Money Multiplier under Reserve Option Mechanism

Transcription

1 MPRA Munch Personal RePEc Archve Money Multpler under Reserve Opton Mechansm Halt Akturk and Hasan Gocen and Suleyman Duran Melksah Unversty June 2015 Onlne at MPRA Paper No , posted 5. June :42 UTC

2 Money Multpler under Reserve Opton Mechansm * Halt Aktürk Hasan Göcen Süleyman Duran Department of Economcs Melkşah Unversty Kayser, TURKEY June 2015 Abstract Ths paper ntroduces a generalzed money (M2) multpler formula to the lterature for a monetary system wth Reserve Opton Mechansm (). Varous features of the proposed multpler are then explored usng monthly Turksh data durng the decade 2005 to We report a step ncrease n the magntude and a slght upward adjustment n the long-run trend of the multpler wth the adopton of. We provde evdence for substantal change n the seasonal pattern of the multpler, cash rato, requred and excess reserves under. We show that money (M2) multpler s less volatle n a monetary system wth and dscuss the subsequent stablzng nfluence of more predctable multpler on the foregn exchange market. Keywords: Money multpler, macroprudental polcy, reserve opton mechansm, reserve requrements, fnancal stablty JEL Classfcaton: E51; E52; E58; F31 * Halt Aktürk and Süleyman Duran gratefully acknowledge the fnancal support of the Scentfc and Technologcal Research Councl of Turkey (TUBITAK) under SOBAG Grant No. 114K540. The authors reman responsble for all remanng errors. 1

3 I. Introducton Captal flows, measured by percentage share of GDP, to emergng markets and majorty of developng economes have ncreased more than twofold on average durng the fve years precedng global fnancal crss of (Gallagher & Tan, 2014). Soarng volatlty and sudden stops are consdered as two fundamental elements of captal flows to emergng markets durng and aftermath of global fnancal crss. Surges n nternatonal captal flows and fnancal stablty concerns led many emergng market central banks to adopt alternatve macroprudental polcy nstruments lke reserve requrements and captal requrements. Reserve requrement ratos have been mportant monetary polcy nstruments for many central banks. Federco, Vegh, and Vuletn 2014 reports around two thrds of developng economes have used reserve requrement polces for macroeconomc stablzaton purposes n comparson to none ndustralzed country snce In ths research, we focus on Reserve Opton Mechansm whch we consder as a modern verson of reserve requrement polces. Wth macroeconomc and fnancal stablty n mnd, the Central Bank of the Republc of Turkey (CBRT) developed Reserve Opton Mechansm () an unconventonal reserve requrement polcy amng to reduce adverse effects of short term captal flows and exchange rate volatlty aftermath of global fnancal crss of s a mechansm that allows commercal banks to keep a certan percentage of ther Turksh lra requred reserves n foregn currences: USD and Euro, and gold 1. s consdered actng as an automatc stablzer n the face of external 1 Foregn currences that can be mantaned for Turksh lra requred reserves have been lmted to USD only as of August

4 9/16/2011 1/16/2012 5/16/2012 9/16/2012 1/16/2013 5/16/2013 9/16/2013 1/16/2014 5/16/2014 9/16/2014 1/16/2015 Percent (%) Percent (%) mbalances and hence t promses to be a potentally useful reserve requrement polcy nstrument for macroeconomc and fnancal stablty (Alper, Kara, & Yorukoglu, 2013). In ths paper, we study the behavor of money multpler n a monetary system where Reserve Opton Mechansm s effectvely used as a polcy nstrument. To ths end, we analytcally derve a generalzed money (M2) multpler formulae and assess the mpact of Reserve Opton Mechansm on the multpler and ts components. Usng Turksh data, we show that money (M2) multpler ncreased n a stepwse fashon due to ntroducton of. We dscuss how the rse n the value of the multpler renforces the role of as an automatc stablzng polcy tool. In addton, we study seasonal patterns and stablty of money (M2) multpler and ts components. We report that seasonal patterns of money (M2) multpler has changed and money (M2) multpler became sgnfcantly less volatle wth the ncepton of faclty to the monetary system. Fgure 1. FX Reserve Opton Mechansm Utlzaton Rates Notes: Data has bweekly frequency (per-mantenance perod) coverng the perod from 9/30/2011 to 3/27/2015. The dagram shows the FX Reserve Opton Mechansm utlzaton rates for the bankng sector. Source: Electronc Data Delvery System (EDDS) of the Central Bank of Republc of Turkey. 3

5 Reserve Opton Mechansm () conssts two prncpal elements: reserve opton rato (ROR) and reserve opton coeffcents (ROCs) whch are both set by the central bank. The upper lmt for the proporton of domestc currency requred reserves that s allowed to be held n foregn currency (FX) and gold s governed by reserve opton rato (ROR). The amount of FX and gold that can be held aganst per domestc currency requred reserves s called the reserve opton coeffcent (ROC). Our focus n ths research s on FX opton of due to the fact that Turksh bankng sector has been utlzng FX opton under ntensvely as depcted by Fgure 1. Due to Turksh bankng sector s approach to FX opton, the composton of requred reserves held at the central bank has changed dramatcally as shown by Fgure 2. Fgure 2. Requred Reserves to be mantaned for TL Labltes under /30/ /30/2011 1/31/2012 3/31/2012 Bllon USD 5/31/2012 7/31/2012 9/30/ /30/2012 1/31/2013 3/31/2013 5/31/2013 7/31/2013 9/30/ /30/2013 1/31/2014 3/31/2014 5/31/2014 7/31/2014 9/30/ /30/2014 1/31/2015 Turksh Lra FX (USD and Euro) Gold Notes: Data has bweekly frequency (per-mantenance perod) coverng the perod from 9/30/2011 to 3/27/2015. Source: Electronc Data Delvery System (EDDS) of the Central Bank of Republc of Turkey. The Turksh central bank has been employng as an alternatve macroprudental monetary polcy tool qute actvely. Fgure 3 presents reserve opton ratos and the correspondng FX reserve opton coeffcents snce the ntroducton of n Sept The dagram shows that there has been constant amendments n both reserve opton ratos and the correspondng FX reserve opton coeffcents untl Mar

