inflation dynamics and business cycles
TRANSCRIPT
2013
Borsa İstanbul Finance & Economics Conference (BIFEC) 2013
“Policy Issues and Challenges in the Global Financial System and Economies”
www.bifec.com
September 30 – October 1, 2013İstanbul
Book of Abstracts
Borsa İstanbul Finance & Economics Conference (BIFEC) 2013
“Policy Issues and Challenges in the Global Financial System and Economies”
September 30 – October 1, 2013
www.bifec.com
ISBN: 978-605-5076-04-7
BIFEC Book of Abstracts & Proceedings
Volume I Issue II
(Proceedings)
March 2014, İstanbul
The views expressed in this book are those of the author(s) and do not necessarily represent the official views of Borsa İstanbul or its members. The authors of the individual papers are responsible for
technical, content, and linguistic correctness. The refereeing process is managed by the Research and Business Development Department of Borsa İstanbul.
This e-book is published by Borsa İstanbul
Copyright © 2014
i BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Contents (Volume I Issue II) - Proceedings
Formal and Informal Regulations for Credit Card Payment Services ................................ 1
Güzin Gülsün AKIN & Ahmet Faruk AYSAN & Gültekin GÖLLÜ & Levent YILDIRAN
Assembling International Equity Datasets – Review of Studies on the Cross-Section of Common Stocks .......................................................................................................... 34
Antonina WASZCZUK
Trading Puzzle, Puzzling Trade .......................................................................................... 66
Orhan ERDEM & Evren ARIK & Serkan YÜKSEL
The effect of investors' confidence on monetary policy- economic growth relationship: a Multivariate GARCH approach................................................................................. 82
Chiara GUERELLO
Agency and Transparency in Financial Markets ............................................................... 110
Sadettin Haluk ÇİTÇİ
Inflation Dynamics and Business Cycles ........................................................................... 121
Süleyman Hilmi KAL & Nuran ARSLANER & Ferhat ARSLANER
Comovement and Polarization of Interest Rate and Stock Market in Turkey ................130
Ahmet DURAN & Burhaneddin İZGİ
The Effect of Global Shocks and Volatility on Herd Behavior in Borsa Istanbul ............ 142
Mehmet BALCILAR & Rıza DEMIRER
Volatility and Transparency of Financial Markets in the MENA Region ......................... 173
Hamid MOHTADI & Stefan RUEDIGER
Hedging Strategy for Electricity Market Price Volatility: The Case of Turkish Electricity Market ...................................................................................................... 196
Sezer Bozkuş KAHYAOĞLU & M. Vedat PAZARLIOĞLU
The Impact of Financial Innovation on Firm Stability ..................................................... 211
Fabian Kuehnhausen
Index of Authors ............................................................................................................... 240
121 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Süleyman Hilmi KAL; Central Bank of the Republic of Turkey
Nuran ARSLANER; Central Bank of the Republic of Turkey
Ferhat ARSLANER; Borsa İstanbul
The paper aims to investigate whether the effect of the backward-looking inflation expectations,
nominal effective exchange rate, money supply, gross domestic product and import prices on
inflation depends on business cycle. For this purpose, a two states Markov Switching Auto
Regression model with time varying transition probabilities to a generic inflation model is
implemented for the period 2003-2013. In this model the states are assigned whether output gap is
positive or negative. The inflation forecasting in-sample and out-of-sample is also utilized by
adopting mean squared error and Diebold Mariano test to measure explanatory and forecasting
power of our model. Our main finding provides that the determinants of inflation have different
dynamics during boom periods as compared to recessions.
Keywords: Inflation; Output Gap; Exchange Rate Pass-Through; Markov Switching
Autoregressions; Business Cycles
JEL: C32, E30, E31, E37, E58
1. Motivation
After 2007-2008 financial crises and ensuing the Great Recession, despite the fact that money
supply increased substantially by the central banks especially in the United States, in the
European Union, and in the other countries, the inflationary pressures still nonexistent; instead
there are concerns of deflation emerged in these regions. This anomaly leads us to make the
following argument that there may be some other variables influence the relationship between
inflation and the factors that determine the inflation such as money supply, exchange rate pass
trough and output. In economic literature, these variables are usually named as state variables. In
this paper, we will attempt to evaluate whether the output gap as one of the possible factors
determining the different dynamics, which take important role in explaining inflation during
various phases of the economy, i.e. the business cycles.
Inflation Dynamics and Business Cycles
122 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Figure 1. The Keynesian View
Our motivation comes from the Keynesian view that when the economy is below its potential (YFE
in Figure 1) , a shift in aggregate demand or in money supply may not have an important impact
on price changes. So, basically, this is the theoretical framework and the motivation behind our
work.
