Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our User Agreement and Privacy Policy.

Slideshare uses cookies to improve functionality and performance, and to provide you with relevant advertising. If you continue browsing the site, you agree to the use of cookies on this website. See our Privacy Policy and User Agreement for details.

Successfully reported this slideshow.

Like this presentation? Why not share!

- Inflation Targeting by G Garcia 2118 views
- The Essential Toolkit for Your: EDR... by Steven Oest 207 views
- Adopting Inflation Targeting for Mo... by iosrjce 330 views
- Targeting inflation and growth by Dushyant Bheda 557 views
- Inflation targeting group 7 section... by SHASHANK SHARMA 129 views
- Inflation targeting in Emerging Mar... by Sarthak Luthra 427 views

881 views

Published on

IHU, Greece, University of Liverpool, UK , University of Macedonia, Greece, LSE, UK and RCEA, Italy.

November 2015. Open Seminar at Eesti Pank

Published in:
Economy & Finance

No Downloads

Total views

881

On SlideShare

0

From Embeds

0

Number of Embeds

8

Shares

0

Downloads

18

Comments

0

Likes

1

No embeds

No notes for slide

- 1. Does Inflation Targeting matter? T. Dergiades1, C. Milas2, T. Panagioditis3 1IHU, Greece, 2University of Liverpool, UK , 3University of Macedonia, Greece, LSE, UK and RCEA, Italy. November 2015
- 2. 2 1. Background • Inflation Targeting (IT) became a popular framework for the conduct of monetary policy replacing other targeting regimes (monetary or exchange rate targeting). • It features an explicit target for inflation and greater emphasis on central banks’ transparency, credibility, and accountability in conducting monetary policies. • Main benefit the alleviation of the dynamic inconsistency problem and leads to lower (expectations) of inflation and inflation variability. • From the 1990s inflation rates became lower, less persistent and less variable. These were regarded as indications that IT was successful. 2
- 3. • Bernake (2011): flexible inflation targeting “increased the effective scope of monetary policy” and “the basic principles of flexible inflation targeting – the commitment to a medium-term inflation objective…seem destined to survive”. • Mervyn King (2012) reflecting on twenty years of IT : “…the results in terms of low and stable inflation have been impressive. There have been pronounced reductions in the mean , variance and persistence of inflation in Britain and elsewhere.” 3
- 4. •But Carney (2014): “That fight (against inflation) culminated in the adoption of an inflation target, which helped secured 15 years of price stability and sustained economic growth. However, with time, a healthy focus became a dangerous distraction”. 4
- 5. 5 Background • Svesson (1997) argues that IT implies base drift of the price level, suggesting that the price level has a unit root and inflation is stationary. • Previous studies have tested for stationarity of the level of inflation by employing either linear or nonlinear unit root tests. • Culver and Papell (1997) employ sequential break and panel unit root tests and Hassler and Wolters (1995) fractional unit root tests for international inflation data. 5
- 6. • Advantages: an official announcement of the target increases the credibility of the policy of the CB, alleviates the dynamic incontinency problem, anchors expectations and secures price stability. Empirical evidence further strengthens the efficiency of the policy. • Opponents of IT argue that it is merely a conservative window dressing (Romer 2006 p532), it reflects a more general process of reforms (Gertler 2005) such as central bank independence (Kohn 2005) and/or the CB’s communication policy (Mankiw 2005). 6
