Effectiveness of Monetary Policy Communication in Indonesia and Thailand

Transcription

1 Effeciveness of Moneary Policy Communicaion in Indonesia and Thailand Sahminan Sahminan 1 Bank Indonesia (Firs Draf: 3 June 8) Absrac In his paper we invesigae he effeciveness of Bank Indonesia s and Bank of Thailand s moneary policy communicaion. We focus on wo channels of communicaion: moneary policy saemens, and iner-meeing saemens. Alhough he srucure of Bank Indonesia s and Bank of Thailand s moneary policy saemens have some differences, mos of he saemens conain policy inclinaion. In addiion, during iner-meeing periods, members of heir board of governors ofen convey saemens ha conain policy inclinaion. Our empirical resuls show ha o some exen Bank Indonesia s and Bank of Thailand s moneary policy saemens and iner-meeing saemens move shor-erm ineres raes effecively. We find ha here is asymmery in he effecs of he saemens, ha is, he saemens wih loose policy inclinaion end o be more effecive relaive o he saemens wih igh policy inclinaion. JEL codes: E43, E5, E58 Keywords: communicaion, effeciveness, moneary policy, Bank Indonesia, Bank of Thailand. 1 I would like o hank Ramon Moreno, William Nelson and Vongsinsirikul Sukpinnara for helpful discussions. This paper was wrien while I was a visiing fellow a he Bank for Inernaional Selemens in Basel. The views expressed in his paper are hose of he auhor and do no necessarily reflec he views of Bank Indonesia. All errors are my own. address: sahminan@bi.go.id.

2 1. Inroducion In he pas en years or so Bank Indonesia (BI) he cenral bank of Indonesia and Bank of Thailand (BOT) he cenral bank of Thailand have inroduced moneary policy communicaion as an imporan par of heir moneary policy implemenaion. In communicaing moneary policy, he BI and he BOT have used various channels of communicaion, including press releases, publicaions, speeches, esimonies, and inerviews. Through hose channels of communicaion, he BI and he BOT provide a wide coverage of informaion including economic and financial daa, inerpreaion of daa, research and policies. There are a number facors conribuing o his developmen. The mandae obained by he BI and he BOT as an independen cenral bank is cerainly one of he main facors 3. As he BI and he BOT become more independen hey need beer accounabiliy and ransparency, and o achieve beer ransparency hey need effecive communicaion. Anoher imporan facor behind his developmen is he fac ha he BI and he BOT have implemened inflaion argeing and recognized ha communicaion plays an imporan role in such a framework. While he imporan role of communicaion in moneary policy has been widely recognized and communicaion has become an imporan par of moneary policy in many counries including emerging marke counries such as Indonesia and Thailand he quesions on he effeciveness of such communicaion o suppor moneary policy remain. The lieraure on moneary policy communicaion does no provide a conclusive resul on which communicaion sraegy ha can be used effecively by a paricular cenral bank. As recenly poined ou by Blinder e al (8), for example, wheher or no here are beer/worse ways a cenral bank communicaes is sill largely an empirical quesion. The purpose of his paper is o examine he effeciveness of moneary policy communicaion in Indonesia and Thailand. The analysis will focus on wo channels of moneary policy communicaion: moneary policy saemens referring o press releases 3 Filardo and Guinigundo (8) provide a comprehensive survey on recen developmens in communicaion pracices in cenral banks in he Asia Pacific region. In Indonesia, as poined ou by Goelom (7), before he BI became an independen cenral bank in 1999, he BI was kep quie and is roles and funcions were communicaed in line wih governmen policy implemenaion. 1

3 delivered by he cenral bank immediaely afer moneary policy meeings; and iner-meeing saemens referring o he saemens conveyed by members of he cenral banks board of governors during iner-meeing periods. These wo channels of communicaion are wached closely by financial marke paricipans in par because hey ofen conain cenral bank views on he economic oulook and possible consequences for moneary policy in he near erm. To beer undersand he differences and similariies in he BI s and he BOT s communicaion sraegies, we firs examine he srucure and conen of he wo cenral banks moneary policy and iner-meeing saemens. We hen examine wheher hose wo channels of communicaion move shor-erm ineres raes. The availabiliy of informaion on policy inclinaion, raher han jus informaion on he acions aken, allows us o examine wheher or no communicaion moves ineres raes in he cenral banks inended direcion. We expec ha effecive communicaion will move ineres raes in a manner consisen wih he saed policy inclinaion. The paper proceeds as follows. In secion we provide a lieraure review on cenral bank communicaion. In secion 3 we briefly discuss moneary policy decision-making process in Indonesia and Thailand. In secion 4 we analyze he BI s and he BOT s moneary policy saemens and iner-meeing saemens. In secion 5 we presen an empirical analysis on he effecs of moneary policy communicaion on inerbank ineres raes. Finally, he paper concludes wih Secion 6.. Lieraure Review Recenly here have been a large number of sudies focusing on he role of communicaion in moneary policy. The heoreical lieraure assumes ha moneary policy only has direc effecs on shor-erm ineres raes; he effecs on long-erm ineres raes and oher asse prices are channelled hrough expecaions (Blinder 1998). Communicaion is believed o help reduce uncerainy facing economic agens due o he presence of asymmeric informaion beween moneary policy makers and oher economic agens (Geera, ). Through he

