncholas (2004) studed the causal assocatons between nflaton uncertanty and output growth usng panel data from the G7 countres. The result emphaszed on nflaton nfluences on output growth and caused nflaton uncertanty. Other study mentoned the turnng ponts of nflaton can be ndcated through examnng the nfluences of the leadng ndcators n the economy. It s very mportant to control monetary polcy and mantan the nflaton at stable rate when peaks and troughs are dentfed clearly n nflaton cycles (Healther and Sopha, 2001). Furthermore, nflaton s studed n many dfferent countres lke Bangladesh, Span, Inda, Mexco, Turkey and etc, that has been shown n the papers of Akhand (2010), Marka et al. (2008), Narayana and Narayan (2013), Pazarbaşoğlu and Ötker (1997) and Zya and Süleyman (1990), respectvely. A number of studes have been conducted as an mportant step toward lnkng macroeconomcs and monetary varables wth nflaton (deflaton) exstng tme. The correlaton between nflaton (deflaton) perods can be found through the researches of Cagan (1974), Francsco (1999), Harrson (2000), Arze et al. (2004), Celasun (2006), Bll et al. (2008), Marka et al. (2008) and Dura and Bhadur (2009). The mpacts of monetary polces receved the attentons from scholars such as Eugene (1981), Davs and Toma (1995), Gonzalez et al. (1997), Peter (2008), Gann and Gabrel (2013) and Syed and Sajd (2013). The presence of macroeconomcs and monetary varables mght be explaned for the dfferences and may nfluence on the lkelhood of nflaton and deflaton perods. 2

3 Ths research dscusses about the survval model that are derved from bologcal scences and appled n busness effectvely by of Pazarbaşoğlu and Ötker (1997), Gonzalez et al. (1997), Catanach and Perry (2001), Tudela (2004) and Chen (2010). The survval model consders about the ext tmngs and analyzes of death n bologcal statstcs and falure of economcs. Ths knd of model s utlzed popular n bologcal scences, however, stll lmted n busness, especally n fnance. To the best of our knowledge, no formal emprcal lterature has studed yet the nfluence of the process of recurrence of the nflaton (deflaton) n a perod of tme. Ths paper employs parametrc shared-fralty models to nvestgate the nflaton and deflaton n twenty sx countres and examnes the nflaton (deflaton) cycle reoccurrence. To deal wth questons such as how long an economy can be stayed n nflaton or deflaton perod and what knd of determnants affects the tme of nflaton (deflaton) perod, ths study uses sx macroeconomcs varables such as ndustral producton ndex, stock ndex, gross domestc producton (GDP) per capta, real exchange rate, unemployment and government expendture; and fve monetary varables lke money supply, nterest rate, hot money growth rate, reserve assets and M2/reserve rato. The contrbutons of ths paper are as follows. Frst, ths study mplements a new model that known as the parametrc shared-fralty model n ndcatng the determnants that nfluence on the nflaton (deflaton) perod through forty-two years perod of tme across twenty sx countres. The relatonshps between macroeconomc and monetary varables and nflaton (deflaton) perod have been represented. Second, the paper consders the nfluence of macroeconomc and monetary determnants whether have sgnfcant negatve (postve) mpacts on each recurrence tme of nflaton (deflaton) perods. Thrd, ths study examnes whether any dfference between the macroeconomc and monetary determnants that mpacts on the nflaton and deflaton perod and how they shorten or lengthen the reoccurrence of nflaton (deflaton) perod. The paper s structured as follows. The next secton brefly revews the methodology of parametrc shared-fralty model. Secton 3 ntroduces the data and hypothess. Secton 4 presents the emprcal results. Fnally, secton 5 concludes. 