2. Short Literature Review
Unfortunately, there is not much literature about the effects of business cycle on inflationary
factors even though there is considerable litreture on exchange rate pass-through. Joey Chew et al
(2011) find for Singaporean economy that there is asymmetric exchange rate pass through. They
documented that the exchange rate pass-through occurs more during recessionary periods than it
occurs during expansionary periods.
Nidhaleddine (2012) detected that there is nonlinear exchange rate pass through for 12 EU
Countries. Higher exchange rate pass-through is detected when inflation exceeds a certain
threshold.
Oinonen et al (2013)’s study provided evidence that economic dynamics of inflation US has
changed and output gap became more important in US since 1990.
In the following graphs (Figure 2, 3 and 4), bar graphs are output gap, red line is the rate of
inflation and blue line is broad definition of money supply which is M2, import price index and
manufacturing industrial production index. One can see from the figures that, there is no strong
123 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
linear relationship between inflation and money supply. Similarly, there is not any linear pattern
between inflation and import prices and between inflation and industrial production index. We
analyze the relationships among those variables using Markov switching regression method.
124 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
3. Methodology
Markov switching regression model was first used pretty early in 1953 by Quant and Goldfeld and
became popular by Hamilton’s work in 1990s
Markov switching regime models are basically space models. The relationship between two
variables depends on observed state variable. The unobserved state variable follows a Markov
chain.
Markov regime switching model assume that the observed changes in a variable between two-
consecutive periods are a random draw from two distributions. The unobserved state variable
evolves according to Markov chain. Probabilities of switching from one state to another -
transition probability- is not exogenous but endogenous. Basically we use two normal
distributions each of which has a different mean and standard deviation in this article.
State variables (St) determine the distribution for the period. In a two-state case, when St =1
the observed changes are a random draw from: yt /st= 1~ N ( μ1 ,σ21)
When St=2 from: yt /st= 2~ N ( μ2 ,σ
22)
125 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Mean and the variance of the yt depend on the state. Density of yt is conditional on St and is as
follows:
4. Model
We use a generic inflation model (Goldberg and Knetter, 1997) by running monthly data for the
period 200369 and 2014 which is as follows:
inft = α1infet + α2Δmst+ α3Δprodt + α4Δneert + α5Δimpt + εt
where
inft : The rate of inflation in consumer prices,
infet : Backward looking inflation expectation
Δms : Change in money supply,
Δprod : Change in industrial production index,
Δneer : Change in nominal effective exchange rate ,
Δimp : Change in import price index,
εt : Error Term.
If output gap is used as state variables,
State 1: When output gap is positive.
State 2: When output gap is negative
69
2003 is especially picked for the beginning of the analysis period because it is the start of inflation targeting regime in Turkey.
)2
)y(exp(
2
1)sy(
2
st-ttt
st
if
126 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
5. Results
Markov regime switching regression has two sets of coefficients; one for each state. In other
words coefficients for the recessionary periods (negative output gap) are different than the
coefficients for the expansionary periods (positive output gap).
According to the results, when the output gap is negative which is State 1, a change in money
supply interestingly reduces the rate of inflation as opposed to the expectations. Moreover, an
increase in manufacturing industrial production increases the inflation. They are statistically
highly significant. However, regarding to nominal effective exchange rate and import price index,
they are not statistically significant. This is just we found the opposite of Singaporean economy.
During the recessionary times, we do not see exchange rate pass through for the Turkish economy
when the output gap is negative whereas the powerful pass-through exists when the output gap is
positive that is to say, during the boom times, expansionary periods. Money supply and industrial
production are still statistically insignificant, yet if you look at the pass-through coefficient,
nominal effective exchange rate and import prices are highly significant. Their total pass-through
effect is 0.07.
Table 1. Coefficients Estimated by Two-State Markov Switching Process
States Variables Coefficients t-ratios
Negative INFt-1 α11 1.0535 31.1791***
Output ΔM2 α12 -0.0307 -2.4326***
Gap ΔPROD α13 0.1019 3.4213***
(State 1) ΔNEER α14 -0.0114 -0.9241
ΔIMP α15 -0.0196 -1.3885*
Positive INFt-1 α21 0.8948 17.6955***
Output ΔM2 α22 0.0212 1.2409
Gap ΔPROD α23 -0.0296 -1.0038
(State 2) ΔNEER α24 0.0268 2.2324***
ΔIMP α25 0.0456 2.6861*** *, ** and *** denote statistical significance at the 10%, 5% and 1% levels, respectively.
inf2t
= α21
infe
t + α
22Δms
t+ α
23Δprod
t + α
24Δneer
t + α
25Δimp
t + ε
2t
inf1t
= α11
infe
t + α
12Δms
t+ α
13Δprod
t + α
14Δneer
t + α
15Δimp
t + ε
1t
127 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Additionally, Wald test is implemented in order to see whether the coefficients in both states are
statistically different from each other or not. Wald test results are encouraging. According to
these results, all the coefficients are significant. Therefore, in each state for all the variables are
statistically significantly different from each other.