4 expecaion channel, cenral bank s communicaion plays an imporan role in influencing long-erm asse prices ha are imporan for he economy 4. Alhough he heoreical lieraure recognizes he imporance of cenral bank communicaion, some heoreical sudies show ha a more ransparen moneary policy is no necessarily desirable. A widely-cied sudy by Morris and Shin () explains condiions under which more public informaion may reduce welfare. They show ha he economic effecs of public informaion arise from is role in conveying fundamenal informaion and in serving as a focal poin for coordinaion. Public informaion can cause damage because privae agens may over-reac o such informaion and suppress heir own informaion. Svensson (6) has challenged he resuls of Morris and Shin (), showing ha only under very special circumsances does more public informaion resuls in lower welfare. Even wih a conservaive benchmark when he qualiy of privae informaion equals he qualiy of public informaion social welfare is sill higher han social welfare under he siuaion wihou public informaion. Svensson shows ha public informaion can only be welfare reducing if privae informaion conains a leas eigh imes he precision 5 of public informaion. Woodford (5) argues such a condiion is very unlikely and i is implausible ha public informaion provided by a cenral bank based on is bes guess could reduce welfare as implied by Morris and Shin s model. If communicaion is desirable for he effeciveness of moneary policy, he quesion is wha ype of informaion needs o be communicaed publicly by a cenral bank. Woodford (5) proposes four iems: (1) inerpreaion of economic condiions, () conen of policy decisions, (3) sraegy ha guides decision, and (4) oulook of fuure policy. While all hese issues migh be of he ineres o he public, he exen o which ransparency can help achieve sabilizaion goals can vary. As Woodford poined ou, communicaing he inerpreaion of economic condiions and conen of policy decisions are he leas conroversial, and have been pu ino pracice by many cenral banks. There is less agreemen on he exen o which cenral banks should communicae is fuure policy decisions. A radiional argumen agains ransparency (which has become less acceped among cenral bankers), is ha cenral bank 4 5 The lieraure on moneary policy provides a variey of explanaions on how communicaion affecs shorerm ineres rae expecaion, long-erm ineres rae and he economy. See for example, Blinder e al (8). Precision of he informaion is measured in erms of he noise in he signal: he higher he noise he lower is he precision. 3

5 inervenion will be effecive only o he exen ha i surprises he markes. Anoher argumen is ha here are possible disadvanages o public informaion provision. A cenral bank has limied informaion ha marke paricipans wish o know abou, and bad informaion provided by cenral banks could harm marke paricipans and public in general. Nowihsanding he preceding argumens, many believe ransparency and communicaion in moneary policy are imporan. The resul is a grea deal of heerogeneiy in he pracices of communicaion among cenral banks. To address his issue, many empirical sudies have aemped o examine sraegies and he effeciveness of moneary policy communicaion implemened by paricular cenral banks. Guhrie and Wrigh () analyze he effecs of communicaion by he Reserve Bank of New Zealand (RBNZ) on ineres raes, and hey find ha he RBNZ has used communicaion sysemaically and effecively in conrolling shor-erm ineres raes. Demirlap and Jorda () examine he exen o which he announcemen of a change in he Federal Fund rae affecs erm srucure of Treasury securiies, which is a key ingredien of moneary ransmission mechanism in he US. They find ha wih he announcemen of he policy decision and he FOMC schedule, markes can beer anicipae iming and naure of policy moves. Kohn and Sack (4) show ha saemens by he Federal Open Marke Commiee (FOMC) and esimony by Chairman Greenspan significanly affec marke ineres raes in he US. In paricular, hey find ha moneary policy saemens significanly affec shor-erm ineres raes while a saemen on asse valuaion has a less imporan effec. A number of cross-counry sudies assess he effeciveness of moneary policy communicaion under differen sraegies. Connolly and Kohler (4) examine he effecs of news relaed o he expeced pah of moneary policy on ineres rae fuures in six developed counries: Ausralia, Canada, he Euro area, New Zealand, he UK and he US. They find ha boh macroeconomic news and policy news significanly affec ineres rae expecaions. Commenary wih rae decisions, moneary policy repor, and parliamenary hearings are he channels of communicaion ha have large influences on ineres rae expecaions in all six counries. On he oher hand, oher channels such minues of meeings have significan effecs only in some counries. Ehrmann and Frazcher (7) examine he sraegies and he effeciveness of moneary policy communicaion underaken by he Federal Reserve, he European Cenral 4

6 Bank (ECB), and he Bank of England. They show ha he Fed ends o use an individualisic 6 communicaion sraegy while he ECB and he Bank of England end o use a collegial communicaion sraegy. Nowihsanding hese differences in communicaion sraegies, he predicabiliy of policy decisions and financial marke responsiveness o communicaions by he Fed and he ECB are equally successful in heir effeciveness. While many sudies have analyzed he effecs of moneary policy communicaion in developed counries, only a few sudies have examined he effeciveness of moneary policy communicaion in emerging marke counries. Rozkru e al (7), for example, sudy how cenral banks communicaion in Czech Republic, Hungary and Poland affec financial markes. They find ha cenral banks communicaion sraegies in hose hree counries are quie differen, and he effeciveness of communicaion in influencing moneary policy predicabiliy depends on he cenral banks moneary policy srucure and communicaion sraegy. Recenly, Garcia-Herrero and Remolona (8) examine he effeciveness of moneary policy communicaion in Asia and he Pacific. For ha purpose, hey propose and use wo differen mehodologies, ha is, a es on he efficiency hypohesis of he erm srucure of ineres raes, and a es on wheher a surprise due o he presence of policy saemens is larger han a surprise due o he presence of macroeconomic daa releases. They find ha o some exen he yield curve provides informaion on he fuure policy rae, and relaive o macroeconomic news, policy saemens conain a larger elemen of surprise. This paper conribues o he empirical lieraure on cenral bank communicaion in wo imporan ways. Firs, his paper provides an analysis of he srucure and conen of moneary policy saemens and iner-meeing saemens released by cenral banks in emerging marke counries. Second, his paper provides evidence on he effeciveness of moneary policy saemens and iner-meeing saemens in moving shor-erm ineres raes in emerging marke counries. Insead of esing only he presence of a surprise as a resul of moneary policy communicaion, his paper assesses wheher or no ineres raes move in he same direcion as he direcion of he saed policy inclinaion. 6 Individualisic communicaion sraegy refers o a sraegy in which wha he individual commiee members say have a high degree of dispersion. 5