2. Parametrc shared-fralty models The parametrc shared-fralty model has been mentoned n the studes of Pazarbaşoğlu and Ötker (1997), Gonzalez et al. (1997), Catanach and Perry (2001) and Tudela (2004). Ths model s utlzed to estmate the survval tmng that used to mplement analyses of death n medcal statstcs and now s appled n economcs. Based on theoretcal econometrc lterature, ths study uses the shared-fralty regresson specfcaton to examne the tmng of nflaton and deflaton n 26 countres. Clayton (1978) and Hougaard (2000) addressed the shared-fralty model whch concentrated on the survval tme of repeated measurement. The survval model s assumed to be correlated for observatons through countres n the same group; the hazard s dentfed as followed: t x, ht x h, (1) where s a fralty that s proposed to have a mean of one and a varance, θ stands for some unobserved observaton-specfc effect. denotes the covarates for the observaton (j = 1,. ) wthn a group 3

4 through dfferent countres ( ) at tme t. Furthermore, f < 1 or > 1 s a decrease or an ncrease n the hazard, afterward the subjects experenced a decreased rsk or an ncreased rsk, respectvely (Cleves et al., 2004). The relatonshp between hazard and survvor functons can be dentfed as: S t x, S t x, (2) where S t x dentfes the survval functon related to a standard parametrc model. Ths study has utlzed the accelerated falure-tme metrc to estmate the transformed hazard functon, whch mght be x specfed as ln t w xx, where e s a postve hazard rate parameter and p 1 known as a postve scale parameter. Note that x addresses the parameters. The applcaton of fve parametrc log-lnear survval dstrbutons: Webull, Exponental, Gompertz, Log-normal and Loglogstcs (Greene, 2002) has denoted n Table 1. Accordng to the parametrc survval dstrbutons as functoned by ( t 0, t ], the lkelhood of the shared-fralty model s denoted as: t t S x c L t t0, h t x. (3) S 0 x The ncorporatng between the fraltes L G 0 c n S S t t j1 0 x x for the th group (G) mght be denoted as below: c h t x g d, (4) where n C c s the number of nflaton and deflaton perods across countres (Guterrez, 2001). j1 The uncondtonal survval functon n ths study has been appled as: t x S t x g d 0 S, (5) where s explaned for the fralty varance that dentfes the degree of heterogenety among the g denotes the probablty observed countres. Accordng to the assumpton of Cleves et al. (2004), densty functon that has been selected wdely for computatonal applcaton wth a gamma dstrbuton. Therefore, the survval functon wth a gamma dstrbuton s descrbed as below: g e 1 1, (6) where s assumed as a gamma functon. 4

5 Table 1: The Dstrbuton of Survval Functon and Hazard Functon Survval Functon Hazard Functon Dstrbuton S t x h t x p Webull exp t pt p1 Exponental exp t Gompertz (( )( )) ( ) Log-Normal log p t 1 pt Log-Logstc p 1 t p1 Note: represents the standard normal cumulatve dstrbuton. 1 t p Shape Monotonc hazard rate Monotonc hazard rate Monotonc hazard rate (Exponental ncrease or decrease) Non-monotonc hazard rate Non-monotonc hazard rate 3. Data Accordng to the developed of Pazarbaşoğlu and Ötker (1997), Gonzalez et al. (1997), Catanach and Perry (2001) and Tudela (2004), ths study focuses on the recurrence tme for nflaton and deflaton cycles n dfferent economes as dependent varables. The avalable quarterly data emanated from the Internatonal Fnancal Statstcs (IFS) database publshed by the Internatonal Monetary Fund (IMF) from January 1970 to December Other data s collected from webste: Tradng Economcs 1 and ndex mund 2. Ths ncludes twenty-sx countres from the Asa, the Amercas, the Australa, Western Europe and Eastern Europe, whch were dstngushed as nflaton and deflaton perod accordng to the volatlty of consumer prce ndex. The ndependent varables encompass the effects of macroeconomcs varables (such as ndustral producton ndex, stock ndex, GDP per capta, real exchange rate, unemployment and government expendture) and monetary varables (such as money supply, nterest rate, hot money growth rate, reserve assets and M2/reserve rato). Ths data s yearly data bass and derved from IFS. These ndependent varables are classfed nto two categores that are dscussed n ths secton. 