- 7. • Batini and Lexton (2007): Other disadvantages (1) decreased discretion by the central bank that leads to declines in output growth; (2) too much discretion that results in the inability to influence inflation expectations; (3) higher exchange rate volatility as inflation targeting ignores exchange rate levels; and (4) inability of inflation targeting to be successful in countries that do not meet strict preconditions 7
- 8. 8 2. Existing Literature • Ball and Sheridan (2005) focusing on twenty major developed OECD countries (7 of which adopted IT regime) and through the implementation of the standard “differences in differences” approach they concluded that the IT regime seems to have no significant impact upon the improvement of inflation, output and interest rates. 8
- 9. Literature • Gregoriou and Kontonikas (2006) looked at 5 OECD countries (UK, Canada, Sweden, Australia and NZ) and two high inflation non-OECD countries (Chile and Israel). Sample starts from the establishment of IT in each country until 2004. ADF fail to reject the non- stationarity null. ESTAR unit root test points towards stationarity (nonlinear mean reverting). 9
- 10. 10 Existing Literature continued • Gonçalves and Salles (2008) argue that the sample selection bias problem contaminates the results reported by Ball and Sheridan (2005). In their effort to overcome the bias selection problem, they examined a sample of 36 emerging economies (13 of which adopted IT regime). Gonçalves and Salles (2008) found that the IT regime delivers irrefutable gains, with respect to inflation and growth volatility. 10
- 11. •Lin and Ye (2007) by implementing a variety of propensity score matching methods for 22 major industrial countries (7 inflation targeters) managed to take in consideration the self- selection problem. The authors showed that such a policy adoption is non-effective in industrial countries with respect to inflation, inflation variability, long-term interest rates and income velocity of money. 11
- 12. 12 Existing Literature continued • Lin and Ye (2009) using for once more the propensity score matching methodological framework and focusing on 52 developing countries for the years 1985 to 2005, they offered evidence for the effectiveness of the IT regime. Specifically, Lin and Ye (2009) showed that the IT regime for the developing countries constituted a significant parameter for lowering both inflation and inflation variability. • These results come to confirm the Gonçalves and Salles (2008) empirical findings.
- 13. 13 3. Econometric Methodology To examine a change in persistence we implement the modified Busetti and Taylor (2004) tests (BT), as suggested by Harvey et al (2006). BT to test a change in persistence from I(0) to I(1) proposed the following statistics (the null is a constant I(0) process) : 13 1 [ , ] max ( ) l u M MH K K 3 [ , ] log exp ( )K K l u M MH d 2 2 1,[ ] 1 [ ] 1 2[ ]2 0,1 1 ˆ[(1 ) ] ( ) ˆ[ ] T t it T i T M T t it i T T K (1) (2) with (3) 2 [ , ] ( ) l u M MH d K K