7 3. Moneary Policy Decision Making a he BI and he BOT Following he Asian crisis in he lae 199s, he BI and he BOT have become more independen and moved owards he implemenaion of inflaion argeing. The BOT sared o implemen inflaion argeing in May, while he BI sared in abou five years laer. Under he inflaion argeing, he BI and he BOT have he auhoriy o se he direcion of moneary policy in which he overriding arge is price sabiliy. In line wih he implemenaion of inflaion argeing, a significan change in he operaional arge used by he BI and he BOT also ook place. Saring in July 5, he operaional arge used by he BI changed from base money o an ineres rae, he 1-monh Bank Indonesia cerificae (SBI) rae his operaional arge is also called BI Rae 7. A he BOT he use of ineres rae as an operaional arge sared in May. Unil mid January 7 he BOT used he 14-day repo rae, and hen changed o he 1-day repo rae. Moneary policy decisions a he BI are made by a board of governors comprising he governor, he senior depuy governor, and all oher depuy governors. A he BOT, moneary policy decisions are made by a moneary policy commiee (MPC) comprising he op managemen of he BOT he governor, he depuy governor for moneary sabiliy, and he depuy governor for financial insiuions sabiliy and disinguished expers from ouside he BOT. The BI holds moneary policy meeings every monh, while he BOT holds he meeing every six weeks. The policy meeing evaluaes sae of he economy, and ses he direcion of moneary policy. The schedule for policy meeings a he BI is eiher on firs Tuesday or firs Thursday of every monh. If for some reason policy meeing canno be held eiher on Tuesday or Thursday, he policy meeing is held on a differen day. A he BOT, policy meeings are mosly held on Wednesday. All policy meeing days a he BI and he BOT are pre-announced publicly. A he BOT he schedule of all eigh policy meeings hroughou he year is announced publicly a beginning of each year. Boh he BI and he BOT announce moneary policy decisions immediaely afer he meeing is concluded. A he BI moneary policy saemens are released by Office of he 7 Saring in July 9 h, 8, Bank Indonesia began using an overnigh rae. 6

8 Governor while a he BOT hey are released by he Communicaion and Relaion Office. The press releases on he policy rae are published by he BI and he BOT in heir respecive naional languages as well as in English. Thus domesic and foreign invesors are expeced o receive he same messages conveyed in he press releases. In addiion o he press release on he moneary policy decision, he BI and he BOT also publish a moneary policy repor every quarer. The informaion and analysis in he moneary policy repors conain much more deail han wha is provided in he press releases. Moreover, he BI and he BOT also provide a wide range of economic and financial daa and informaion. All press releases, moneary policy repors, and various daa and informaion are available in he he BI s and he BOT s websies. Oher han press releases, repors, and publicaions, he BI s and he BOT s governors and depuy governors also occasionally make saemens regarding he moneary policy. The saemens may come ou in speeches a various evens such as conferences and workshops, inerviews, or esimonies before parliamen. 4. Srucure and Conen of he BI s and he BOT s Moneary Policy Communicaion In his secion we discuss he srucure and conen of he BI s and he BOT s moneary policy communicaion, and he way we measure such communicaion. As we have menioned a he ouse, in his paper we focus on analyzing wo channels of cenral bank communicaion, namely moneary policy saemens released immediaely afer each policy meeing, and inermeeing saemens by members of he board of governors (BI) or moneary policy commiee (Bank of Thailand). 4.1 Moneary Policy Saemens To idenify he srucure and conen of he moneary policy saemens, we look a each of he English version of he BI s and he BOT s press releases on moneary policy decisions from January 4 o December 7. Alhough he BI had no implemened he inflaion argeing before July 5, he BI had underaken regularly moneary policy meeings o decide he moneary policy sance. To look a he srucure of moneary policy saemens, we coun he oal number of words in he saemens, number of paragraphs, average number of words per paragraph, number of words in he decision paragraph, and average senence lengh in he decision 7

9 paragraph. Informaion on he srucure of he moneary policy saemens is expeced o give some idea on he conciseness and he consisency of moneary policy saemens over ime. Summary saisics of he srucure of moneary policy saemens show ha he BOT has much shorer and more concise moneary policy saemens compared wih hose of he BI (Table 1). The average number of words in he BI s moneary policy saemens is 769, while he average number of words in he BOT s moneary policy saemens is only 56. The BI covers policies on no only moneary policy sance, bu also ofen on oher issues such as issues in banking secor. On he oher hand, he conen of he BOT s moneary policy saemens is very much closely relaed o moneary policy sance. This facor seems o be par of he reasons why he BI s moneary policy saemens are in general longer han hose of he BOT. If we look a he changes in he srucure of he moneary policy saemens over ime, he BOT s moneary policy saemens have ended o change less. As shown in Figure 1, number of words, number of paragraphs, number of words per paragraph and average lengh of senences in he BI s moneary policy saemens flucuae subsanially. On he oher hand, number of words, number of paragraphs, and number of words per paragraph in he BOT s moneary policy saemens do no change much. Moreover, he number of words in he BOT s moneary policy saemens has ended o decrease over ime. Of course, a shorer saemen is no necessarily clearer han a longer saemen. However, i has been widely known ha normal human being have maximum capaciy o process informaion. A sudy by Cowan (1), for example, argues ha he capaciy of working memory of normal adul human is abou four chunks defined as a collecion of conceps ha have srong associaion o one anoher bu much weaker associaion o oher chunks concurrenly used. Alhough he exac amoun of informaion people can process can be debaed 8, given ha people have limied capaciy or working memory, a shorer saemen migh well provide a clearer signal of he main message he saemen inends o convey. Thus differences in saemen srucure may accoun for some of he empirical resuls obained below. Evidence on he imporance of clariy and readabiliy of moneary policy saemens or repors for reducing uncerainy in he ineres rae movemens can be found, for example, in Fracasso, Genberg, and Wyploz (3). 8