3.1 Macroeconomc varables: (1) Real exchange rate (RER): RER s derved from the nomnal exchange rate that deflated the rato of domestc and foregn currency levels. Celasun (2006) found that determnants of nflaton played an mportant role n nflaton dynamcs. Arze et al. (2004) explaned through eghty-two countres wth the current floatng exchange-rate era and stated that nomnal exchange rate volatlty can have a postve and sgnfcant effect on the volatlty of nflaton. Joshua et al. (2010) showed the response to real exchange rates was strongest n these countres followng nflaton targetng polces that were relatvely ntensve n exportng basc commodtes. Ths paper expects that the hgher apprecaton n RER, the greater probablty a country can rapdly revve the economy resultng to shorter ext nflaton perod. The hgher apprecaton n RER, the less probablty a country exts from the deflaton perod

6 (2) Stock return (SIR): Cagan (1974) and Choudhry (2001) examned the postve relatonshp between stock returns and nflaton and provded the evdence that stock returns acted as a hedge aganst nflaton n hgh nflaton countres. However, Eugene (1981) and Dura and Bhadur (2009) tested and explaned the negatve relatons between stock return and expected and unexpected nflaton rate durng the post Therefore, the nfluence of stock ndex return on nflaton stll contnues beng as controversal ssues. The ncreasng n stock ndex return can facltate a faster recovery tme of a country from a recesson resultng to shorter ext nflaton tme. Moreover, the more ncrease n stock return, the shorter tme a country can escape from the deflaton perod. (3) Industral producton ndex (IPI): IPI s an economc ndcator that measures the amount of output from the manufacturng, mnng, electrc and gas ndustres. Harrson (2000) mentoned that f the new products were substtuted for old ones, the nflaton would rose. Thus, the paper antcpates that the more contnued ncrease n IPI, the longer lkelhood of survval tme n nflaton perod. The sgnal of IPI ndex coeffcent n deflaton perod s forecasted to have the postve mpact on the deflaton recurrence perod. (4) Gross domestc product per capta (GDP): GDP per capta s measured by the value of all the fnshed goods and servces produced wthn a country whch dvded for number of people n a specfc perod of tme. Bruno and Easterly (1998) and Ncholas (2004) contended that the nflaton and output had causal effects when nvestgated panel data from hgh nflaton countres and G7 countres. Gruben and McLeod (2002) consoldated the postve correlaton between hgh nflaton and economc growth and poverty by lookng nto 112 countres n data samples. In vew of ths, a negatve relatonshp between GDP per capta and nflaton would be antcpated for nflaton perod. However, the hgher n GDP per capta s assumed to lead to the longer recurrence tme n deflaton perod. (5) Unemployment rate (UER): The relatonshp between nflaton and unemployment rate has been studed through Phlp curve that regarded the negatve correlaton because of an accurate and relable gude to future prce nflaton. But Fredman (1977) and Marka et al. (2005, 2008) addressed the tradeoff between long-run nflaton-unemployment and the persstence of nflaton and unemployment s related to the slope of the long-run Phllps curve. Nevertheless, Samuelson and Solow (1960) found that changes n nomnal wages were postvely correlated wth general nflaton, and then there s an nverse relatonshp between unemployment and overall prce nflaton. The Phlp curve that trades off between the nflaton and unemployment s only mantaned n the short run. For the long run, the actual nflaton s adjusted by the expected nflaton and the short-run Phlp curve shfts. As a consequence, the long-run Phlp curve shfts to be vertcal at the natural unemployment rate. Therefore, a hgher lkelhood of survval tme for contagon s drven by the decreasng unemployment rate assocated wth hgh nflaton. The ncreasng n recurrence rate, the greater probablty a naton can stay longer n exstng tme of deflaton. (6) Government expendture (GE): The structural relatonshp between nflaton, money growth, and government expendture exploted by Francsco (1999). If there s a stable between nflaton and money growth rates, they wll have assocated wth each of the two possble government spendng regmes. Ths paper proposes the greater lkelhood of recurrence tme for contagon s drven by ncreasng n the government expendture related to rsng up n nflaton rate. The hypothess also s proposed for the deflaton perod, wth the hgher n government expendture, the shorter an economy stayng n a recurrence tme. 6