- 14. 14 Methodology continued where, τ(0,1), [τl ,τu] is a sub-interval of (0,1), T is the sample size and finally, ε0,i and ε1,i are the OLS residuals from the regression of the examined series on a constant for the samples t = 1,…,[τΤ] and t = [τΤ],…,T, respectively. BT to test for a change from I(1) to I(0) proposed the use of the reciprocal of , that is , with J=1,2,3 However, the above statistics are severely oversized when the true generating process is constantly I(1). M K 1 /j MH K
- 15. 15 Methodology continued min,min expj M j Mj H bJ H K K 9 1 , 1, ,[ ]i t t i ti k y t u t T x Harvey et al (2006) to test the null of a constant I(0) or I(1) generating process against a change from I(0) to I(1), proposed the following modification: where, b is a constant provided by Harvey et al (2006) and ,with equals to 1/T times the Wald statistic that corresponds to the test of the joint hypothesis , in the regression below: min [ , ] 1,[ ]min l u Tj j 1,[ ]Tj 1 9... 0k (4) (5)
- 16. 16 Methodology continued When the alternative examined hypothesis is of an opposite direction (that is, I(1) to I(0)), the statistic used in equation (4) is substituted by its reciprocal analogue and the factor is replaced with , where is defined as , with the exception that specification (5) is now estimated for t=[τT],…,T. The modified statistic is denoted as : 1,[ ]Tj min [ , ] [ ],min l u R T Tj j [ ],T Tj 1,[ ]Tj ,min 1/j M j H K
- 17. 17 Methodology continued The location of the breakpoint for a change that might be either from I(0) to I(1) or from I(1) to I(0), BT propose the use of (5) and (6), respectively. [ , ] ˆ arg max l u [ , ] arg min l u 22 1,[ ] 1 2 2 0, 1,[ ] 22 0,1 ˆ ˆ ˆ ˆ where and definedaspreviously, are T tt T t tT tt T T T (6) (7)
- 18. 18 4. Data • Our study implements tests for change in persistence for 45 countries individually as well as for three groups of countries (G7, OECD and OECD Europe). • The frequency of the data used is monthly over the period 1980:1 to 2010:6. • For each country (or croup of countries) the annualized inflation is used which is estimated by the logarithmic difference of the respective consumer price index. • Exception consist the countries of Australia and New Zealand given that there is no availability of monthly observations. For the two abovementioned countries the annualized inflation series is used in quarterly frequency for the period 1980:q1 to 2010:q2. 18
- 19. 19 Data continued • Primary source of the data used is the International Financial Statistics (IFS) database of the International Monetary Fund (IMF), while the secondary source of data is the Main Economic Indicators (MEI) database provided by OECD. • The MEI database has been used as source of data only for the following countries (or group of countries): OECD total, OECD Europe, G7 countries, China, Germany and Ireland.
- 20. 20 5. ResultsTable 1. Change in Persistence Testing Statistics for testing I(0)→I(1) Statistics for testing I(1)→I(0) Country Sample Break date H1(K Μ (.))min H2(K Μ (.))min H3(K Μ (.))min Break date H1(1/K Μ (.))min H2(1/K Μ (.))min H3(1/K Μ (.))min Euro-zone countries Austria 1980m01-2010m06 2004m06e 0.134 0.038 0.014 1987m05 87.073*** 30.455*** 39.343*** Belgium 1980m01-2010m06 2004m06e 0.028 0.006 0.002 1987m10 144.839*** 75.477*** 67.742*** Cyprus 1980m01-2010m06 2004m06e 0.058 0.027 0.010 1987m03 79.357*** 35.680*** 35.732*** Finland 1980m01-2010m06 2004m06e 0.095 0.015 0.004 1994m05 83.397*** 45.496*** 36.660*** France 1980m01-2010m06 2004m06e 0.009 0.001 0.000 1992m05 478.584*** 232.202*** 229.055*** Germany 1980m01-2010m06 2004m06e 0.057 0.025 0.008 1994m03 40.1148*** 17.167*** 16.927*** Greece 