6 Fgure 3. Evoluton of Reserve Opton Ratos (ROR) and correspondng FX Reserve Opton Coeffcents (ROCs) Reserve Opton Rato (ROR, %) Sep Oct Nov Jun Jul Aug Aug Oct Nov Dec Reserve Opton Coeffcent (ROC) Apr May Jun Jan Feb Realzed Average FX ROCs (Rght Scale) Notes: Data has bweekly frequency (per requred reserves mantenance perod) coverng the perod from 9/30/2011 to 3/27/2015. The fgures nsde the stacked bar chart show effectve foregn exchange (FX) Reserve Opton Coeffcents (ROCs) and the correspondng tranches as of the gven dates (requred reserves mantenance perods) on the horzontal axs. For nstance, effectve FX ROCs as of Aug are 1.1, 1.4, 1.7, 1.9, and 2.0 for the correspondng tranches of 0-30, 30-35, 35-40, 40-45, 45-50, 50-55, and percent. The ROCs 3.30, 3.50, 3.70, 3.90, and 4.10 located at the top rght corner of the dagram correspond the tranches of 55-56, 56-57, 57-58, 58-59, and percent, respectvely. The sold black lne s obtaned by connectng ffteen realzed average FX ROC data ponts for the bankng sector for the correspondng ffteen dates (requred reserves mantenance perods) on the horzontal axs. Realzed average FX ROCs are obtaned by takng the rato of requred reserves kept n FX to total requred reserves multpled by the percentage use of FX faclty. Source: Electronc Data Delvery System (EDDS) of the Central Bank of Republc of Turkey. II. Lterature Revew There s a long and establshed lterature on money multpler that lnks reserve money to supply of broad money. In ths tradton, monetary base s vewed as an mportant monetary polcy nstrument through whch central banks control the money supply. In modern economcs, though, the use offcal nterest rates s consdered as the man 5

7 polcy nstrument rather than reserve requrements. Havng sad that, monetary authortes n emergng economes stll use reserve requrements extensvely as an accompanyng monetary polcy nstrument. In contrast to several work on money multpler, studes on Reserve Opton Mechansm () are not much whle the number of studes are on the rse. Alper et al., 2013 ntroduce Reserve Opton Mechansm and explan the way ths mechansm functon through banks decson makng on reserves. Küçüksaraç and Özel, 2012 compute optmal Reserve Opton Coeffcents (ROCs), mportant elements of faclty, where they provde evdence for hgh senstvty of breakeven ROCs to funds denomnated n foregn currency. Değerl and Fendoğlu, 2015 study the mpact of n reducng the volatlty of foregn exchange expectatons. Oduncu, Akçelk, and Ermşoğlu, 2013 show that s an effectve polcy tool n reducng volatlty of Turksh lra. Ths paper touches upon two lnes of work on money multpler and Reserve Opton Mechansm. There are two recent and closely related studes to current research on. Aslaner, Çıplak, Kara, and Küçüksaraç, 2014 use bank level data where they show varatons n banks utlzaton rates are manly explaned by varatons n short term domestc nterest rates. Hence, they argue, counter actve polcy responses to captal flows, hgher/lower nterest rates durng outflows/nflows, undermne automatc stablzng feature of the faclty. Ths paper shows that domestc short term nterest rates automatcally re-adjust upward/downward to some extent durng captal outflows/nflows va changes n money (M2) multpler whch n turn depends on banks utlzaton rate of the faclty. Hence, excessve counter actve nterest rate polcy responses may reduce automatc stablzng ablty of the faclty rather than the conduct of nterest rate polcy tself n response to captal flows. 6

8 Unlke Jayaratne and Morgan s, 2000 fndng on lmted bank lendng channel of monetary polcy, Alper, Bnc, Demralp, Kara, and Özlü, 2014 fnd banks lendng behavor respond sgnfcantly to reserve requrement nduced monetary polcy. Alper et al., 2014 produce cost based effectve reserve requrement rato whch negatvely depends on the banks utlzaton rate of faclty as long as banks fnd FX borrowng s less costly than that of n domestc currency. They argue short term borrowng from the central bank s not a close substtute for deposts for banks. Therefore, banks lendng behavor respond sgnfcantly to reserve requrements and the ntroducton of faclty. Our research shows that, snce reserve requrements under nfluence banks lendng behavor, broad money supply must be lnked to reserve requrements va a modfed money multpler under. III. Data descrpton and prelmnary emprcs Ths study uses monthly frequency data from Turkey coverng the perod of December 2005 to February All data s secondary and gathered from Electronc Data Delvery System (EDDS) of the Central Bank of Republc of Turkey. Data table n Appendx I reports raw data used to produce dagrams, obtan summary statstcs and estmates, and conduct varous statstcal tests. We adopt IMF defnton of reserve money (monetary base MB) where MB equals to the sum of currency n crculaton C, total domestc currency reserves R, and demand deposts of banks held at the Central Bank. In the data table n Appendx I, there s neglgble amount of dscrepancy between the monetary base MB and the sum of currency n crculaton C and total domestc currency reserves R s due to deposts of banks held at the Central Bank. Followng the standard defnton, money supply (M2) 2 Unless t s stated otherwse underneath fgures, tables etc. gven n the text. 7