7 3.2 Monetary varables: (1) Money supply (MS): The government can decrease the nflaton through controllng money supply. If the central bank pushes a large of money quantty nto the economy, t leads to the prce level wll ncrease. The nflaton wll mantan n a longer tme when the central bank contnung ncreases the supply of money. Zya and Süleyman (1990) found the exstence of nteractons between exchange rate, prce level and money supply. Accordng to Akhand (2010), there s a causal relatonshp between money supply growth and nflaton. Gann and Gabrel (2013) llustrated money supply provdes an mportant early warnng ndcator for rsks to prce stablty. Ths study expects that the larger the quantty of money, the greater probablty a country n a recesson resultng to a longer exst tme n nflaton. Moreover, for deflaton perod, the hgher money supply, the greater probablty a country n a recesson resultng to a longer recurrence tme. (2) Interest rate (IR): Interest rate s a long-term 10 year nterest rate that a central bank charges from the borrowng depostory nsttutons. Based on the study of Peter (2008), f central bank charges nterest rate at hgh rate, t wll ncrease the margnal costs of producton and certanly rse up the nflaton. Syed and Sajd (2013) ndcated the nfluences of the nterest rate polcy n controllng nflaton. Both the cost channel and natonal debt create large challenges to less developed economes. Thus, an ncreasng rato of nterest rate mght facltate a longer recurrence tme of a country from recesson resultng to longer exst tme n nflaton perod. For the deflaton tme, the sgn of the nterest rate coeffcent s assumed has postve mpact on the recurrence tme. (3) Hot money growth rate (HMR): Hot money s amount of money that dentfes the flow of captal transfer from one country to another through a short-term proft on nterest rate dfferences and/or the shfts of antcpated exchange rate. Hot money s equal to change n foregn exchange reserves mnus net exports mnus net foregn drect nvestment. In vew of ths, ths paper predcts that the hgher n hot money growth rate, the lower probablty of a country can rapdly revse the economy resultng to shorter recurrence perod n the nflaton perod. For the deflaton perod, a hgher lkelhood of recurrence tme for hot money growth rate s foreseen to represent the opposte sgnal as n the nflaton perod. (4) Reserve assets (RAs): Reserve assets are known as the currency or other store of value that s prmarly used by countres for ther foregn reserves. Davs and Toma (1995) ndcated that when the nflaton rate ncreased the tax on banks, then they are requred to hold reserves, and causes banks to lower depost rates. Frankel and Rose (1996) dscussed the problem of a country wth less reserve assets and hgh domestc credt growth mght hde the bad sgn of economc growth. Therefore, a hgher lkelhood of recurrence tme for contagon s drven by growng nflaton rate assocated wth the large number of reserve assets. Nonetheless, ths paper expected that the hgher the number of reserve assets, the less probablty a country can revve the economy resultng to shorter recurrence tme n deflaton perod. (5) M2 to reserve rato (M2R): M2R s calculated by M2 dvded by reserve assets. Chen and Chen (2012) have referred the postve mpact of M2R on probablty of ext tme and speculatve attacks when studyng fnancal crses leads to the rapd depleton of reserve assets under the expectatons of central banks. Ths study has assumed that the greater rato of M2 to reserve rato, the hgher probablty a naton mght swft to revve the economy that leads to longer recurrence perod of nflaton tme. Furthermore, the ncreasng n the M2 to reserve rato s forecasted to have postve nfluence on the recurrence tme of 7