1980m01-2010m06 1986m02e 11.102 2.923 3.914 1999m10 247.639*** 48.409*** 108.655*** Ireland 1980m01-2010m06 2004m06e 0.013 0.008 0.002 1987m06 18.452** 13.561*** 7.513** Italy 1980m01-2010m06 1986m02e 0.002 0.000 0.000 1996m09 664.274*** 235.192*** 328.272*** Luxembourg 1980m01-2010m06 2004m06e 0.018 0.004 0.001 1988m04 147.160*** 62.256*** 67.814*** Malta 1980m01-2010m05 2004m06e 0.002 0.001 0.000 1986m05 181.266*** 90.587*** 85.549*** Netherlands 1980m01-2010m06 2000m12 0.012 0.008 0.001 1988m06 60.488*** 13.874*** 26.200*** Portugal 1980m01-2010m06 1986m02e 2.871 0.287 0.334 1995m10 76.208*** 30.398*** 32.446*** Slovakia 1994m12-2010m06 1999m06 9.634 2.294 2.862 2004m12 2.191 0.560 0.407 Slovenia 1994m12-2010m06 2007m06e 0.439 0.173 0.063 1998m01e 2.454 1.438 0.786 Spain 1980m01-2010m06 2004m06e 0.317 0.039 0.014 1995m07 78.909*** 37.857*** 35.696*** Inflation-targeting countries Australia 1980 q01-2010 q02 1986 q03 0.918 0.142 0.074 2001 q02 799.178*** 140.774*** 392.197*** Brazil † 1996m01-2010m06 2002m11 0.359 0.141 0.050 2007m07e 22.772** 3.471 8.597** Canada 1980m01-2010m05 2004m06e 0.003 0.001 0.000 1995m01 163.941*** 77.138*** 77.101*** Chile 1980m01-2010m01 1989m11 13.390 3.123 4.755 1997m04 67.754*** 29.205*** 29.743*** Columbia 1980m01-2010m01 1988m03 43.726*** 13.075*** 18.853*** 2003m09 113.813*** 11.254*** 49.018*** Czech Reb. 1994m12-2010m06 1998m01e 7.077 0.257 1.155 1999m12 16.628* 7.914*** 6.387** Ghana 1980m01-2010m06 1995m03 0.265 0.104 0.051 2004m01 307.032*** 35.878*** 145.705*** Hungary 1980m01-2010m06 1987m07 768.189*** 65.894*** 367.710*** 2001m09 295.583*** 42.756*** 136.369*** Iceland 1984m01-2010m06 2005m03e 0.377 0.089 0.039 1993m01 14.401* 7.581*** 5.069* Indonesia 1980m01-2010m05 1997m12 24.433** 4.255* 8.278** 2000m06 45.793*** 5.721** 18.221*** Notes: see next page
- 21. 21 Results continued Table 1 (continued). Change in Persistence Testing Statistics for testing I(0)→I(1) Statistics for testing I(1)→I(0) Country Sample Break date H1(K Μ (.))min H2(K Μ (.))min H3(K Μ (.))min Break date H1(1/K Μ (.))min H2(1/K Μ (.))min H3(1/K Μ (.))min Inflation-targeting countries (continued) Israel 1980m01-2010m06 1986m02e 0.016 0.003 0.001 1999m08 926.556*** 586.374*** 676.896*** Korea† 1980m01-2010m05 2005m11e 0.024 0.002 0.004 1984m09 581.440*** 142.162*** 281.137*** Mexico 1980m01-2010m05 1986m05 3.838 0.339 0.429 2001m10 724.031*** 337.753*** 373.500*** New Zeal. 1980m01-2010m06 1986 q03 1.486 0.115 0.073 1999 q04 958.553*** 256.141*** 564.051*** Norway 1980m01-2010m06 2003m12 0.033 0.005 0.001 1992m09 258.789*** 139.025*** 125.209*** Peru 1992m01-2010m06 2006m11e 0.000 0.000 0.000 1998m12 641.159*** 208.118*** 227.242*** Philippines 1980m01-2010m05 2004m06e 0.434 0.132 0.066 1987m04 162.917*** 41.325*** 76.607*** Poland 1989m01-2010m06 1993m05e 0.004 0.005 0.000 2001m08 261.734*** 306.316*** 192.123*** Romania 1991m10-2010m06 1995m07e 0.197 0.033 0.012 2006m03 102.608*** 154.718*** 37.065*** South Africa† 1980m01-2010m05 1985m01 67.162*** 5.794** 28.631*** 2000m03 10.548 3.726* 3.568* Sweden 1980m01-2010m06 1986m02e 0.528 0.089 0.036 1990m02 192.208*** 62.820*** 88.167*** Switzerland 1980m01-2010m06 1986m02e 1.461 0.294 0.155 1994m10 131.723*** 45.908*** 61.252*** Thailand† 1980m01-2010m06 2005m11e 0.012 0.013 0.003 1985m01 17.372* 7.760*** 6.030** Turkey 1980m01-2010m06 1994m03 10.746 4.555* 3.029 2004m03 263.594*** 97.625*** 123.579*** UK 1980m01-2010m06 2004m06e 0.016 0.006 0.001 