9 s a broad money measure comprsng domestc currency n crculaton, checkable and tme deposts (excludng foregn currency deposts) held by the publc at commercal banks and other depostory nsttutons. Tme deposts TD conssts of one-month, threemonths, sx-months, one-year and longer deposts. Common reserve requrement ratos rr 1 and rr 2 apples to domestc currency checkable deposts and tme deposts, respectvely untl January 6, Tme deposts reserve requrement rato rr 2 s the weghted average of the requred ratos for one-month, three-months, sx-months, oneyear and longer deposts. Monthly nterbank money market rates r t are calculated by takng smple average of the daly weghted averages of nterbank overnght (O/N) rates. Fgure 4. Realzed Average FX ROCs and FX Rates 3.9 Realzed Average FX ROCs FX Rate (Turksh Lra/USD) /30/ /30/2011 1/31/2012 3/31/2012 5/31/2012 7/31/2012 9/30/ /30/2012 1/31/2013 3/31/2013 5/31/2013 7/31/2013 9/30/ /30/2013 1/31/2014 3/31/2014 5/31/2014 7/31/2014 9/30/ /30/2014 1/31/2015 Notes: Data has bweekly frequency (per requred reserves mantenance perod) coverng the perod from 9/30/2011 to 3/27/2015. The black and red lnes show realzed average FX ROCs for the entre bankng sector and 14-day smple average FX rates (Turksh lra per USD). Realzed average FX ROCs are obtaned by takng the rato of requred reserves kept n FX to total requred reserves multpled by the percentage use of FX faclty. Source: Electronc Data Delvery System (EDDS) of the Central Bank of Republc of Turkey. 3 In Press Release No of Dec. 17, 2010, the CBRT announced the Turksh lra requred reserve rato would be dfferentated accordng to the maturty structure of deposts startng January

10 The model n Secton IV shows that threshold reserve optons coeffcents depend on spot exchange rates. To confgure these relatonshp, we use daly nomnal spot exchange rate e t (domestc currency per USD) to obtan 14-day smple averages and monthly rates 4. The correlaton coeffcent between realzed average FX reserve opton coeffcents (ROCs) and nomnal spot exchange rates s about as llustrated by Fgure 4. IV. The model The exstng lterature lnks money supply MS to monetary base MB through money multpler m. Ths relatonshp s usually depcted as followng. MS t = m t MB t (1) The total amount of reserves R n the bankng system equals to the sum of total requred reserves RR and excess reserves ER.e. R t = RR t + =1 =1 ER t where RR t and ER t stand for requred and excess reserves held by bank at tme perod t. Accordng to ths formulaton N banks operate n the bankng sector. Assume reserve requrement rato on domestc currency checkable deposts D t s rr 1 and on tme deposts TD t s rr 2 where rr 2 < rr 1 < 1. Note that how much domestc currency checkable vs. tme deposts to be held by publc at commercal banks and other depostory nsttutons would depend on market nterest rate r. We assume the publc hold a fracton of ts domestc currency checkable deposts as tme deposts.e. =N =N =1 TD t = ρ(r) =1 D t and the publc tend to utlze domestc currency tme deposts more as market nterest rses.e. ρ (r) > 0. Hence, n a monetary system wthout =N =N 4 The labltes subject to reserve requrements are calculated on Frday every two weeks n Turksh bankng sector.e. bweekly requred reserves mantenance perod apples. 9

11 Reserve Optons Mechansm (), total requred reserves (n terms of domestc currency) n the bankng system equals to =N =N RR t = rr 1 D t + rr 2 TD t = =1 =1 =N [rr 1 + ρ(r)rr 2 ] D t =1 (2) Suppose banks hold k 1 (r) porton of checkable deposts and k 2 (r) porton of tme deposts as excess reserves. We assume precautonary demand for excess reserves for checkable deposts s greater than that of tme deposts.e. k 2 (r) < k 1 (r). Snce excess reserves don t earn any nterest, excess reserve demand for any gven level of checkable and tme deposts s negatvely related to market nterest rates k 1 (r) < 0, k 2 (r) < 0. Hence, the total excess reserves n the bankng system s gven by =N =N =N ER t = k 1 (r) D t + k 2 (r) TD t = [k 1 (r) + ρ(r)k 2 (r)] D t =1 =1 =1 Puttng equatons (2) and (3) gve total reserves R (n domestc currency (3) equvalent unts) n the bankng system as descrbed n equaton 4 below. R t = {rr 1 + ρ(r)rr 2 + k 1 (r) + ρ(r)k 2 (r)} D t (4) In a monetary system wth Reserve Opton Mechansm the banks are allowed to keep up to reserve opton rato (ROR) of ther domestc currency requred reserves n foregn exchange (FX) and/or gold. Hence, total requred reserves held by the banks wshng to utlze the faclty can be wrtten as =N =1 =K =K RR t = [rr 1 + ρ(r)rr 2 ] D t (1 ROR t ) + [rr 1 + ρ(r)rr 2 ] D t ROR t ROC(ROR t ) =1 =1 (5) In equaton 5, the frst term stands for the total requred reserves held n domestc currency whereas the second term shows the total foregn currency or gold held by the banks aganst domestc currency requred reserves (n domestc currency 10