Table 2. Wald Test Results
Variables Test Results
INFt-1 8.0107 ΔM2 7.9264
ΔPROD 15.1797 ΔNEER 4.194 ΔIMP 10.0332
Note: All numbers indicate that a variable is significant at 5% significance level according to the χ²-distribution.
We also did in-sample estimation in order to understand the success of our model. In-sample
estimation mimics the realized inflation. We also used random walk and ordinary least square
estimation of the same model whether the model is success. In terms of mean square errors,
Markov processed model produces significantly lower errors than that of OLS and also the
random walk.
Figure 5. Coefficients Estimated by Two-State Markov Switching Process
Graph 4 shows the transition probabilities of the output gap which we will be derived with the HP
filter, nominal exchange rate and money supply of State 1. In addition to in-sample estimation,
0 50
100 0
0.1
0.2
Realized Inflation
0 50 100 0
0.1
0.2
In-sample Estimation
0 50
100 -20
0
20
Output Gap
0 50 100 -0.5
0
0.5
Nominal Exchange Rate
0 50
100 0
0.5
1
Transition Probability of State 1
0 50 100 -0.5
0
0.5
Money Supply
2003 2005 2007 2009 2011 2013 2003 2005 2007 2009 2011 2013
128 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
out-of-sample estimation is also done. It is seen that the model successfully predicts mimics of
ups and downs the rate of inflation. We used 16 periods as out-of-sample observations (Figure
5&6). The R2 of the model is 80 per cent. We take out the previous inflation from the model is still
around above 30 per cent. Therefore, the model from that point of view seems to be promising.
Figure 6. Realized Inflation and Out-of-Sample Estimation
6. Conclusions
Our main conclusion is that inflation dynamics depends on the business cycles. So, the
determinants of inflation have different dynamics during boom periods as compared to
recessions.
Agenda for future research is that states may be determined whether the expected inflation is
realized or not and reel effective exchange rate is overvalued or undervalued. Cross country
analysis can also be done by using the same model and methodology.
0 2 4 6 8 10 12 14 160.06
0.08
0.1
0.12Gerceklesen Enflasyon
0 2 4 6 8 10 12 14 160.04
0.06
0.08
0.1
0.12Ornekleme Disi Enflasyon Tahmini
Realized Inflation
Out-of-Sample Inflation Estimation
2003 2005 2007 2009 2011 2013
2003 2005 2007 2009 2011 2013
129 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Reference
Goldberg, L., and M. M. Knetter. 1997. “Goods Prices and Exchange Rates. What have We
Learned?” Journal of Economic Literature 35, 1243-72.
Goldfeld, S. M. and R. E. Quandt 1973. “A Markov Model for Switching Regressions”, Journal
of Econometrics 1, pp. 3-16.
Hamilton, James D. 1990. “Analysis of Time Series Subject to Changes in Regime”, Journal of
Econometrics 45, pp. 39-70.Joey Chew et all, 2011
Nidhaleddine, B. C. 2012. “Asymmetric exchange rate pass-through in the Euro area: New
evidence from smooth transition models”, Economics - The Open-Access, Open-Assessment
E-Journal, Kiel Institute for the World Economy, vol. 6(39), pages 1-28.
Oinonen, S., M. Paloviita and L. Vilmi 2013. “How have inflation dynamics changed over time?
Evidence from the euro area and USA”, Bank of Finland Research Discussion Papers 6
Quandt, Richard E. 1958. “The Estimation of Parameters of Linear Regression System Obeying
Two Separate Regimes”, Journal of the American Statistical Association 55, pp. 873-880.
240 BIFEC Book of Abstracts & Proceedings (Volume I Issue II)
Index of Authors
Ahmet DURAN, 130
Ahmet Faruk AYSAN, 1
Antonina WASZCZUK, 34
Burhaneddin İZGİ, 130
Chiara GUERELLO, 82
Evren ARIK, 66
Ferhat ARSLANER, 121
Gültekin GÖLLÜ, 1
Güzin Gülsün AKIN, 1
Hamid MOHTADI, 173
Levent YILDIRAN, 1
M. Vedat PAZARLIOĞLU, 196
Mehmet BALCILAR, 142
Nuran ARSLANER, 121
Orhan ERDEM, 66
Rıza DEMIRER, 142
Sadettin Haluk ÇİTÇİ, 110
Serkan YÜKSEL, 66
Sezer Bozkuş KAHYAOĞLU, 196
Stefan RUEDIGER, 173
Süleyman Hilmi KAL, 121