10 Regarding he conen of he moneary policy saemens, for each moneary policy saemen we look a hree ypes of informaion i.e. assessmen of he economic condiion, direcion of he fuure policy, and possible risk o he oulook. Examples of he saemens are provided in Appendix 1. While all he BI s and he BOT s moneary policy saemens conain an assessmen of he economic condiion, hey do no always conain policy direcion and possible risks o he oulook. Seven ou of he 48 BI s moneary policy saemens do no conain a saemen on possible risks o he oulook, and five ou of he 3 BOT s moneary policy saemens do no conain a direcion of fuure policy. Based on he keywords presened in Table, we classify moneary policy saemens according o wheher hey conain a igh policy inclinaion, neural policy inclinaion or loose policy inclinaion. Using such a classificaion, we find ha over he period from January 4 o December 7, policy inclinaion in he BI s moneary policy saemens consiss of 58 percen igh, 19 percen neural and 3 percen loose (Table 3). Over he same period, policy inclinaion in he BOT s moneary policy saemens consiss of 53 percen igh, 31 percen neural and 16 percen loose. 4. Iner-meeing Saemens Informaion on he iner-meeing saemens by members of he BI s and he BOT s board of governors is searched by using Google News. The keywords used in he searching are names of he governors and depuy governors. Insead of searching all members of he board of governors, we only limi our search for he ones who are mos likely providing saemens relaed o moneary policy. For he BI we include he saemens by he governor, he senior depuy governor, he depuy governor for moneary secor and he depuy governor for he financial marke secor. Meanwhile, for he BOT we include he saemens by he governor, he depuy governor for moneary sabiliy, and he depuy governor for financial insiuions sabiliy. In some cases, due o a replacemen over he period sudied, a posiion for a governor or a depuy governor can be held by wo differen persons. If his is he case hen we include he saemens of hose wo persons during heir enure. Afer we pull ou all he news relaed o he members of he board of governors, we examine he conen of each of he news and idenify wheher or no he news conains 8 A classical Miller s (1956) rule says ha he working memory of normal people is abou seven, and his rule has been ofen used as a reference for communicaion. 9

11 saemens on moneary policy inclinaion 9. When we can access he aricle fully from is source hen we examine he quoed saemens. Bu when we canno access he aricle fully, we only use he headline of he news. Based on he keywords presened in Table 4, all of he news conaining moneary policy inclinaion is classified ino igh, neural or loose moneary policy inclinaion. The examples of he iner-meeing saemens wih policy inclinaion are presened in Appendix. Summary saisics of he policy inclinaion in he iner-meeing saemens are presened in Table 5. Over he period sudied, he number of iner-meeing saemens wih moneary policy inclinaion conveyed by he BI s and he BOT s governors are quie similar. The BI governor had 19 saemens while he BOT governor had 1 saemens. In he case of he BI, oher han he governor, he senior depuy governor also ofen conveys iner-meeing saemens ha conain policy inclinaion. Oher han he senior depuy governor, depuy governors a he BI and he BOT convey only a few iner-meeing saemens. Neverheless, if we compare he numbers of iner-meeing saemens by depuy governors a he BI and he BOT, he BI depuy governors convey iner-meeing saemens more ofen. The iner-meeing saemens conveyed by differen members of he BI s board of governors as well as by differen members of he BOT s MPC are quie consisen. As shown in Figure, over he period sudied here are only wo cases a he BI, and none a he BOT in which he saemens conveyed by differen members of board of governors or MPC conain differen policy inclinaion during he same iner-meing period. 5. Empirical Analysis While in he previous secion we discuss he srucure and conen of he moneary policy communicaion, in his secion, using economeric echniques we assess he effeciveness of he moneary policy communicaion in moving shor-erm ineres raes. 5.1 Economeric Model The model we use in examining he effecs of communicaion on ineres raes is Exponenial Generalized Auoregressive Condiional Heeroskedasiciy (EGARCH), a model proposed by 9 In fac, members of he board of governors may give saemens on a wide range of issues ha are no necessarily direcly relaed o moneary policy. 1

12 Nelson (1991) and has been widely used in analyzing he effecs of announcemens on financial markes 1. This model provides esimaes of he effecs of communicaion no jus in level bu also in volailiy. Moreover, his model also has he advanage of correcing for kurosis, skewness, and ime-varying volailiy of asse prices, as well as avoiding he nonnegaiviy consrains on he condiional second momen. As he dependen variables in he models, we use daily changes in inerbank ineres raes and daily changes of implied forward ineres raes 11. To check for he robusness of he resuls, we esimae he models for differen mauriies of ineres raes and implied forward raes. Ineres rae mauriies range from 1 monh o 1 year, while he mauriies of forward rae are 3 monh and 6 monh. The implici forward rae is calculaed as follows. Le r i and rj be inerbank ineres raes wih i - and j - monh mauriies, respecively. Then, inerbank forward rae (i - j ) is defined as: where i ri F( i j) = 1 x1 j r (1) j j i > j. To capure he effecs of policy inclinaion conained in he saemens, insead of using a single cardinal variable, we use differen dummy variables for differen direcions of policy inclinaion. The advanage of his approach is ha i does no require cardinaliy resricions, and allows us o capure he presence of asymmery in he effecs of communicaion under differen direcions of policy inclinaion. Thus, dummy variables for moneary policy saemens are defined as: M = 1 if he saemen conains igh policy inclinaion; oherwise + M = 1 if he saemen conains neural policy inclinaion; oherwise M = 1 if he saemen conains loose policy inclinaion; oherwise 1 11 See for example, Ehrmann and Frazer (7), and Connlly and Kohler (7). In a number of sudies, ineres rae expecaion is measured by fuure ineres rae. In his paper due o he lack of fuure rae daa, we use implied forward rae. Indeed, as poined ou by Brooke and Cooper (), forward rae is also commonly used as a measure for ineres rae expecaion. 11