8 a naton that stayng n the deflaton recurrence tme. The predcton for the sgns of all varables s summarzed n Table 2 for both nflaton and deflaton recurrence tme. Table 2: Notaton and defntons of varables used Category Varables Notaton Sgn n nflaton perod Sgn n deflaton perod Macroeconomc varables Real exchange rate RER - + Stock ndex return SIR +/- +/- Industral producton ndex IPI + - Gross domestc product per capta GDP - + Unemployment rate UER +/- +/- Government expendture GE - + Monetary varables Money supply MS +/- +/- Interest rate IR + - Hot money growth rate HMR - + Reserve assets RAs + - M2 to reserve rato M2R Emprcal results Table 3 llustrates the results of both the nflaton and deflaton perods that estmated by the parametrc shared-fralty models. Cleves et al. (2008) stated two crtera to evaluate the best fttng model based on the hghest Log-lkelhood value and the lowest Akake Informaton Crteron (AIC). The best fttng dstrbutons for both specfcatons are Gompertz and lognormal dstrbuton among fve parametrc shared-fralty models. Fgure 1 descrbes the ext tme for nflaton and deflaton perod n panel (A) and (B), respectvely. For the nflaton perod, there are seven varables are sgnfcant such as money supply, stock return, ndustral producton ndex, real exchange rate, GDP per capta, government expendture and reserve assets. The stock return s found to have negatve sgnfcant nfluence on the ncreasng of nflaton at 5 percent level. The nfluence of stock return on nflatons stll s controversal attentons from scholars because of ts sgnal. Ths fndng s consstent wth the research of Eugene (1981) and Dura and Bhadur (2009). They found the negatve relatons between real stock returns and nflaton durng the post-1953 perod and explaned by monetary polcy. Choudhry (2001) has exhbted the negatve relaton between stock returns and nflaton n hgh nflaton countres, s nconsstence wth Fsher s theory that argued no relaton between them. L et al. (2010) assessed the relaton between real stock returns and nflaton whch depended on dfferent regmes. Moreover, real exchange rates lkely present a negatve effect on the exstng tme for the perod of nflaton and sgnfcant at 5 percent level. Ths emprcal result conformed to the hypothess. It means that constant ncrease n value of ths varable probably leads to shorter nflaton recurrence tme. Ths result s explaned n many open-economy countres, an apprecaton n exchange rate wll effect lower nflaton snce the prce to mport goods from foregn countres wll not ncrease as rapdly wth the apprecaton of the currency. An apprecaton of the exchange rate may declne the level of output and nflaton as mentoned n the study of John (2001). As we expected n the 8

9 lterature revew, the ndustral producton ndex proves a sgnfcant mpact on nflaton perod at 1 percent level. Ths nfluence on nflaton s postve and n accordance wth the fndng of Harrson (2000). The ncrease n producton level can be nsttutonalzed that leads to the excesses produced. Consequently, there s an ncrease n supply of goods as well as export f well managed and whch may thus nduce a reducton n nflaton. Wth the hgher ndustral producton ndex, the ext tme of nflaton perod wll lkely be longer n controllng nflaton of a country. For developed countres, the ndustral producton ndex s often much hgher than developng countres, thus the perod of nflaton n these countres s lkely longer and more stable than that of developng ones. As ths revew as shown, the contrastng sgns but neglgble values of GDP per capta and government expendture can be related to the study of Bruno and Easterly (1998) n hgh nflaton among thrty-one countres that s further developed by Ncholas (2004) among G7 countres, showng the negatve relatonshps of long-run growth and recurrence the tme of nflaton. Usng panels of annual, 5-year, or even 10-year averages, the evdence ndcated that the lower growth rate, the hgher nflaton. In addton, when the state has budget defct, the government creates new money to