1991m07 100.182*** 60.403*** 46.175*** Other countries and Groups China 1994m01-2010m02 2006m12e 0.000 0.000 0.000 1999m06 10.250 7.069** 3.678 Denmark 1980m01-2010m06 2004m06e 0.009 0.001 0.000 1990m03 545.549*** 302.392*** 263.543*** Japan 1980m01-2010m05 2004m06e 0.081 0.035 0.010 1992m09 52.361*** 16.131*** 22.663*** US 1980m01-2010m06 2004m06e 0.002 0.001 0.000 1987m01 71.582*** 38.464*** 32.731*** G7 1980m01-2010m04 2004m04e 0.000 0.000 0.000 1992m04 175.663*** 84.571*** 82.724*** OECD Europe 1980m01-2010m04 1994m07 0.025 0.025 0.004 2002m12 116.406*** 12.696*** 52.345*** OECD 1980m01-2010m04 1986m02e 0.184 0.064 0.018 1998m07 136.591*** 27.647*** 62.882*** Notes: (a) ***, ** and * a change in persistence at the 0.01, 0.05 and 0.10 significance level, respectively. (b) The subscript e signifies that the identified breakpoint is located on an extreme of the initial search interval (0.20-0.80). (c) The superscript † implies that an extreme breakpoint is also combined with a significant change in the persistence. In these cases we amend the search interval to (0.15-0.85) and the tests for a change in persistence are implemented from the beginning. (d) On those cases where there is no evidence of a significant change in the persistence, given the identification of an extreme breakpoint, then the extreme breakpoint is disregarded and the testing procedure is not repeated for the extended search interval.
- 22. 22 Results continued • From the 45 countries included in the sample, 35 countries (13 from the euro-zone group, 19 from the IT group and 3 from the rest countries) appear to experience change in persistence from I(1) to I(0) in the annualized inflation (criterion used: 1% significance level for all the implemented statistics). - Exception are the following 10 countries: For the Euro zone group: Ireland, Slovakia, Slovenia For the IT group: Brazil, Czech Republic, Iceland, Indonesia, South Africa, Thailand For the rest countries: China
- 23. 23 Results continued • The 3 examined groups of countries (G7, OECD Europe and OECD) appear also to experience change in persistence from I(1) to I(0) in the annualized inflation (criterion used: 1% significance level for all the implemented statistics). • From the 45 countries included in the sample, only 2 countries (both from the IT group) appear to experience change in persistence from I(0) to I(1) in the annualized inflation (criterion used: 1% significance level for all the implemented statistics). These two countries are the following: Columbia and Hungary
- 24. 24 Results continued Figure 2. Difference between the formal adoption date and the estimated break date. -200 -160 -120 -80 -40 0 40 80 120 160 200 A us tralia *** Canada *** Chile *** Colum bia *** Cz ec h Rebublic *** G hana Hungary *** Ic eland Indones ia Is rael *** *** K orea M ex ic o *** New Zealand *** *** Norway *** P eru *** P hilippines P oland *** Rom ania *** S outh A fric a*** S weden *** S witz erland Thailand *** Turk ey *** UK (94) (48) (79) (48) (23) (-40) (3) (-98) (-61) (91) (-163) (33) (109) (-102) (-37) (-177) (34) (7) (1) (-35) (-63) (-184) (-22) (-15) adoption datem onths before m onths after Notes: (a) Figures within the parentheses denote the time difference in months between the formal adoption date and the estimated break date for all the inflation targeting counties, (b) *** denotes a 0.01 significant change in persistence provided that there is an agreement among the three alternative implemented tests, (c) Finally, Brazil is excluded since the estimated break date has been found to be an extreme breakpoint.