13 And dummy variables for iner-meeing saemens are defined as: I = 1 if he iner-meeing saemen conains igh policy inclinaion; oherwise + I = 1 if he iner-meeing saemen conains neural policy inclinaion; oherwise I = 1 if he iner-meeing saemen conains loose policy inclinaion; oherwise Oher han dummy variables for communicaion, we also ake ino accoun auoregressive behaviour in he dependen variables, he effecs of changes in he policy rae, and he effecs of changes in ineres raes in inernaional financial markes. As he policy rae for Indonesia we use 1-monh SBI rae, and as he policy rae for Thailand we use 1-day repo rae. As a proxy for ineres raes in inernaional financial markes we use Fed Fund rae. Besides communicaion variables and ineres raes, we also include conrol variables consis of macroeconomic announcemens and days of he week. We capure macroeconomic announcemens by using a dummy variable for GDP announcemens and a dummy variable for inflaion announcemens. Specifically, he model is formulaed as follows. Le r be he changes in ineres raes, hen he mean equaion of he EGARCH model is given by: 4 4 = r α αir i βi pi γ i ff i + μ1m + μm + μ3 i= 1 i= 1 M where p is he change in policy rae, + θ I + θi + θ3i + ζx ε () ff is he change in Fed Fund rae, and X is a vecor of conrol variables. In his model, i is assumed ha ε = σ. v where v is independenly idenical disribued random variable wih zero mean and uni variance. The effec of communicaion on he volailiy of he dependen variables is capured hrough he variance equaion: ε ε log( σ ) = ϖ + ζ log( σ ) + ϕ + ϑ + τ M + τ I + δx (3) σ 1 σ 1 In he variance equaion, insead of aking ino accoun direcion of he policy inclinaion, we only include dummy variables ha capure he presence of he saemens, ha is: M = 1 if here is a moneary policy saemen; oherwise 1

14 I = 1 if here is an iner-meeing saemen; oherwise The hypohesis for he mean equaion is ha if communicaion is effecive hen i moves ineres rae in he inended direcion. Thus, we expec ha a saemen conains igh policy inclinaion leads o higher marke ineres raes, a saemen conains neural policy inclinaion leads o unchanged marke ineres raes, and a saemen conains loose policy inclinaion leads o lower marke ineres raes. The hypohesis for he variance equaion is less sraigh forward. Some sudies argue ha if communicaion provides new informaion for he markes hen he presence of communicaion is expeced o resul in higher volailiy of ineres raes 1. However, higher volailiy can also be inerpreed as an indicaion of higher uncerainy and hereby less effecive communicaion, given ha he goal of communicaion is o reduce uncerainy in he markes. Here, insead of esing a hypohesis on he direcion of he effec of communicaion on he volailiy, we assess he exen o which he volailiy reacs o he communicaion. 5. Esimaion Resuls Using he model specified in equaions () and (3) we esimae mean equaions and variance equaions for he effecs of he BI s and he BOT s moneary policy communicaion. The resuls for he mean equaions are presened in Table 6, and he resuls for volailiy equaions are presened in Table 7. Moneary Policy Saemens The esimaion resuls for he BI s moneary policy saemens show ha he saemens conaining loose moneary policy inclinaion significanly lower ineres rae changes. However, he resuls also show ha he saemens wih neural moneary policy inclinaion lower ineres rae changes. The resuls are quie robus across differen mauriies of ineres raes, in which he magniudes of he coefficien range from.3 o.6 for loose policy inclinaion, and.4 o.9 for neural policy inclinaion. 1 This idea is followed, for example, by Kohn and Sacks (4), and Connolly and Kohler (4). 13

15 The policy inclinaion in he BOT s moneary policy saemens in general moves ineres raes in he inended direcion. While loose policy inclinaion in general significanly lowers ineres rae changes, neural policy saemens do no significanly affec ineres raes, as expeced. On he oher hand, igh moneary policy inclinaion ends o have mixed effecs. In erms of he magniudes, when he effecs are significan, he effecs of loose policy inclinaion in Thailand are generally larger han he magniudes in Indonesia. The effecs of he BOT s loose policy inclinaion on 1-monh and 3-monh inerbank ineres raes in Thailand are even more han 1 basis poins, while he effecs of he BI s loose policy inclinaion for he same mauriy of ineres raes are less han 5 basis poins. If we compare hese resuls wih he evidence from hree major cenral banks he Federal Reserve, he Bank of England, and he ECB found by Ehrmann and Frazscher (7), when hey are significan, he effecs of he BI s and he BOT s communicaion end o be much larger. Ehrmann and Frazscher (7) find ha he effecs of policy inclinaion for hose hree major cenral banks range from 1.5 o.5 basis poins. Looking a he influence of he BI s moneary policy saemens on he volailiy of ineres raes in Indonesia, he resuls show ha in mos cases moneary policy saemens do no significanly affec volailiy of ineres raes. On he oher hand, excep for he volailiy of ineres raes wih 1-monh mauriy, he BOT s moneary policy saemens significanly increase volailiy of inerbank ineres raes in Thailand, and he magniudes of he coefficiens are consisenly larger han one. Iner-Meeing Saemens Similar o he effecs of loose policy inclinaion in he BI s moneary policy saemens, loose policy inclinaion in he BI s iner-meeing saemens also significanly lower ineres raes, in which heir magniudes range from.8 o.4. The main difference wih he effecs of moneary policy saemens is ha a neural policy inclinaion in he iner-meeing saemens does no significanly affecs ineres raes. The effecs of he BOT s iner-meeing saemens in some cases are in line wih he direcion of he policy inclinaion. A igh policy inclinaion in he iner-meeing saemens significanly leads o higher 1-monh ineres raes, 3-monh forward raes and 6-monh forward raes. On he oher hand, a loose policy inclinaion in he iner-meeing saemens significanly leads o lower 1-monh ineres raes and 3-monh forward raes. However, a neural policy inclinaion also significanly leads o lower 1-monh and 1-monh ineres raes. 14