12 equvalent unts). Accordng to ths formulaton ROR t stands for utlzaton rate of by bank at tme perod t, and ROC(ROR t ) s the correspondng reserve opton coeffcent as set by the Central Bank. Any bank decdng on ROR t at any tme perod t automatcally determnes whch trench of ROC t must use. Ths formulaton assumes K N banks utlze faclty n the bankng sector. Usng equaton 5 along wth equatons 3 and 4 gve us total reserves at tme perod t n a monetary system wth as descrbed by equaton 6. =K =N R t = ([rr 1 + ρ(r)rr 2 ] D t (1 ROR t ) + [rr 1 + ρ(r)rr 2 ] D t =1 =K+1 =N + [k 1 (r) + ρ(r)k 2 (r)] D t ) =1 (6) =K + {[rr 1 + ρ(r)rr 2 ] D t ROR t ROC(ROR t )} =1 In equaton 6, the frst term nsde large parentheses and the second term nsde large curly bracket stand for reserves held n domestc currency and reserves held n foregn currency (n domestc currency equvalent unts). Accordng to equaton 6, K banks utlze faclty whereas N K banks don t. Because the monetary base MB equals to the sum of total domestc currency reserves and currency n crculaton C, we generate equaton 7 lnkng the monetary base to checkable deposts and currency n crculaton by droppng the term nsde large curly bracket n equaton 6 to obtan: =K =N MB t = [rr 1 + ρ(r)rr 2 ] D t (1 ROR t ) + [rr 1 + ρ(r)rr 2 ] D t =1 =K+1 =N + [k 1 (r) + ρ(r)k 2 (r) + c t ] D t =1 (7) 11

13 where we assume non-bank publc holds fracton c of the total amount of checkable deposts as currency.e. C t = c t =N =1 D t. Insertng equaton 7 nto equaton 1 yelds equaton 8 lnkng money supply MS to checkable deposts va money (M2) multpler m n a monetary system where faclty s n use. =K MS t = m 2,t [[rr 1 + ρ(r)rr 2 ] D t (1 ROR t ) =1 =N + [rr 1 + ρ(r)rr 2 ] D t + [k 1 (r) + ρ(r)k 2 (r) + c t ] D t ] =K+1 =N =1 (8) Followng the standard defnton, money supply (M2) s a broad money measure comprsng domestc currency n crculaton, checkable and tme deposts (excludng foregn currency deposts) held by the publc at commercal banks and other depostory nsttutons. Accordng to ths defnton money (M2) supply can be wrtten as followng. =N =N =N MS t = D t + TD t + c t D t =1 =1 =1 =N = [1 + ρ(r) + c t ] D t =1 (9) Insertng equaton 9 nto 8 and solvng for m 2,t produces money (M2) multpler under as summarzed by equaton 10. m 2,t = 1 + ρ(r) + c t [rr 1 + rr 2 ρ(r)] {1 =K D t =1 ROR t (10) =N } + [k 1 (r) + k 2 (r)ρ(r)] + c t =1 D t Accordng to equaton 10, money (M2) multpler under s a functon of the currency rato c and ρ(r) chosen by depostors, the excess reserve rato k 1 (r) + k 2 (r)ρ(r), the number of banks K utlzng, and the utlzaton rate of reserve 12

14 opton mechansm ROR t optmzed by the banks, and, fnally, the requred reserve ratos rr 1 and rr 2 set by the monetary authorty 5. Note that money (M2) multpler under turns nto a standard money (M2) multpler m 2 as n Goodfrend, 1982 gven n equaton 11 n the absence of faclty (.e. ROR t = 0 ) and/or when no bank utlzes faclty (.e. K = 0). m 2,t = 1 + ρ(r) + c t rr 1 + rr 2 ρ(r) + [k 1 (r) + k 2 (r)ρ(r)] + c t (11) It s straghtforward to show that money (M2) multpler under m 2,t s at least as large as the standard money multpler m 2,t n the absence of faclty due to the =K fact that D =1 t ROR t 0, rr 1 >, rr 2 > 0 and ρ(r) > 0. Accordng to equaton 10, calculatng M2 multpler requres faclty utlzaton rates at bank level. Snce banks may consder utlzaton rates as commercal secret, such bank level data s not publcally avalable. However, ths doesn t mean the formulae n equaton 10 can t be evaluated. Suppose there s at least a bank utlzng the faclty (.e. K > 0), and the utlzaton rate of by banks s common at any tme perod t (.e. ROR t = ROR t > 0 = 1, 2, K). Then, the M2 multpler under turns nto formulae as n equaton 12. m 1 + ρ(r) + c t 2,t = (12) [rr 1 + rr 2 ρ(r)]{1 ROR t θ t } + [k 1 (r) + k 2 (r)ρ(r)] + c t where θ t s the share of total domestc currency checkable deposts at the banks utlzng the faclty at tme perod t.e. θ t = =K =1 D t =N D =1 t. Accordng to ths 5 Note that banks use of reserve opton mechansm ROR must le wthn the lmts as set by monetary authorty. 13