- 25. 25 Results continued Table 2. Descriptive statistics for the inflation targeting countries. Before the FAD After the FAD Before the EBD After the EBD Country FAD EBD Mean Median St. dev. Mean Median St. dev. Mean Median St. dev. Mean Median St. dev. Inflation-targeting countries (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Australia 1993m04 2001 q02 7.25 7.79 3.15 2.69 2.55 1.42 5.36 5.45 3.57 2.94 2.80 0.94 Brazil 1999m06 2007m07e 7.98 5.49 6.01 6.79 6.27 3.08 - - - - - - Canada 1991m01 1995m01 6.36 4.71 3.09 2.01 1.87 1.26 5.27 4.36 3.38 1.94 1.99 0.91 Chile 1990m09 1997m04 21.67 21.05 8.22 7.04 4.69 6.16 18.55 17.83 8.51 3.73 3.61 2.25 Columbia 1999m09 2003m09 23.06 23.44 4.74 6.17 6.10 1.92 20.50 21.88 7.16 5.16 5.26 1.49 Czech Reb. 1998m01 1999m12 8.94 8.89 1.23 3.31 2.65 2.92 8.05 8.70 3.30 2.72 2.56 1.99 Ghana 2007m05 2004m01 33.48 23.80 30.83 15.53 16.43 3.75 36.36 26.39 31.84 14.07 13.09 3.55 Hungary 2001m06 2001m09 15.23 13.55 8.73 5.51 5.59 1.85 15.16 13.03 8.70 5.40 5.40 1.74 Iceland 2001m03 1993m01 11.74 5.75 12.28 6.47 5.39 4.02 19.94 19.27 12.00 4.71 3.72 3.63 Indonesia 2005m07 2000m06 11.34 9.03 12.16 8.65 6.72 4.67 12.01 9.11 13.43 8.65 7.41 3.86 Israel 1992m01 1999m08 113.17 103.58 121.78 5.31 4.25 4.63 73.18 17.93 107.74 2.12 2.10 2.28 Korea 1998m04 1984m09 7.44 5.25 6.85 3.16 2.98 1.53 13.36 9.34 11.10 4.34 3.94 2.25 Mexico 1999m01 2001m10 46.69 28.58 39.62 6.19 4.83 3.70 42.23 26.52 38.88 4.55 4.44 0.82 New Zealand 1990m03 1999 q04 11.89 13.14 5.13 2.38 2.19 1.42 7.09 4.70 6.22 2.58 2.64 0.88 Norway 2001m03 1992m09 5.27 4.00 3.59 1.99 1.96 1.31 7.31 6.56 3.29 2.11 2.16 1.05 Peru 2002m01 1998m12 19.91 10.09 24.55 2.44 1.99 1.86 27.38 11.81 26.28 2.63 2.51 1.74 Philippines 2002m01 1987m04 11.43 8.87 10.91 5.17 4.33 2.68 17.52 11.96 16.51 7.25 7.13 4.01 Poland 1998m10 2001m08 133.01 35.74 270.33 3.97 3.67 2.87 104.92 31.06 243.41 2.60 2.44 1.44 Romania 2005m08 2006m03 83.46 43.87 84.72 6.25 6.30 1.71 80.42 41.33 84.29 5.93 5.99 1.57 South Africa 2000m02 2000m03 12.22 13.08 3.87 6.13 5.96 3.20 12.18 13.06 3.91 6.17 5.98 3.19 Sweden 1993m01 1990m02 7.81 7.92 3.25 1.50 1.44 1.41 7.95 7.69 3.05 2.34 1.66 2.82 Switzerland 2000m01 1994m10 2.81 2.64 2.01 0.96 0.96 0.82 3.53 3.40 1.82 0.90 0.83 0.77 Thailand 2000m05 1985m01 5.33 4.49 4.41 2.55 2.20 2.23 8.49 5.09 7.17 3.61 3.47 2.38 Turkey 2006m01 2004m03 57.46 60.36 27.57 9.03 9.61 1.99 61.07 62.75 25.16 9.30 9.72 1.78 UK 1992m10 1991m07 7.28 5.68 4.14 2.65 2.76 1.25 7.62 6.34 4.22 2.75 2.89 1.27 Notes: (a) FAD denotes the formal adoption date. (b) EBD denotes the estimated break date. (c) Columns (1), (2), (3) and (4), (5), (6), present the mean, the media and the standard deviation before and after the formal adoption date of the inflation targeting policy. (d) Columns (7), (8), (9) and (10), (11), (12), present the mean, the media and the standard deviation before and after the estimated break date. (e) The descriptive statistics in columns (7), (8), (9) and (10), (11), (12) are not reported for Brazil, since the EBD is located on an extreme of the search interval.