16 In erms of he magniudes, he resuls show ha for Indonesia s case here is no sysemaic difference beween he effecs of iner-meeing saemens and he effecs of moneary policy saemens. On he oher hand, in Thailand he effecs of iner-meeing saemens end o be smaller han he effecs of moneary policy saemens. If we inerpre ha he difference in he magniudes indicaes he difference in he imporance of differen channels of communicaion for inerbank ineres raes, hen in Thailand moneary policy saemens seem o be more imporan relaive o iner-meeing saemens. Regarding he effecs of iner-meeing saemens on he volailiy of ineres raes, in mos cases he iner-meeing saemens do no significanly affec volailiy of ineres raes in Indonesia and Thailand. Given ha he policy inclinaion in he iner-meeing saemens moves ineres raes significanly, he insignificance effec of he iner-meeing saemens on he volailiy canno be inerpreed as an indicaion of lack of new informaion in he inermeeing saemens. Insead, such insignificance seems o be more reasonable inerpreed as an indicaion of he lack of ineres rae uncerainy due o he iner-meeing saemens. Comparisons of he Resuls The esimaion resuls above show ha o some exen moneary policy saemens and iner-meeing saemens have been effecive in moving domesic inerbank ineres raes in Indonesia and Thailand. If we look a he effecs of he saemens under differen direcions of policy inclinaion, he resuls show ha here is asymmery in he effeciveness of he saemens, in which he saemens wih loose policy inclinaion end o be more effecive relaive o he saemens wih igh policy inclinaion. Moreover, he resuls also show ha while he BOT s moneary policy saemens wih neural policy inclinaion is also quie effecive, he BI s moneary policy saemens wih neural policy inclinaion end o bring lower ineres raes. While he difference in he srucure of moneary policy saemens may accoun for he differences beween he resuls for Indonesia and Thailand, i does no explain why here is asymmery in he effeciveness of he saemens under differen direcions of policy inclinaion. To explain such asymmery we examine he associaion beween policy inclinaion in he saemens and he changes in policy rae following he saemens. As shown in Figure 3 and Figure 4, i is quie obvious ha he BI s and he BOT s saemens wih loose policy inclinaion are mosly followed by lower policy raes. On he oher hand, when he saemens conain igh or neural policy inclinaions, he BI and he BOT show differen paerns. In many cases he BI s saemens wih igh policy inclinaion are followed 15

17 by unchanged policy raes, and he BI s saemens wih neural policy inclinaion are ofen followed by lower policy raes. On he oher hand, he BOT s saemens wih igh policy inclinaion are mosly followed by unchanged policy rae and he BOT s saemens wih neural policy inclinaion are mosly followed by unchanged policy rae. The associaion beween he policy inclinaion and he change in he following policy rae is summarized in Table 8. As we can see for Indonesia s case, only 36 percen of he BI s moneary policy saemens wih igh policy inclinaion is followed by a higher policy rae, while 91 percen of he saemens wih loose policy inclinaion are followed by a lower policy rae. Moreover, we can also see ha 44 percen of he BI s moneary policy saemens wih neural policy inclinaion are followed by a lower policy rae. When we look a Thailand s case, he BOT s moneary policy saemens wih neural policy inclinaion and loose policy inclinaion are quie consisen wih he changes in he policy rae. Niney percen of he BOT s moneary policy saemens wih neural policy inclinaion are followed by an unchanged policy rae, and 8 percen of he he BOT s moneary policy saemens wih loose policy inclinaion are followed by a lower policy rae. If we look a he associaion beween policy inclinaion in he iner-meeing saemens and changes in he policy rae, we can also see ha he policy inclinaion in he BOT s iner-meeing saemens is much more in line wih he changes in is policy raes. While only around 6 percen of he policy inclinaion in he BI s iner-meeing saemens ha is in line wih he direcion of he change in he BI s policy rae, more han 75 percen of he policy inclinaion in he BOT s iner-meeing saemens is in line wih he direcion of he change in he BOT s policy rae. One possible reason why policy inclinaion in he BI s saemens are no necessarily followed by corresponding changes in he policy rae could be because he BI jus sared o use ineres raes as he operaing arge in July 5. As shown in Figure 3, before July 5 for abou a year he SBI rae as a proxy for he BI s policy rae before July 5 hardly changed alhough moneary policy saemens conain igh policy inclinaion. Afer July 5 we find ha while all moneary policy saemens wih loose policy inclinaion are sill followed by lower policy raes, mos of he moneary policy saemens wih igh policy inclinaion are no followed by higher policy rae (Table 9). The main difference is observed in he associaion beween policy inclinaion in he iner-meeing saemens and he changes in policy rae, in which he saemens wih igh policy inclinaion are mosly followed by higher policy rae. 16

18 The quesion now is wheher he change in he BI s moneary policy framework since July 5 has improved he effeciveness of is moneary policy communicaion. To address ha quesion we esimae he empirical model using a sample of daa covering he period from July 5, and he esimaion resuls are presened in Table 1 and Table 11. The resuls for he mean equaion using daa since July 5 do no show a significan difference from he resuls for he mean equaion using he full sample. Using daa from July 5, he esimaion resuls show ha moneary policy saemens wih loose or neural policy inclinaion sill lead o a lower ineres rae change, while moneary policy saemens wih igh inclinaion sill do no significanly lead o a higher ineres rae change. However, he effecs of he moneary policy saemens on he volailiy of ineres raes using daa since July 5 is significan in almos all cases, in which moneary policy saemens resul in higher ineres rae volailiy. This resul is differen from he esimaion resul using daa from January 4, in which only he volailiy of 6-monh ineres rae ha is significanly affeced by moneary policy saemens. The effecs of he BI s iner-meeing saemens since July 5 also end o be negaive when policy inclinaion is loose, alhough in mos cases he effecs are insignifican. The iner-meeing saemens since July 5 only significanly move 1-monh inerbank ineres rae, in which loose policy inclinaion leads o a lower ineres rae changes while neural policy inclinaion leads o a higher ineres rae changes. If we look a he variance equaion, he effecs of iner-meeing saemen on he ineres rae volailiy do no change subsanially in which in mos cases he iner-meeing saemens do no significanly affec he volailiy of ineres raes. 6. Conclusions In his paper we invesigae he effeciveness of moneary policy communicaion in Indonesia and Thailand. We focus on wo channels of moneary policy communicaion, namely saemens on moneary policy decisions, and iner-meing saemens by members of he board of governors. We firs analyse he BI s and he BOT s moneary policy saemens and iner-meeing saemens over he period from January 4 o December 7. Looking closely a he srucure and conen of he moneary policy saemens, we find he following paerns. In erms of number of words and number of paragraphs, he BI s moneary policy saemens in general are much longer and show more flucuaion compared wih he BOT s moneary policy saemens. The BOT s moneary policy saemens are relaively shor and 17

19 he number of words in he saemens ends o decrease over ime. If we look a he conen of he BI s and he BOT s moneary policy saemens, almos all saemens cover he sae of he economy, direcion of he fuure policy and possible risks o he oulook. Beside communicaion hrough moneary policy saemens on meeing days, we also find ha during iner-meeing periods members of he board of governors ofen convey saemens ha conain moneary policy inclinaion. Such saemens are conveyed in various occasions, including conferences, inerviews, or esimonies before he parliamen. No surprisingly, he governors and he senior depuy governor in he case of he BI are he ones who deliver saemens mos ofen relaive o oher member of he board of governors. If differen members of he board of governors happen o deliver saemens during he same iner-meeing periods, he policy inclinaion conveyed by differen person are quie consisen. Afer analyzing he srucure and conen of he moneary policy saemens and inermeeing saemens, we examine empirically how he saemens affec financial markes, and he findings are as follows. Firs, he BI s moneary policy saemens ha conain loose policy inclinaion end o lower inerbank ineres raes. However, while he BI s saemens wih igh policy inclinaion do no move ineres raes, he saemens wih neural policy inclinaion also end o lower he ineres raes. Similar o he effecs of he BI s moneary policy saemens wih loose policy inclinaion, he BOT s saemens wih loose policy inclinaion also significanly lower ineres raes. On he oher hand, while he saemens wih igh policy inclinaion give mixed resuls, he BOT s saemens wih neural policy inclinaion does no move ineres raes. Regarding he effecs of moneary policy saemens on he volailiy of ineres raes, he effecs in Indonesia and Thailand are quie differen. In Indonesia, in mos cases moneary policy saemens do no significanly affec ineres rae volailiy, while in Thailand such saemens have significan effecs and he magniudes of he effecs are larger han one. Second, iner-meing saemens by member of he BI s board of governors seem o be effecive when he iner-meing saemens conain loose or neural policy inclinaion, while he saemens wih igh policy inclinaion do no significanly move he ineres raes. If we look a he effecs of iner-meeing saemens on he volailiy of ineres raes, only in a few cases he BI s iner-meeing saemens significanly resul in higher ineres rae volailiy. The effecs of iner-meeing saemens by members of he BOT s board of governors in some cases are in line wih he direcion of he policy inclinaion. On he oher 18

20 hand, in all cases he volailiy of ineres raes in Thailand is no significanly affeced by he iner-meeing saemens. Finally, he implemenaion of inflaion argeing in Indonesia since July 5 has no seem o improve he effeciveness of moneary policy communicaion in moving ineres raes. Moneary policy saemens wih loose or neural policy inclinaion end o lower ineres rae while he saemens wih igh policy inclinaion do no have significan effecs. Moreover, alhough he effecs of iner-meeing saemens wih loose policy inclinaion remain negaive, in mos cases he effecs are insignifican. The main difference is he effecs of moneary policy saemens on he volailiy of ineres rae, in which he effecs of he saemens become more pronounced. In summary, he empirical findings in his paper show ha, o some exen, he BI s and he BOT s moneary policy saemens and iner-meeing saemens have been effecive in moving shor-erm ineres raes. Neverheless, here seems o be asymmery in he effeciveness of he saemens, in which he saemens wih loose policy inclinaion end o be more effecive compared o he saemens wih igh policy inclinaion. In he case of Thailand, neural policy inclinaion in moneary policy saemens is also quie effecive in he sense ha such saemens do no move ineres raes. The effeciveness of he communicaion seems o be associaed wih he consisency beween policy inclinaion in he saemens and he following change in policy rae. In his respec, he BOT s moneary policy saemens seem o be more consisen relaive o he BI s moneary policy saemens. 19

21 References: Blinder, Alan. (1998). Cenral Banking in Theory and Pracice. MIT Press: Cambridge. Blinder, Alan, Michael Ehrmann, Marcel Frazscher, Jacob De Haan, and David-Jan Jansen. (8). Cenral Bank Communicaion and Moneary Policy: A Survey of Theory and Evidence. NBER Working Paper, No Brooke, Marin, and Neil Cooper. (). Inferring Marke Ineres Rae Expecaions from Money Marke Raes. Bank of England Quarerly Bullein, November. Connolly, Ellis, and Marion Kohler. (4). News and Ineres Rae Expecaions: a Sudy of Six Cenral Banks. Reserve Bank of Ausralia Research Discussion Paper, 4-1. Cowan, Nelson. (1). The Magical Number 4 in Shor-Term Memory: a Reconsideraion of Menal Sorage Capaciy. Behavioral and Brain Sciences, 4, Demirlap, Selma, and Oscar Jorda. (). The Announcemen Effec: Evidence from Open Marke Desk Daa. Federal Reserve Bank of New York Economic Policy Review, 8, Ehrmann, Michael, and Marcel Frazscher. (7). Communicaion by Cenral Bank Commiee Members: Differen Sraegies, Same Effeciveness? Journal of Money, Credi and Banking, Vol.39, No-3. Filardo, Andrew, and Diwa Guinigundo. (8). Transparency and Communicaion in Moneary Policy: a Survey of Asian Cenral Bank. Mimeo. Fracasso, Andrea, Hans Genberg, and Charles Wyploz. 3. How Do Cenral Banks Wrie? Geneva Repors on he World Economy. Geneva/London: ICMB and CEPR. Garcia-Herrero, Alicia, and Eli M. Remolona. (8). Managing Expecaions by Words and Deeds: Moneary Policy in Asia and Pacific. Mimeo. Geeras, Pera.. Cenral Bank Transparency. The Economic Journal, 11(483): Goelom, Miranda. (7). Managing Moneary Policy in an Uncerain Environmen: he Challenge of Cenral Bank Communicaion in Indonesia. In Essays in Macroeconomic Policy: The Indonesian Experience. Jakara: Gramedia Pusaka Uama.

22 Guhrie, Graeme, and Julian Wrigh. (). Open Mouh Operaions. Journal of Moneary Economics, 46(), Kohn, Donald L., and Brian Sack. (4). Cenral Bank Talk: Does I Maer and Why? In Macroeconomics, Moneary Policy, and Financial Sabiliy. Oawa: Bank of Canada, Miller, George A. (1956). The Magical Number Seven, Plus or Minus Two: Some Limis on Our Capaciy for Processing Informaion. Psychological Review, 63, Morris, Sephen, and Hyun Song Shin. (). Social Value of Public Informaion. American Economic Review, 9(5), Nelson, Daniel. (1991). Condiional `Heeroscedasiciy in Asse Reurns: A New Approach. Economerica, 59, Rozkru, Marek, Krzyszof Rybiński, Lucyna Szaba, and Radoslaw Szwaja. (7). Ques for Cenral Bank Communicaion: Does I Pay o Be Talkaive? European Journal of Poliical Economy, 3, Svensson, Lars E.O. (6). Social Value of Public Informaion: Morris and Shin Is Acually Pro Transparency, No Con. American Economic Review, 96(1), Woodford, Michael. (5). Cenral Bank Communicaion and Policy Effeciveness. In The Greenspan Era: Lesson for he Fuure. Kansas Ciy: Federal Reserve Bank of Kansas Ciy,

23 Table 1: Summary Saisics of he Srucure of Moneary Policy Saemens Bank Indonesia Bank of Thailand Mean Min Max Mean Min Max Toal Number of Words Number of Paragraph Number of Words in Decision Paragraph Senence Lengh of Decision Paragraph Table : Keywords for Policy Inclinaion in he Moneary Policy Saemens Tigh Neural Loose Bank Indonesia: - guided owards cauious - mainain igh bias - o remain neural - unchanged sill provide simulus easing - allowed room for cauious easing - coninue igh bias - hold seady course in is igh bias - will be kep on rack - possibiliy o gradually lower - more measured, cauious acions - mainain presen course - maybe reduced furher - consisenly guide expecaions - calling for vigilance - will mainain pruden moneary policy - will remain on pruden rack - will remain building sabiliy - more aggressive reducion will be made - reinvigorae consumer& business - indicae room for furher reducion Bank of Thailand: - low level has become less necessary - would sand ready o respond - should coninue a an upward rend - Should be raised again - curren rae is appropriae - moneary policy could be eased Table 3: Summary Saisics of Policy Inclinaion in he Moneary Policy Saemens Moneary Policy Saemens Tigh Neural Loose Toal Bank Indonesia % 19% 3% 1% Bank of Thailand % 31% 16% 1%

24 Table 4: Keywords for Policy Inclinaion in he Iner-Meeing Saemens Tigh Neural Loose Bank Indonesia - will keep igh bias policy - no plan o raise - may be cu - raising ineres rae is possible - will keep - will cauiously cu - lowering BI rae is uncerain - will raise ineres rae - will ease moneary policy - coninue o increase ineres rae - will no impose igh policy - here is a room o furher cu - may raise ineres rae - will no cu - i is possible o lower - migh no raise - lower ineres rae rend - expecs he benchmark o go down - i makes sense o cu Bank of Thailand - will add ineres rae increase - oo soon o say furher rise - will remain low - could raise key rae - will seek o mainain balance - will keep low - needs o be increased - no need o push furher - would sar o ease keep increasing - won' be cu - will be furher cu - would have o rise - uncerain abou decision on rae - may cu benchmark rae - would coninue upward rend - cu unlikely - he increase will help - migh no have o rise - wans higher rae - migh have o go up Table 5: Summary Saisics of Policy Inclinaion in he Iner-Meeing Saemens Bank Indonesia Bank of Thailand Tigh Neural Loose Toal Tigh Neural Loose Toal Governor % 17% 61% 1% 43% 43% 14% 1% Senior Depuy Governor % 5% 33% 1% Depuy Governors % 36% 55% 1% 67% 17% 17% 1% Toal % 34% 49% 1% 48% 37% 15% 1% 3

25 Table 6: Esimaion Resuls: Mean Equaion Moneary Policy Saemen Iner-Meeing Saemen Tigh Neural Loose Tigh Neural Loose Bank Indonesia: 1-monh Inerbank *** -.3*** ** (-.66) (-4.79) (-3.3) (-.16) (-1.7) (-.7) 3-monh Inerbank. -.48*** -.49*** ** (.1) (-4.31) (-.79) (1.63) (-1.1) (-.38) 6-monh Inerbank ** -.47*** ** (.48) (-.5) (-.61) (.68) (-.34) (-.7) 1-monh Inerbank.1 -.4** -.59*** *** (.16) (-.48) (-3.1) (.18) (.93) (-.1) 3-monh Forward (.49) (-.49) (.53) (.3) (.3) (-1.43) 6-monh Forward ** (-.95 (-1.17) (-.1) (.8) (1.1) (.45) Bank of Thailand: 1-monh Inerbank ***.4*** -.** -.14*** (-1.1) (.6) (-8.91) (3.8) (-.) (-8.41) 3-monh Inerbank.19*** ***..1.1*** (.94) (-.77) (-8.8) (-.54) (.8) (.96) 6-monh Inerbank (-.1) (-.39) (-.9) (.64) (1.41) (-.55) 1-monh Inerbank -.11*** ***. -.3**.4 (-3.94) (.81) (-3.34) (.11) (-.55) (.8) 3-monh Forward...7.4**. -.1** (.3) (1.9) (.8) (1.97) (.3) (.31) 6-monh Forward **..5 (.41) (.18) (-.89) (.3) (-.6) (.99) Noes: Numbers in he brackes are sandard error; ***,**,*) indicae significance a he 99%, 95%, and 9% levels. Sample for Indonesia runs from 1/1/4 o 1/31/7, and sample for Thailand runs from 6/8/5 o 1/31/7. 4