15 formulaton, ceters parbus, the hgher the share of domestc currency checkable deposts (or requred reserves to be mantaned) at the banks utlzng the faclty (hgher θ), the larger the M2 multpler. Each bank s utlzaton rate of the faclty depends on the relatve cost of acqurng domestc currency and/or FX fundng and other bank specfc factors. Suppose all banks have the same level of ablty obtanng domestc currency fundng and/or FX fundng.e. there s not any credt constrant and no bank specfc factors exst. When bank decdes to fulfll domestc currency requred reserves wthout utlzng the faclty, the total cost of acqurng domestc currency fundng at tme perod t equals to [rr 1 D t + rr 2 TD t ] r t. When bank decdes to fulfll domestc currency requred reserves utlzng faclty at a rate of ROR t, the total cost of acqurng fundng at tme perod t equals to the cost of domestc currency fundng (1 ROR t )[rr 1 D t + rr 2 TD t ] r t plus the cost of acqurng FX fundng ROR t ROC(ROR t ) [rr 1 D t + rr 2 TD t ] r t E t [e t+1 ] where r t s domestc market nterest rate, r t s foregn market nterest rate, e t s nomnal spot exchange rate (domestc currency per USD) at tme perod t, and E t s the expectaton operator condtonal on the nformaton observed through tme perod t. The term ROC(ROR t ) stands for the correspondng trench of reserve optons coeffcent based on banks choce of utlzaton rate of the faclty at tme perod t. The bank optmzes the utlzaton rate ROR t at any tme perod t where the assocated costs for utlzaton and non-utlzaton of the faclty are equal. Hence, the threshold tr reserve optons coeffcent at any tme perod t ROC t can be expressed as followng e t 14

16 ROC t tr = r t r t E t[e t+1 ] e t (13) Note that each bank s threshold reserve optons coeffcent may dffer due to each bank s utlzaton rate of the faclty ROR whch n turn s determned by solvng each bank s proft maxmzaton problem consderng the relatve cost of domestc and FX fundng as descrbed n equaton (13) plus credt constrants and other bank specfc factors. Our focus n ths paper s not on banks ndvdual proft maxmzaton behavor towards the use of faclty. Comparatve statcs for M2 multpler under :. Changes n the requred reserve ratos rr 1 and/or rr 2 : When requred reserve ratos rr 1 and/or rr 2 rse/fall, m 2,t falls/rses.e. rr 1 and/or rr 2 and m 2,t are negatvely related. m 2,t < 0; rr 1 m 2,t < 0 rr 2. Changes n the currency deposts rato c: When currency deposts rato c t rses/falls, m 2,t falls/rses.e. c t and m 2,t are negatvely related. m 2,t < 0 c t. Changes n elements of excess reserve rato k 1 (r) and/or k 2 (r): When k 1 (r) and/or k 2 (r) rse/fall, m 2,t falls/rses.e. k 1 (r) and/or k 2 (r)and m 2,t are negatvely related. m 2,t < 0; k 1 m 2,t k 2 < 0 15

17 v. Changes n the number of banks utlzng the faclty K: When the number of banks K utlzng the faclty rses/falls, m 2,t rses/falls.e. K and m 2,t are postvely related. m 2,t K > 0 v. Changes n the banks utlzaton rate of the reserve optons mechansm: When the banks utlzaton rate ROR of the faclty rses/falls, m 2,t rses/falls.e. ROR and m 2,t are postvely related. m 2,t ROR > 0 v. Changes n the domestc money market nterest rates r: When money market nterest rates rses/falls, m 2,t rses/falls.e. r and m 2,t are postvely related. Usng M2 multpler formulae n equatons 10, 11, and 12, partal dervatves descrbed by equatons A1, A2, and A3 n Appendx II prove that m 2,t > 0 r The ratonale behnd ths fndng s smple. When r rses t s more costly to fnd domestc currency fundng than FX fundng and therefore banks start utlzng the faclty more ntensely (gven that the bank s not already utlzng the faclty fully).e. ROR t rses when r rses and ths would lead to an ncrease n threshold reserve opton coeffcent ROC t tr and fnally the utlzaton rate rses snce reserve opton rato and reserve opton coeffcent are postvely related. Accordng to equatons 10 and 11, money (M2) multpler n a monetary system wth s more senstve to the domestc market nterest rate changes as shown by equatons A4 and A4 n Appendx II. 16

18 Dec-05 Apr-06 Aug-06 Dec-06 Apr-07 Aug-07 Dec-07 Apr-08 Aug-08 Dec-08 Apr-09 Aug-09 Dec-09 Apr-10 Aug-10 Dec-10 Apr-11 Aug-11 Dec-11 Apr-12 Aug-12 Dec-12 Apr-13 Aug-13 Dec-13 Apr-14 Aug-14 Dec-14 V. Emprcal analyss and results We frst calculate M2 multpler usng Turksh data from Dec to Feb and llustrate our computatons. Fgure 5 plots M2 multpler under and w/o aganst the log of monetary aggregates: monetary base and M2 along wth market nterest rates. The fgure reveals three key features. Frst, there s a Fgure 5. Money (M2) multpler, Monetary aggregates and nterest rates M2 multpler under Log of MB (Rght scale) Log of nterbank overnght rates (Rght scale) M2 multpler w/o Log of M2 (Rght scale) Notes: Data has monthly frequency coverng the perod from December 2005 to February M2 multpler under s calculated assumng θ = 0.90 due to Aslaner et al., 2014 whch reports that the share of total domestc currency checkable deposts (or reserve requrements to be mantaned) at the banks utlzng the faclty consttutes 91% of the bankng sector. M2 multpler w/o s calculated by settng θ = 0. sharp drop and a subsequent sgnfcant rse n M2 multpler durng Second, M2 multpler under s sgnfcantly larger. Thrd, a decsve upward trend n M2 17