- 26. 26 Results continued Table 3. D escriptive statistics for the non-inflation targeting countries or groups. Before the EBD After the EBD Country EBD M ean M edian St. dev. M ean M edian St. dev. Euro-zone countries Austria 1987m05 4.45 4.97 2.00 2.19 2.05 1.00 Belgium 1987m10 5.73 6.52 2.66 2.11 2.09 1.08 Cyprus 1987m03 6.74 6.03 4.07 3.32 3.23 1.63 Finland 1994m05 6.16 5.83 3.24 1.50 1.32 1.19 France 1992m05 6.60 3.78 4.22 1.62 1.69 0.73 G erm any 1994m03 3.23 2.95 1.99 1.54 1.50 0.72 G reece 1999m10 15.48 16.42 6.54 3.19 3.24 0.93 Ireland 1987m06 11.53 10.20 6.52 2.66 2.78 2.19 Italy 1996m09 8.91 6.27 5.56 2.17 2.19 0.68 Luxem bourg 1988m04 5.15 6.26 3.57 2.32 2.27 1.01 M alta 1986m05 5.05 1.51 6.51 2.45 2.58 1.54 N etherlands 1988m06 3.22 2.80 2.73 2.16 2.13 0.87 Portugal 1995m10 14.30 12.29 7.50 2.55 2.71 1.26 Slovakia 2004m12 7.85 7.24 3.07 3.06 3.04 1.21 Slovenia 1998m01 e - - - - - - Spain 1995m07 8.58 6.90 3.85 2.81 2.94 1.25 O ther countries and G roups China 1999m06 9.39 7.40 9.76 1.68 1.28 0.96 D enm ark 1990m03 6.87 5.51 3.25 2.10 2.11 0.78 Japan 1992m09 2.61 2.31 2.05 0.08 -0.10 0.97 U S 1987m01 6.15 4.20 4.04 2.95 2.90 1.33 G 7 1992m04 5.35 4.30 2.99 2.03 2.13 0.85 O ECD Europe 2002m12 8.11 7.67 2.79 2.59 2.52 0.83 O ECD 1998m07 7.60 6.77 2.73 2.83 2.79 1.04 N otes: (a) EBD denotes the estim ated break date. (b) C olum ns (1), (2), (3) and (4), (5), (6), present the m ean, the m edia and the standard deviation before and after the estim ated break date. (c) The descriptive statistics are not reported for Slovenia, since the EBD is located on an extrem e of the search interval.
- 27. 27 6. Does Inflation Targeting matter? • On the basis of the above implemented tests for a change in the annualized inflation persistence from I(1) to I(0), theme of interest is whether a significant change in persistence is significantly associated with those countries that implement an inflation targeting policy. • This is a binary choice problem, which may be handled through a simple probit regression model: 0 1ln 1 i i i i b b D where, πi is the probability that the country i experiences a significant change in persistence from I(1) to I(0). Di is a dichotomous dummy variable which indicates whether the country i is an IT country or not. b0 and b1 are coefficients to be estimated and is as usual the error term.
- 28. 28 Does Inflation Targeting matter? continued • Provided that each country’s πi is non-observable, we introduce the binary yi variable, which receives the value of 1 if a significant change in persistence is signified simultaneously by the three implemented Statistics at the 0.01 significance level, and 0 otherwise. • A positive and significant value for the b1 coefficient, it would be an indication for the validity of the proposition that IT countries face a significant higher probability, over the rest countries, for a change in the inflation persistence from I(1) to I(0).
- 29. 29 Does Inflation Targeting matter? continued • The parameter estimates of the simple probit regression model are analytically illustrated in Table 2. Table2. ThebinaryProbit estimationresultsanddiagnosticStatistics. Variable Coefficient Std. error z-Statistic p-value Marginal effect(Std. error) Constant 0.8416 0.3194 2.6343 0.0084 - Di -0.1353 0.4213 -0.3211 0.7481 -0.0400(0.1376) Diagnosticstatistics McFaddenR-squared 0.0021 LRStatistic 0.1034 Log-likelihood -23.7850 LRStatisticp-value 0.7477 • Based on a sample of 45 observations (countries) the estimated b1 coefficient is negative and non-significant. Therefore, it can be said that countries which are inflation targeters face similar probability for a significant 1% change in the inflation persistence from I(1) to I(0), compared with other countries.