meet the expendture demand. It nduces the ncrease n money supply leads to nflaton n these countres as the regard of Francsco (1999). Table 3: Test results: The survval model Varables Notaton Inflaton perod Deflaton perod Real exchange rate RER (0.022)** (0.137) Stock return SIR (0.015)** (0.026)** Industral producton ndex IPI (0.038)** (0.540) GDP per capta GDP (0.004)*** (0.000)*** Unemployment rate UER (0.207) (0.771) Government expendture GE (0.042)** (0.434) Money supply MS (0.008)*** (0.982) Interest rate IR (0.470) (0.628) Hot money growth rate HMR (0.346) (0.754) Reserve assets RAs (0.024)** (0.000)*** M2 to reserve rato M2R (0.477) (0.230) Constant (0.000)*** (0.000)*** Log Pseudo Lkelhood Best fttng model Gompertz Lognormal Note: *, ** and *** are sgnfcance at 10, 5 and 1% levels, respectvely; p-values are n parentheses. Monetary varables have the mpact of sgnfcances on nflaton perod such as money supply and reserve assets. A sgnfcance at 1 percent level result for money supply seems make sure ts negatve mpacts on the exstng tme for nflaton whch s stll debated by researchers. The fndng supports the clams of Francsco (1999) and Akhand (2010) about the exstence of long-run causal relaton between money supply growth and nflaton n dfferent countres. The nflaton can probably be shfted from the low to hgh nflaton depends on the lagged money growth measurement calculated n the real tme. Thus, the money supply growth can represent a warnng sgnal for shftng nflaton, and vce versa. Ths statement s developed by Gann and Gabrel (2013) when fndng the early warnng sgnal of money supply on the consumer prce ndex volatlty. The last varable that performs the sgnfcant estmaton s reserve assets, explaned by Davs and Toma (1995) and Frankel and Rose (1996). Among the sgnfcant 9

10 results, the ndustral producton ndex s presented to be the major factors n affectng the nflaton recurrence tme of twenty-sx countres. Fgure 1: Survval tmes for nflaton and deflaton perod. Panel (A) Panel (B) For the deflaton perod, there are only three varables are sgnfcant at 1 and 5 percent levels such as stock return, GDP per capta and reserve assets. Most of macroeconomc varables are observed to be more nfluent than monetary varables consderng the weght of sgnfcant varables performed. The stock return n ths perod s descrbed wth a negatve sgns and the same sgn wth nflaton perod whch s n lne wth the prevous fndngs such as Eugene (1981) and Dura and Bhadur (2009). Interestngly, GDP per capta s lkely to have postve coeffcent but opposte sgn comparng to nflaton perod. These results have been studed by Ncholas (2004) when lookng nto the nflaton n G7 countres. It reflects the postve relatonshp between aggregate demand and supply n the economy. Heather and Sopha (2000) developed further that f demand s approachng capacty output and contnue to ncrease, the economy wll get nflaton pressures. The nfluence of reserve assets presents a negatve sgn on deflaton recurrence tme unlke the fndngs mentoned on the prevous studes of Davs and Toma (1995) and Frankel and Rose (1996), the deflaton recovery tme wll be slower f a country has much reserve assets. Ths leads to many expected effects on the economy. Comparng to the nflaton perod, the number of sgnfcant varables n deflaton perod s lower and ther coeffcents are relatedly small. Ths means that the deflaton recurrence perod s less effcent to determne the nfluenced factors over nflaton extng perod. Accordng to Cox and Snell (1968) and appled study of Chen and Chen (2012), Fgure 2 makes the comparson of forecastng precse of the models by applyng the Cox-Snell resdual method. A straght lne s llustrated for the exactly ftted model wth an exponental dstrbuton equal to 1. As can be seen from the graph, both nflaton and deflaton occurrence perods represent the best ftted model among fve survval dstrbutons. The plottng lne reveals that the parametrc survval models perform a comparatvely explanable ft to the data for both nflaton and deflaton survval tme. It seems that the nflaton perod s closer to ft wth the data. 10