19 Dec-05 May-06 Oct-06 Mar-07 Aug-07 Jan-08 Jun-08 Nov-08 Apr-09 Sep-09 Feb-10 Jul-10 Dec-10 May-11 Oct-11 Mar-12 Aug-12 Jan-13 Jun-13 Nov-13 Apr-14 Sep-14 Feb-15 Percent (%) Percent (%) multpler become evdent wth the ncepton of to the monetary system. The sharp drop and the subsequent rse n M2 multpler durng Dec to Sept seem to reflect changes n requred reserve ratos rather than changes n other components of M2 multpler. In Dec. 17, 2010, the CBRT announced Turksh lra requred reserve rato would be dfferentated accordng to the maturty structure of deposts startng Jan In addton, sgnfcant ncreases were ntroduced n the requred reserves ratos. From Dec to Sept. 2011, common reserve requrement rato sharply rose from 5.5 to 16 and about 13 percent on checkable and tme deposts, respectvely. The rates were re-adjusted agan n Oct where requred reserves ratos were set to about 11 percent. Fgure 6 shows the tme path of requred reserve ratos on checkable and tme deposts. Fgure 6. Reserve requrement ratos Tme deposts RRR Checkable deposts RRR Notes: Data has monthly frequency coverng the perod from December 2005 to February Tme deposts reserve requrement rato (RRR) s the weghted average of the requred ratos for one-month, three-months, sx-months, one-year and longer deposts snce The sharp drop n M2 multpler durng Dec to Sept s manly reflected by step ncrease n the reserve requrement ratos as ndcated n Fgure 6. Subsequently, 18

20 when the Central Bank set requred reserves ratos back to about 11 percent, M2 multpler pcked back up from a bottom of 2.5 to 3.5 as ndcated n Fgure 5. Accordng to Fgure 5, average M2 multpler under and w/o s about 4.3 vs. 3.6 durng Sept to Feb. 2015, respectvely. The p-value assocated wth Student s pared t-test wth one taled dstrbuton s ndcatng sgnfcantly larger M2 multpler under. There s a defnte upward trend n M2 multpler along wth monetary base and M2. Average slope of upward trend n M2 multpler under equals to 0.24 whereas 0.17 for that of w/o. The rse n the sze and altered trend behavor of M2 multpler seem to reflect the ntroducton of faclty to the monetary system and the changes t trggered n the components of M2 multpler. Fgure 7 gves a vsual summary of the underlyng components of M2 multpler under and how they change over tme. There s steady long-run declnng trend n requred reserves rato (as descrbed by [rr 1 + rr 2 ρ(r)]{1 ROR t θ t }), excess reserves rato, and currency depost rato c from Dec and onwards except for the perod of Apr to Oct. 2011, the perod wth sharp drop and a subsequent rse n M2 multpler. The declnng trend n requred and excess reserves ratos seem to be renforced by the ntroducton of around Sept Comparatve statstcs,, and v predct negatve relatonshp between M2 multpler and currency depost rato, excess reserves rato, and utlzaton rate. The data suggests that the slght upward trend found n M2 multpler around Dec to Aug s due to slght declnng trend n currency depost rato, excess reserves and requred reserves ratos (as descrbed by [rr 1 + rr 2 ρ(r)]) w/o. M2 multpler rses even further due to large decreases n excess and requred reserves ratos around Sept. 2011, the date for ncepton of to the monetary system, and onwards. Requred reserves rato refers to [rr 1 + rr 2 ρ(r)]{1 ROR t θ t } aftermath of Sept by the ntroducton of 19

21 . Hence, the decrease n requred reserves rato could be due to changes n [rr 1 + rr 2 ρ(r)] and/or changes n utlzaton rate.e. n {1 ROR t θ t }. We fnd about 60 percent of the ncrease n the multpler s due to the ncepton of faclty and ts utlzaton by banks. Fgure 7. Money (M2) multpler components Dec-05 Apr-06 Aug-06 Dec-06 Apr-07 Aug-07 Dec-07 Apr-08 Aug-08 Dec-08 Apr-09 Aug-09 Dec-09 Apr-10 Aug-10 Dec-10 Apr-11 Aug-11 Dec-11 Apr-12 Aug-12 Dec-12 Apr-13 Aug-13 Dec-13 Apr-14 Aug-14 Dec Excess reserves rato Currency depost rato Requred reserves rato Notes: Data has monthly frequency coverng the perod from December 2005 to February Excess reserves and currency depost ratos refers to total excess reserves dvded by total checkable deposts and currency n crculaton dvded by total checkable deposts, respectvely. Requred reserves rato refers to [rr 1 + rr 2 ρ(r)]{1 ROR t θ t }. Table 1 gves estmated long-run trends n M2 multpler under and ts components. The estmates provde evdence for sgnfcant shfts n the long-behavor of M2 multpler, requred and excess reserves wth the ntroducton of to the monetary system. Trend estmates for M2 multpler rse from 0.13 to 0.61 percent when the perod of estmaton s narrowed from full sample of 2005m m2 to 2011m9-2015m2 n whch s n effectve use. Abrupt declne n M2 multpler durng Dec to Oct s due to sgnfcant upward shfts n excess and requred reserves as shown n Fgure 6. Sgnfcant rse n excess reserves seem to reflect cost effcency as ndcated by sgnfcant drops n market nterest rates depcted n Fgure 5. 20