- 30. 30 Does Inflation Targeting matter? continued • Another subject of interest is whether the inflation targeting countries present higher values for the three estimated statistics used to test the change in persistence • For this purpose, we regress the binary Di variable (defined as previously) over the three modified Busetti and Taylor (2004) Statistics. The three subsequent specifications are illustrated below: 0 1 ,mini j M ij i D b b H K 1,2j with and 3 • Positive and significant b1 coefficient for each specification, it would be an indication that IT countries face indeed higher probability to receive greater values for the implemented persistence change Statistics.
- 31. 31 Does Inflation Targeting matter? continued • The estimation results for the three similar probit regressions are provided in Tables 6, 7 and 8, below: Table 3. The binary Probit estimation results and diagnostic Statistics. Variable Coefficient Std. error z-Statistic p-value Marginal effect (Std. error) Constant -0.1627 0.2621 -0.6207 0.5347 - H1j(1/K Μ (.))min 0.0014 0.0009 1.5626 0.1181 0.0005 (0.0003) Diagnostic statistics McFadden R-squared 0.0519 LR Statistic 3.2121 Log-likelihood -29.3072 LR Statistic p-value 0.0731 Table 4. The binary Probit estimation results and diagnostic Statistics. Variable Coefficient Std. error z-Statistic p-value Marginal effect (Std. error) Constant -0.0685 0.2510 -0.2729 0.7849 - H2j(1/K Μ (.))min 0.0024 0.0020 1.1862 0.2355 0.0009 (0.0007) Diagnostic statistics McFadden R-squared 0.0299 LR Statistic 1.8501 Log-likelihood -29.9882 LR Statistic p-value 0.1738
- 32. 32 Does Inflation Targeting matter? continued • What is revealed from the estimation results (Tables 3,4 and 5) no matter which Statistic is used as regressor, is that IT countries do not receive significantly higher values for the Statistics used to test the change in persistence. • The above conclusion is justified upon the fact that all the coefficients remain statistically insignificant even at the 0.1 significance level. Table 5. The binary Probit estimation results and diagnostic Statistics. Variable Coefficient Std. error z-Statistic p-value Marginal effect (Std. error) Constant -0.14902 0.2580 -0.5774 0.5636 - H3j(1/K Μ (.))min 0.0027 0.0018 1.4795 0.1390 0.0011 (0.0006) Diagnostic statistics McFadden R-squared 0.0566 LR Statistic 3.5020 Log-likelihood -29.1622 LR Statistic p-value 0.0613
- 33. • Other Considerations: 1.CBI 2.Institutions 3.Endogeneity 33
- 34. 34 Conclusions • We provide econometric evidence that the IT regime has no significant effect upon the change in persistence of the annualized inflation series from I(1) to I(0). • The reported results should be interpreted with caution • Many non-targeting countries in our sample, implement policies close to those of the IT countries. • Our results imply that central banks deeds matter more than their publicly announced intentions.
- 35. Conclusions • Forder (1998) point out that : “ The test might appear to show a statistical regularity, say between the content of the statutes of a central bank and the rate of inflation, but in the absence of a theoretical connection that would be of no interest. We might note that the European German-speaking countries (Germany, Switzerland, and Austria) have low inflation. This does not mean that if we all started speaking German, inflation would fall ”. 35

No public clipboards found for this slide

Be the first to comment