11 Fgure 2: Forecastng for Cox-Snell resduals: Panel (A) Panel (B) 5. Conclusons Ths paper apples parametrc shared-fralty models to estmate the nflaton and deflaton recurrence perods. The results underscore several contrbutons as; frstly, for the nflaton perod, the ndustral producton ndex s the most mportant determnant of recurrence tme. Moreover, other determnants lke stock return, real exchange rate, GDP per capta, government expendture, money supply and reserve assets are proved to have mpact on the nflaton tme. A country wshed to mantan the stable nflaton can consder the nfluence of these determnants when combned the fscal and monetary polces. Ths study provdes addtonal nformaton to polcy makers, wth sgnfcant determnants and referred to prevous lteratures to lengthen or shorten the nflaton recurrence when controllng the macroeconomc polces. Secondly, for the deflaton perod, there are only three determnants represented sgnfcant results but neglgble effect n the study samples. In comparson to nflaton perod, the stock returns has the same weght n both two perods, but GDP per capta and reserve assets have dfferent sgnals. As observed n the sample, the polcymakers can rely on the negatve nfluence of GDP per capta and postve reserve assets to lengthen or shorten the nflaton recurrence tme, however, to do that n deflaton they have to apply postve mpact of GDP per capta and negatve reserve assets mpact. Thrdly, ths study has combned the fscal and monetary polces n consderng the prevous lterature to evdence the mplcaton of parametrc shared-fralty model n macroeconomc ssues. These fndngs pont to the consderatons should be taken by researchers and polcymakers when they make plans and strateges n response to mplement nflaton polcy. Accordng to the fndngs, the paper suggests that the determnants whch nfluences the deflaton perod s more complcated than nflaton recurrence tme and needs to explore the strength of other varables n affectng deflaton perod by broaden the data. 11

12 References: Akhand, A. H. (2010). Monetary targetng for prce stablty n Bangladesh: How stable s ts money demand functon and the lnkage between money supply growth and nflaton? Journal of Asan Economcs 21 (6), Arze, A. C., Malndretos, J. and Nppan, S. (2004). Varatons n exchange rates and nflaton n 82 countres: An emprcal nvestgaton. North Amercan Journal of Economcs and Fnance 15, Bck, A. (2010). Threshold effects of nflaton on economc growth n developng countres. Economc letter 108, Bkker, J. A. and Kennedy, N. O. (1999). Composte leadng ndcators of underlyng nflaton for seven EU countres. Journal of forecastng 18, Bll, R. and Anndya, B. (2008). The long-run Phllps curve and non-statonary nflaton. Journal of Macroeconomcs 30, Bruno, M. and Easterly, W. (1998). Inflaton crses and long-run growth. Journal of Monetary Economcs 41, Cagan, P. (1974). Common stock values and nflaton: The hstorcal record of many countres. Natonal Bureau of Economc Research, Annual Report. Catanach, A. H. J. and Perry, S. E. (2001). An evaluaton of the survval model s contrbuton to thrft nsttuton dstress predcton. Journal of Manageral Issues 13 (4), Celasun, O. (2006). Stcky nflaton and the real effects of exchange rate-based stablzaton. Journal of Internatonal Economcs 70, Chen, J. H. (2010). Gender dfferences and job replacement for mutual fund managers. Qualty and Quantty 44, Chen, J. H. and Chen, C. S. (2012). The study of contagous paces of fnancal crses. Qualty and Quantty 46, Choudhry, T. (2001). Inflaton and rates of return on stocks: Evdence from hgh nflaton countres. Journal of nternatonal fnancal markets 11, Clayton, D. G. (1978). A model for assocaton n bvarate lfe tables and ts applcaton n epdemologcal studes of famlal tendency n chronc dsease ncdence. Bometrka 65, Cleves, M., Gould, W. and Guterrez, R. (2008). An ntroducton to survval analyss usng STATA. STATA Corp. U.S.A. Cox, D. R. and Snell, J. (1968). A general defnton of resduals. Journal of the Royal Statstc Socety 30, Davs, G. and Toma, M. (1995). Inflaton, reserve requrements, and real nterest rates wth drect and ndrect loan markets. Journal of Macroeconomcs 17 (3), Dura, S. R. S. and Bhadur, S. N. (2009). Stock prces, nflaton and output: Evdence from wavelet analyss. Economc Modelng 26, Eugene, F. F. (1981). Stock returns, real actvty, nflaton, and money. Amercan Economc Revew 71(4),

13 Fsher, S. (1993). The role of macroeconomc factors n growth. Journal of Monetary and Econometrcs 32, Francsco, J. R. M. (1999). Government expendture and the dynamcs of hgh nflaton. Journal of Development Economcs 58 (2), Frankel, J. and Rose, A. (1996). Currency crashes n emergng markets: An emprcal treatment. Journal of Internatonal Economcs 41, Fredman, M. (1977). Nobel lecture: nflaton and unemployment. Journal of Poltcal Economy 85 (3), Gann, A and Gabrel, F. (2013). Money growth and nflaton: A regme swtchng approach. Journal of Internatonal Money and Fnance 33, Gonzalez, H. B., Ceyla, P., and Bllngs, R. (1997). Determnants of bankng system fraglty: A case study of Mexco. IMF Staff Pap 44(3), Greene, W. (2002). LIMDEP, User s Manual. Econometrc Software, Inc., Planvew, NY, U.S.A. Guerrero, F. and Parker, E. (2006). Deflaton and recesson: Fndng the emprcal lnk. Economcs Letters 93, Guterrez, R. (2001). On fralty models n STATA. Unted Kngdom STATA User s Group Meetng, No. 17. Harrson, M. (2000). Sovet ndustral producton, 1928 to 1955: Real growth and hdden nflaton. Journal of Comparatve Economcs 28, Healther, D. G. and Sopha, L. (2001). Leadng nflaton ndcators for Greece. Economc modelng 18, Hougaard, P. (2000). Analyss of multvarate survval data, Sprnger, NewYork. John, B. T. (2001). The role of the exchange rate n monetary-polcy rules. Amercan Economc Revew 91(2), Joshua, A., Mchael, H. and Ilan, N. (2010). Inflaton targetng and real exchange rates n emergng markets. World Development 39 (5), Khan, M. S. and Senhadj, A. S. (2001). Threshold effects n the relatonshp between nflaton and growth. IMF Staff papers 48, Koman, J. and Tmothy, P. O. (2010). Inflaton and nflaton uncertanty n the ASEAN-5 economes. Journal of Asan Economcs 21, L, L., Narayanc, P. K. and Zheng, X. (2010). An analyss of nflaton and stock returns for the UK. Internatonal Fnancal Markets, Instturon and Money 20, Marka, K., Hector, S. and Denns, J. S. (2005). A reapprasal of the nflaton unemployment trade-off. European Journal of Poltcal Economy 21 (1), Marka, K., Hector, S. and Denns, J. S. (2008). Long-run nflaton-unemployment dynamcs: The Spansh Phllps curve and economc polcy. Journal of Polcy Modelng 30, Narayana, S. and Narayan, P. K. (2013). The nflaton output nexus: Emprcal evdence from Inda, South Afrca, and Brazl. Research n Internatonal Busness and Fnance 28, Ncholas, A. (2004). Inflaton, output growth, volatlty and causalty: Evdence from panel data and the G7 countres. Economcs Letters 83,

14 Pazarbaşoğlu, C. and Ötker, I. (1997). Lkelhood versus tmng of speculatve attacks: A case study of Mexco. European Economc Revew 41, Peter, T. (2008). Do nterest rates drve nflaton dynamcs? An analyss of the cost channel of monetary transmsson. Journal of Economc Dynamcs and Control 32, Samuelson, P. A. and Solow, R. M. (1960). Analytcal aspects of ant-nflaton polcy. Amercan Economc Revew 5, Syed, Z. A. and Sajd, A. (2013). Inflaton and nterest rates n the presence of a cost channel, wealth effect and agent heterogenety. Economc Modelng 31, Tudela, M. (2004). Explanng currency crses: A duraton model approach. Journal of Internatonal Money and Fnance 23, Zya, Ö. and Süleyman, Ö. (1990). Exchange rates, nflaton and money supply n Turkey: Testng the vcous crcle hypothess. Journal of Development Economcs 32, Gruben, W. C., and McLeod, D. (2002). Captal account lberalzaton and nflaton. Economcs Letters 77(2), Appendx 1: Sample countres lst Countres Perod Countres Perod 1 Canada Belgum Japan Hungary Unted Kngdom Indonesa Unted State Malaysa Italy Phlppne Germany Sngapore France Thaland Australa South Afrca Sweden Inda Austra Brazl Span Russa Swtzerland Turkey Norway Mexco