22 Abrupt changes n requred reserves, on the other hand, reflect pure updates n the reserve requrement ratos as dscussed above. The evoluton of requred and excess reserves seres can be characterzed by two dstnct phases. Trend estmate for requred reserves rses from to 1.61, whereas excess reserves ratos decreases from to percent when the perod of estmaton s narrowed from 2005m m2 to 2011m9-2015m2. The percentage change s trend estmate for currency n crculaton s about 20 percent, a relatvely small drop n c from to percent when entre perod of 2005m m2 s narrowed to 2011m9-2015m2. Theoretcally, decreases n requred reserves and currency n crculaton ratos ncreases M2 multpler whereas rse n excess reserves decreases the multpler. Table 1. Estmated long-run trend n M2 multpler and ts components Trend Perod & Varable (X) t-statstcs p-value per-month Adjusted R-squared 2005m m2 M2 multpler 0.13% Currency n crculaton rato -0.17% Requred reserves rato -0.14% Excess reserves rato -0.80% Tme-Checkable deposts rato -0.24% m m8 M2 multpler -0.15% Currency n crculaton rato -0.07% Requred reserves rato 0.46% Excess reserves rato 0.32% Tme-Checkable deposts rato 0.00% m9 2015m2 M2 multpler 0.61% Currency n crculaton rato -0.20% Requred reserves rato -1.61% Excess reserves rato -2.40% Tme-Checkable deposts rato -0.71% Notes: Trend s estmated from an OLS regresson where log of X s regressed on an ntercept, a lnear tme trend and eleven monthly seasonal dummes. 21

23 Hence, the data suggests that the rse n M2 multpler manly come from the changes n banks behavor wth respect to requred reserves whch n turn are nfluenced by ntroducton of the faclty the monetary system. Takng log and frst dfferencng equaton 1 of the model yelds month-to-month percentage growth n M2 that can be decomposed nto two factors: percentage growth n monetary base and M2 multpler. Table 2 reports summary statstcs for M2, monetary base MB, and the M2 multpler over the full perod from Dec to Feb as well as for two sub-perods wth and w/o. Table 2. The stablty of M2 multpler Average monthly Average monthly varance and covarance growth rates M2 MB m 2 Var(M2) Var(MB) Var(m 2 ) Cov(m 2, MB) 2005m m2 1.34% 1.71% 0.21% m m8 1.62% 1.92% -0.56% m9-2015m2 0.88% 1.36% 1.47% Notes: ^ refers to month-to-month percentage change as calculated by takng the frst log dfference of the varables. Confrmng the graphcal evdence ndcated n Fgure 5, Table 2 reports that monthly average growth rates n monetary aggregates of M2 and MB, and M2 multpler under are postve over the entre perod of Dec to Feb When the full sample s splt nto two sub-perods based on avalablty of faclty, the average growth rate n the multpler changes from to 1.47, respectvely. Average monthly growth rate of M2 multpler changes to from to 0.28 even when the perod of Jan to Aug wth large swngs s dropped from the perod of Dec to Aug Ths fndng confrms M2 multpler under has a larger trend component than that of w/o. The rght hand panel n Table 2 gve varance and covarance estmates to study short-run dynamcs n monetary aggregates and the M2 multpler. Frst, no matter the sze of the sample, the varance of M2 s lower than that of MB ndcatng M2 beng 22

24 Percent (%) more stable than s monetary base n the short-run. Accordng to Table 2, the varance on M2 multpler s reduced from 0.02 to 0.01 when the full sample s narrowed to Sept to Feb Ths fndng suggest that the volatlty n M2 multpler decreases wth the ntroducton of to the monetary system. Fgures 8A and 8B report strong seasonal pattern n M2 multpler under and w/o, requred and excess reserves ratos, and relatvely weaker seasonal pattern n currency n crculaton rato. Accordng to Fgure 8A, seasonal pattern s nfluenced by the ntroducton of faclty to the monetary system. Seasonal behavor durng the frst half of the year seem to become stronger wth the ntroducton of whereas the opposte s true for the second half of the year wth the excepton of November. M2 multpler rses sgnfcantly n January, drop abruptly n June and rebounds by the end of the year. Durng other months, wth the excepton of July and October, there don t seem to be a dstnct seasonal pattern n M2 multpler. Fgure 8A. Seasonal patterns n M2 multpler Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec M2 multpler wth M2 multpler w/o Notes: Data has monthly frequency coverng the perod from December 2005 to February The fgures show average monthly devatons from mean. 23

25 Currency n crculaton rato falls sharply n February, June, and December whle t rses sgnfcantly n August, October, and November as ndcated n Fgure 8B. The substantal ncrease n cash holdngs relatve to deposts n August seem to reflects crop purchases and harvest season. Requred reserves rato rse sgnfcantly n January and February and towards the end of the year whle t drops abruptly durng the March- July perod. A closer look at the seasonal behavor of M2 multpler n Fgure 8A and that of requred reserves n Fgure 8B reveals sgnfcant seasonal comovement between the two seres. In contrast, Fgure 8B provde evdence that excess reserves rato behave wth strong opposte seasonal pattern compared to requred reserves rato. Fgure 8B. Seasonal patterns n RR, ER, and CC ratos Percent (%) Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Excess reserves rato Requred reserves rato Currency n crculaton rato Notes: Data has monthly frequency coverng the perod from December 2005 to February The fgures show average monthly devatons from mean. Requred reserves (RR) rato, Excess reserves rato (ER) and Currency n crculaton (CC) rato are found by dvdng these fgures wth total amount of checkable deposts. Now, we turn our attenton onto the stablzng role of money multpler under under substantal captal outflows and nflows. Frst, consder a hypothetcal stuaton where an economy experences substantal captal outflows. Captal outflows place deprecaton pressure on domestc currency. We argue deprecaton pressure on 24

26 domestc currency s reduced/softened va two channels: a drect and an ndrect channels. The drect channel s due to changes n banks behavor n use of faclty whch alters the supply of foregn exchange at the FX market va changes n currently held nternatonal reserves of the monetary authorty as explaned n Alper et al., Perods wth captal outflows are usually characterzed by ncrease n FX fundng cost relatve to domestc currency and tghtenng of external borrowng constrants. Increases n foregn exchange relatve to domestc currency fundng costs nduce banks to use faclty less ntensely. Lower utlzaton of would release some of the requred reserves kept n FX at the Central bank to the FX market whch would reduce the deprecaton pressure on domestc currency. The second and ndrect channel s due to changes n the money multpler stmulated by banks decson on the utlzaton of the faclty. Lower utlzaton of the faclty reduces money (M2) multpler as proved n comparatve statcs v 6. If the monetary authorty allows prvate market forces to determne the domestc nterest rate.e. f the monetary authorty reman passve and does not engage n domestc open market operatons, domestc nterest rate rse. The stmulated ncrease n domestc nterest rate would contan captal outflows lmtng deprecaton pressure on domestc currency. Ths fndng provde bass for Aslaner et al., 2014 where the authors argue actve nterest rate polcy responses to captal flows undermne automatc stablzng feature of the faclty. Captal nflows, on the other hand, nduce banks to utlze faclty more ntensely whch contans apprecaton pressure on domestc currency as dscussed n 6 Snce bank use domestc currency more ntensely to satsfy reserve requrements, lower utlzaton of the faclty reduces domestc money supply drectly, as well. 25

27 Alper et al., Moreover, hgher utlzaton of ncreases money (M2) multpler as proved n comparatve statcs v. If the monetary authorty allows prvate market forces to determne the domestc nterest rate, domestc nterest rate fall. The fall n domestc nterest rate would slow down captal nflows reducng apprecaton pressure on domestc currency. VI. Concluson In ths paper, our objectve was to study Reserve Opton Mechansm (), a recently ntroduced macroprudental monetary polcy nstrument, and ts mpact on the behavor of money multpler. To ths end, we analytcally derve a generalzed money (M2) multpler under and assess sze, stablty and seasonal propertes quanttatvely. Our case study s Turksh economy where we use monthly data durng the decade 2005 to We report one tme sgnfcant rse n the sze of money multpler and slght upward adjustment n the long run trend component of the multpler due to adopton of. Moreover, emprcal evdence suggests substantal change n the seasonal behavor of the multpler, cash rato, requred and excess reserves wth. We report money multpler s strongly seasonal along wth sgnfcant readjustment wthn year movements n the multpler under a monetary system wth. Strong seasonal movements n requred and excess reserves, the components of money multpler, prove to be the underlyng forces behnd strong seasonal behavoral changes n the money multpler. We show that money multpler s substantally less volatle n a monetary system wth. We dscuss the stablzng mpact of Reserve Opton Mechansm n the foregn exchange market va adjustments n money multpler respondng to banks manageral decsons over requred reserves. The actual mechansm through whch the ntroduced money multpler under reduces the sze of varatons n the foregn exchange market reman an open queston for future studes. 26

28 VII. References Alper, K., Bnc, M., Demralp, S., Kara, H., & Özlü P. (2014). Reserve requrements, lqudty rsk and credt growth. Ankara, the Central Bank of Republc of Turkey Workng Paper No. 14/24. Alper, K., Kara, H., & Yorukoglu, M. (2013). Reserve optons mechansm. Central Bank Revew, 13, Aslaner, O., Çıplak, U., Kara, H., & Küçüksaraç, D. (2014). Reserve opton mechansm: Does t work as an automatc stablzer? Ankara, the Central Bank of Republc of Turkey Workng Paper No. 14/38. Değerl, A., & Fendoğlu, S. (2015). Reserve opton mechansm as a stablzng polcy tool: evdence from exchange rate expectatons. Internatonal Revew of Economcs and Fnance, 35, Federco P., Vegh, C. A., & Vuletn, G. (2014). Reserve requrement polcy over the busness cycle. Cambrdge, NBER Workng Paper No Gallagher, K. P., & Tan, Y. (2014). Regulatng captal flows n emergng markets: the IMF and the global fnancal crss. Boston Unversty - Global Economc Governance Intatve (GEGI) Workng Paper, May (5). Jayaratne, J., & Morgan, D. P. (2000). Captal market frctons and depost constrants at banks. Journal of Money, Credt and Bankng, 32(1), Küçüksaraç, D., & Özel, Ö. (2012). Reserve optons mechansm and computaton of reserve optons coeffcents, CBT Research Notes n Economcs, /07. Marvn, G. (1982). A model of money stock determnaton wth loan demand and a bankng system balance sheet constrant. FRB Rchmond Economc Revew, 68(1), Oduncu, A., Akçelk, Y., & Ermşoğlu, E. (2013) Reserve optons mechansm and FX volatlty, Ankara, the Central Bank of Republc of Turkey Workng Paper No. 13/03. The Central Bank of the Republc of Turkey (2010). Press release on requred reserves. Retreved from 982f-11b4d5eac9f4/ANO pdf?MOD=AJPERES&CACHEID=a7fea09b-28c6-403b-982f-11b4d5eac9f4 27

29 VIII. Appendx Appendx I: Data Currency n crculaton Interbank overnght O/N rates Reserve requrement ratos Reserve Opton Rato Requred reserves Excess reserves Checkable deposts Tme deposts Date MB M1 M2 C RR ER D TD r rr 1 rr 2 ROR (n bllons of Turksh lra) (percent, %) Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

30 Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun

31 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar