Embed Size (px)
Citation preview
Economic Modelling 43 (2014) 67–74
Contents lists available at ScienceDirect
Economic Modelling
j ourna l homepage: www.e lsev ie r .com/ locate /ecmod
Pass-through of oil prices to domestic prices: Evidence from anoil-hungry but oil-poor emerging market
Dinçer Dedeoğlu, Hüseyin Kaya ⁎Bahcesehir University, Department of Economics, Istanbul, Turkey
⁎ Corresponding author. Bahcesehir University, İ.İ.B.F., 3Tel.: +90 212 3810497; fax: +90 212 3810020.
E-mail addresses: [email protected],[email protected] (H. Kaya).
1 Approximately 90% of oil demand is imported (EMRAimports accounted for 236.5 billion US Dollars, and almosgoods including unprocessed materials incidental to indudental to industry, unprocessed fuels and oils, parts of invetation vehicles, unprocessed materials of food and beveraand beverages, and processed fuels and oils.
http://dx.doi.org/10.1016/j.econmod.2014.07.0380264-9993/© 2014 Elsevier B.V. All rights reserved.
a b s t r a c t
a r t i c l e i n f oArticle history:Accepted 30 July 2014Available online xxxx
Keywords:Oil pricesDomestic pricesPass-through
This paper aims to investigate the changes in the impact of world oil prices on consumer and producer prices inTurkey, an import dependent country in terms of crude oil and oil petroleum products. According to many re-searchers, the oil price pass-through to domestic prices has been decreasing recently. We estimate a recursiveVAR model on rolling windows to evaluate the changes in the pass-through of oil prices to domestic prices. Forthe period 1990–2012, we identify an increasing trend in the pass-through of oil prices to domestic prices inTurkey. The increasing pass-throughmay be attributed to the changes in relative prices: As oil becomesmore im-portant in the overall cost structure, firms becomemore responsive to its price. The results also suggest that theimpact of oil prices on producer prices is almost two times higher than the impact on consumer prices.Moreover,the gap between the oil price pass-through to producer prices and the pass-through to consumer prices increasedduring this period.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
A large amount of studies show that rising oil prices contributedto falling output and increased inflation during the 1970s and early1980s, and that falling oil prices boosted output and lessened inflationduring the mid to late 1980s (Brown et al., 1995). The moderationin oil prices during the 1980s and 1990s shifted the attention ofmacroeconomists from the study of oil prices. However, since 2000, oilprices have once again risen and taken on a central role in economic re-search. This time the volatility and the level of oil prices are higher. Asdiscussed in LeBlanc and Chinn (2004), in this environment the under-standing of the empirical linkage between oil prices and inflation is veryimportant not only for conducting a propermonetary policy but also forfirms to adopt their pricing policies. This linkage is crucial particularlyfor an emerging economy like Turkey that relies heavily on foreignsources of oil and imports a substantial level of petroleum containingproducts.1
Historically, the Turkish economy has been plagued by high andpersistent inflation. Diboğlu and Kibritçioğlu (2004) argue that one of
4353, Beşiktaş, Istanbul, Turkey.
, 2012). In 2012, Turkey's totalt 74% constituted intermediatestry, processed materials inci-stment goods, parts of transpor-ges, processed materials of food
the reasons for inflation in Turkey is increases in the prices of majorimported inputs. In addition, economic growth was not accompaniedby a stable macroeconomic environment, particularly in the periodfrom 1980 to 2000, and the implemented stabilization policies couldnot succeed. Since the severe financial crisis of February 2001, a struc-tural transformation process has been implemented involving notonly the transition to inflation targeting, but also the introduction ofthe floating exchange rate regime coupled with the new central banklaw, and structural reforms aimed at reducing the public sector burdenon the economy as well as promoting competition and productivity(Basci et al., 2008). These policies have been successful in decreasing in-flation from its historical high levels. However, the high and volatile na-ture of oil prices is one of the main challenges for the new monetarypolicy. In short, Turkey has a substantial current account deficit and aconsiderable part of this deficit is comprised of energy related imports.In such a case vulnerability to the exposed oil price and exchange rateshockswill be high,which in turn rarifies conducting a propermonetarypolicy. This dilemma puts forward the importance of oil price pass-through (OPPT) for countries like Turkey. Furthermore, a clear under-standing of OPPT can also be helpful for policymakers in deciding futureenergy policy.
Our aim is to investigate the OPPT to consumer and producer pricesfor the period 1990–2012. Although the OPPT is very important for de-veloping economies such as Turkey, the number of studies is relativelysmall. Thus, this paper serves to fill this gap by presenting new researchon Turkey. Second, after the financial crisis of February 2001, Turkeyswitched from a high inflation to a low inflation regime, so this studyallows us to test the hypothesis of Taylor (2000) which suggests thatthe OPPTwill be lower in a low inflation environment. Third, a majority
2 Leigh and Rossi (2002) and Kara and Öğünç (2008) use themethodology ofMcCarthy(1999) to evaluate the exchange rate pass-through in Turkey.
3 Making this replacement, themodel can be expressed and estimated as a VAR using aCholesky decomposition to identify the shocks.
68 D. Dedeoğlu, H. Kaya / Economic Modelling 43 (2014) 67–74
of the studies in the literature report that the OPPT has decreasedin many countries over time. This study enables us to examine whetheror not this is valid for Turkey.
To examine theOPPT between 1990 and 2012we employMcCarthy's(1999) model of pricing along a distribution chain, expressed andestimated as a VAR model. To investigate the change in the OPPT overtime we estimate this VAR model on rolling windows, i.e. we employ arolling VAR approach.
The rest of the paper is organized as follows. Section 2 introduces theliterature review. Section 3 describes the model and the data. Section 4reports the empirical results. Section 5 presents a discussion of the find-ings. Section 6 provides a conclusion and policy implications.
2. Literature review
The literature on the impact of oil prices on inflation is mainly dom-inated by studies of developed economies with the majority of themfinding that the impact of oil prices on domestic prices has decreasedover time. Hooker (2002) examines the pass-through issue using quar-terly U.S. data from 1962:Q2 to 2000:Q1. Estimating a Phillips curve, hefinds that the OPPT has been negligible since 1980. De Gregorio et al.(2007) examine 34 developed and developing countries employing aHooker-type approach, as well as rolling VARs, by using quarterly dataspanning from 1965:Q1 to 2005:Q1. According to their results, a declinein theOPPT is a generalized feature ofmany of the 34developed and de-veloping countries. Blanchard andGali (2007) estimate regular VARs forthe U.S., France, the U.K., Germany, Italy, Japan and rolling VARs for theU.S. They find that the dynamic effects of oil shocks have decreased con-siderably over time, which confirms the results of the aforementionedauthors. Chen (2009) estimates a time varying OPPT coefficient usinga Phillips curve for 19 industrialized countries between 1970:Q1 and2006:Q4. The results indicate that the degree of the OPPT varies acrosscountries and is positively correlated with energy imports. In addition,the pass-through coefficient is unstable over time, declining recently.Shioji and Uchino (2010) examine the impact of oil prices on Japanesedomestic prices using a time varying VAR model. They find that theOPPT declined for the 1980–2000 period but increased in the 2000swhen oil prices were on the rise. Valcarcel and Wohar (2013) estimatea Bayesian structural VAR that allows for time-varying parameters andstochastic volatility using the U.S. quarterly data from 1948:Q1 to2011:Q2. Their results imply that the OPPT has become negligiblesince the 1980s.
Researchers propose various reasons for the change in the OPPT.According to De Gregorio et al. (2007) the following factors help to ex-plainwhy oil shocks have had limited inflationary effects: the reductionin oil intensity of economies, a reduction in exchange rate pass-through,a more favorable inflationary environment (price changes by firms takeplace less frequently by rarifying oil price increases to pass through todomestic prices easily) and the fact that the current oil price increaseis largely the result of strong world demand. Blanchard and Gali(2007) show that credible monetary policy, greater wage flexibilityand change in industrial structure have contributed to this decline.Chen (2009) proposes that an appreciation of domestic currency, amore activemonetary policy in response to inflation and a higher degreeof trade openness aremajor causes of the recent decline in the OPPT. Thestudy argues that energy intensity may have played a minor role in theevolution of pass-through overtime. Blinder and Rudd (2008) point outthree widely considered causes of the decline in the OPPT, namely, in-creased credibility of monetary policy, greater wage flexibility andchanging industrial structure. According to Bernanke et al. (1997), oneof the reasons for the low OPPT is tight monetary policy.
Examination of the literature on OPPT reveals that the contributionof studies with respect to emerging economies, in particular Turkey,remains rather limited. Kibritçioğlu and Kibritçioğlu (1999) examinethe pass-through for Turkey between1986:01 and1998:03 by employinga VARmodel. They are unable tofind any evidence in favor of a significant
pass-through effect. However, Berument (2002) used Turkey's input-output table from 1990. Their evidence suggests that when wages andthe other three factors of income (profit, interest and rent) are adjustedto the general price level that includes oil price increases, the inflationaryeffect of oil prices becomes significant.
3. Model and data
In order to examine the pass-through of oil prices to domestic priceswe use a five variable recursive VAR model based on the methodologyof McCarthy (1999).2 The model contains the following endogenousvariables: oil prices, output gap, nominal exchange rate to the U.S.Dollar, producer prices and consumer prices. The methodology relieson a model of pricing along a distribution chain in which inflation inperiod t is comprised of components of expected inflation based onthe available information at the end of period t− 1 and effects of periodt supply, demand, external exchange rate, producer prices and inflationshocks.We assume that supply shocks are identified by the dynamics ofoil price inflation in local currency and that demand shocks are identi-fied by the dynamics of the output gap after considering the contempo-raneous effect of supply shocks. Furthermore, external exchange rateshocks are identified by taking into account contemporaneous supplyand demand shocks. Structural shocks are recovered from the VAR re-siduals using the Cholesky decomposition of the variance–covariancematrix. The system can be represented as follows:
πoilt ¼ Et−1 πoil
t
� �þ εoilt ð1Þ
eyt ¼ Et−1 eyoilt
� �þ εoilt þ εt
ey ð2Þ
Δet ¼ Et−1 Δetð Þ þ εoilt þ εtey þ εΔet ð3Þ
πppit ¼ Et−1 πppi
t
� �þ εoilt þ εt
ey þ εΔet þ εppit ð4Þ
πcpit ¼ Et−1 πcpi
t
� �þ εoilt þ εt
ey þ εΔet þ εppit þ εcpit ð5Þ
where πtoil is oil price (in nominal U.S. Dollars) inflation, eyt is output gap,Δet is change in logarithm of the nominal exchange rate relative to theU.S. Dollar, πtppi is producer price inflation rate, πtcpi is consumer price in-
flation rate, and εoilt ; εtey; εΔet ; εwpi
t ; εcpit are oil price inflation, output gap,change in exchange rate, producer price inflation and consumer priceinflation rate shocks respectively. Et− 1 denotes the expectation of a var-iable conditional on information available at period t− 1. In estimationthe expectations are introduced to themodel by linear projections of thelags of the variables.3 The systemallows for tracing thedynamic effect ofoil price shocks on consumer prices along the supply chain starting fromreal output, moving to the exchange rate, then to the producer pricesthat contain a relatively high proportion of tradable goods and finallyto the consumer prices that contain a smaller proportion of tradablegoods (Leigh and Rossi, 2002).
We use monthly data covering the period 1990:01–2012:02. Aver-age crude oil price in terms of nominal U.S. Dollars is obtained fromGEM commodities published by theWorld Bank. The nominal exchangerate relative to the U.S. Dollar, consumer price index (CPI), producer
69D. Dedeoğlu, H. Kaya / Economic Modelling 43 (2014) 67–74
price index (PPI) and seasonally adjusted industrial production index(IPI) series are obtained from the International Financial Statistics pub-lished by the International Monetary Fund. CPI and PPI are seasonallyadjusted using the Census X12 method. Output gap is derived from IPIby using the Hodrick–Prescott (HP) filter.
4. Results
Since many authors have documented that the OPPT to domesticprices changes over time, in estimation we investigate how the pass-through evolves over time in Turkey. As discussed in Shioji and Uchino(2010) and Chen (2009), considering a gradual change in the pass-through is more plausible than allowing for one-point structural change.
To evaluate the gradual change of the OPPTwe employ a rolling VARapproach. Themain advantage of the rolling VAR estimationmethodol-ogy is that it is an unstructured way of analyzing parameter changesand instability over time (De Gregorio et al., 2007). We estimate ourrecursive VAR model on rolling windows of 120 months.4 The periodof the first window is from 1990:01 to 2000:12 and the last one isfrom 2002:03 to 2012:02. In total the VAR model is estimated for 147windows.
As in Shioji and Uchino (2010), we derive estimates of the cumula-tive pass-through coefficient from the impulse response functions. Theestimates of the oil inflation pass-through coefficient are calculated asfollows:
PTt;tþs ¼DPt;tþs
OPt;tþsð6Þ
where PTt,t + s is the pass-through rate of oil at time horizon s in periodt, DPt,t + s is the cumulative impulse response of domestic price to an oilshock at horizon s in period t and OPt,t + s is the cumulative impulseresponse of oil to an oil shock at horizon s in period t. Since all impulseresponses are cumulative responses, they are the responses of the loglevel of CPI and PPI to one standard deviation oil price shocks.
Instead of enforcing the same lag length for all estimation windows,in each window we determine the lag length by information criterion.In the VAR literature usually one of the well-known information criteriasuch as Akaike Information Criterion (AIC), Schwarz InformationCriterion (SC) or Hannan–Quinn Information Criterion (HQ) is used todetermine the optimal lag length. Rather than relying on only one infor-mation criterion, in each window we estimate VAR based on AIC, SCand HQ separately. Thus, we calculate three pass-through coefficientsfor all windows and then take the average of these three estimations.We believe that this estimation method produces more reliable resultsby reducing the model's uncertainty. We determine the maximum laglength as 8.
In Figs. 1 and 2 we plot the estimated OPPT to CPI and PPI over the147 rolling windows for a 24 month horizon. In order to provide thetrend of the pass-through coefficient, we also plot a 12 month movingaverage of the pass-through coefficients. We find that the maximumvalue of the OPPT to CPI is nomore than 0.04 for the short time horizonand it gradually increases up to 0.085 for the longest time horizon.5 Thismeans that a 1% increase in oil prices leads to CPI inflation of at most0.04% in the short run and 0.085% in the long run. While for some esti-mation windows it turns out to be negative, in most cases it is positive.
4 Since the model includes five variables to have enough degrees of freedomwe set thewindow size as 120. We also consider smaller window size. When the window size is de-creased the general conclusion is not changed but the volatility of the pass-through coef-ficient is increased.
5 We also divide the sample into two sub-samples as pre-2002 and post-2002 periodand estimate the VAR model for both of these two periods. We find that in the pre-2002period responses of producer and consumer prices to the one standard deviation shockin oil prices are negligible. However, in the post-2002 period, the responses turn out tobe significantly positive.
It shows a sharp decline around the estimation window of 1994:04–2004:03.
5. Discussion
This sharp decline in theOPPT is not a surprise because Turkey expe-rienced a currency crisis in April 1994 in which interest rates and ex-change rates skyrocketed. More specifically, in April 1994, CPI inflationand PPI inflation rates soared beyond 20% and the Turkish Lira wasdevalued by about 50%.6 While for the 1 month time horizon a levelshift in the pass-through is present, for the longer time horizons thelevel shift tends to disappear. In general, despite some fluctuations,there seems to be an increasing trend in the OPPT to CPI.
Themagnitude of the OPPT to PPI is higher than to CPI. This finding isconsistent with the fact that the weight of the service sector in the CPIbasket is almost 50%. The maximum value of the OPPT to PPI is around0.08–0.10 in the short time horizon and increases to around 0.15–0.16in the longer timehorizon.While it fluctuates over estimationwindows,the increasing trend of the pass-through to PPI is obvious.
Jongwanich and Park (2011) argue that the gap between these twopass-through rates depends on the ability of producers to pass highercosts on to consumers. In addition to this, under inflation targetingmonetary policy central banks usually set targets on consumer prices,and with the help of monetary policy, they have the ability to reducethe producer price pass-through to consumer prices. When the oilprice pass-through gap between these two prices is investigated overthe estimation windows, it can be seen that the gap has an increasingtrend.7 Turkey has been implementing inflation targeting since2002:01, and this may be one of the reasons for the increasing trendof this gap. On the other hand, the increase in competition8 and low in-flationary environment may decrease the level of producer price pass-through to consumer prices.
It is also worth consideringwhether or not Shioji and Uchino (2010)argument provides an explanation for the increasing oil price pass-through in Turkey. They claim that as oil becomes cheaper it becomesa less important cost item forfirms and thus, they naturally decide to re-spond less to its price changes. Consistent with their claim they findthat, for the period 1980–2000, the main driving force behind the de-cline in pass-through is the price level of oil itself, and for the 2000s,when oil prices were on the rise, pass-through increased.
In Fig. 4, we plot oil prices and the import price of Turkey. As anindicator for import prices we use import unit value index which is col-lected from the IFS. From the figure, it is obvious that import prices andoil prices are highly correlated (correlation is almost 0.95). Additionally,we aim to visualize the trend of import/GDP ratio during this period.In Fig. 5, we plot the ratio of import to GDP. Unfortunately, in Turkeythere is no unique GDP series that is available for the whole periodof 1990–2012.9 Because of that we calculate import ratio by usingtwo available GDP series; the first one is available for the period1987–2007 and the second one is available beginning in 1998. It isclear that there is an increasing trend in the import/GDP ratio whichsuggests that the weight of imported goods in production has beenincreasing over time. (See Fig. 3.)
Moreover, we investigate how the costs of labor and capital, whichare the main cost items for firms, are changing over time. In Figs. 6, 7and 8, we plot the unit labor cost in manufacturing, the real interestrates and real oil prices for Turkey, respectively. The data for unit labor
6 For 1994 currency crises in Turkey see Celasun (1998).7 Gap figures are provided in the Fig. 9.8 The Turkish international trade to GDP ratio was 50.3% on average for the period
2001–2012, up from 40% for the period 1990–2000.9 The earlier national account estimates of GDP (base year 1987) in Turkey were pre-
pared in accordance with the United Nations System of National Account. The new GDPseries (base year 1998) have been prepared in accordance with the European System ofAccounts (ESA-95). With the new methodology the estimates of GDP have increased ap-proximately 30% on average.
-.005
.000
.005
.010
.015
.020
.025
.030
25 50 75 100 125
@MOVAV(C1,12) C1
-.03
-.02
-.01
.00
.01
.02
.03
25 50 75 100 125
@MOVAV(C2,12) C2
-.05
-.04
-.03
-.02
-.01
.00
.01
.02
.03
25 50 75 100 125
@MOVAV(C3,12) C3
-.06
-.05
-.04
-.03
-.02
-.01
.00
.01
.02
.03
25 50 75 100 125
@MOVAV(C4,12) C4
-.08
-.06
-.04
-.02
.00
.02
.04
25 50 75 100 125
@MOVAV(C5,12) C5
-.08
-.06
-.04
-.02
.00
.02
.04
25 50 75 100 125
@MOVAV(C6,12) C6
-.10
-.08
-.06
-.04
-.02
.00
.02
.04
25 50 75 100 125
@MOVAV(C7,12) C7
-.10
-.08
-.06
-.04
-.02
.00
.02
.04
.06
25 50 75 100 125
@MOVAV(C8,12) C8
-.10
-.08
-.06
-.04
-.02
.00
.02
.04
.06
25 50 75 100 125
@MOVAV(C9,12) C9
-.10
-.08
-.06
-.04
-.02
.00
.02
.04
.06
25 50 75 100 125
@MOVAV(C10,12) C10
-.10
-.08
-.06
-.04
-.02
.00
.02
.04
.06
25 50 75 100 125
@MOVAV(C11,12) C11
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C12,12) C12
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C13,12) C13
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C14,12) C14
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C15,12) C15
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C16,12) C16
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C17,12) C17
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C18,12) C18
-.12
-.08
-.04
.00
.04
.08
25 50 75 100 125
@MOVAV(C19,12) C19
-.12
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C20,12) C20
-.16
-.12
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C21,12) C21
-.16
-.12
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C22,12) C22
-.16
-.12
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C23,12) C23
-.16
-.12
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C24,12) C24
Denotes the oil price pass-through to CPI
Denotes 12 month moving average of the documented pass-through
Fig. 1. Oil price pass-through to CPI. Denotes the oil price pass-through to CPI. Denotes 12 month moving average of the documented pass-through.
70D.D
edeoğlu,H.K
aya/Econom
icModelling
43(2014)
67–74
.01
.02
.03
.04
.05
.06
25 50 75 100 125
@MOVAV(C1,12) C1
.01
.02
.03
.04
.05
.06
.07
.08
.09
25 50 75 100 125
@MOVAV(C2,12) C2
.00
.01
.02
.03
.04
.05
.06
.07
.08
.09
25 50 75 100 125
@MOVAV(C3,12) C3
-.02
.00
.02
.04
.06
.08
.10
25 50 75 100 125
@MOVAV(C4,12) C4
-.04
-.02
.00
.02
.04
.06
.08
.10
25 50 75 100 125
@MOVAV(C5,12) C5
-.04
-.02
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125
@MOVAV(C6,12) C6
-.08
-.04
.00
.04
.08
.12
25 50 75 100 125
@MOVAV(C7,12) C7
-.050
-.025
.000
.025
.050
.075
.100
.125
.150
25 50 75 100 125
@MOVAV(C8,12) C8
-.050
-.025
.000
.025
.050
.075
.100
.125
.150
25 50 75 100 125
@MOVAV(C9,12) C9
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C10,12) C10
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C11,12) C11
-.08
-.04
.00
.04
.08
.12
.16
.20
25 50 75 100 125
@MOVAV(C12,12) C12
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C13,12) C13
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C14,12) C14
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C15,12) C15
-.08
-.04
.00
.04
.08
.12
.16
.20
25 50 75 100 125
@MOVAV(C16,12) C16
-.08
-.04
.00
.04
.08
.12
.16
.20
25 50 75 100 125
@MOVAV(C17,12) C17
-.08
-.04
.00
.04
.08
.12
.16
.20
25 50 75 100 125
@MOVAV(C18,12) C18
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C19,12) C19
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C20,12) C20
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C21,12) C21
-.08
-.04
.00
.04
.08
.12
.16
25 50 75 100 125
@MOVAV(C22,12) C22
-.10
-.05
.00
.05
.10
.15
.20
25 50 75 100 125
@MOVAV(C23,12) C23
-.10
-.05
.00
.05
.10
.15
.20
25 50 75 100 125
@MOVAV(C24,12) C24
Denotes the oil price pass-through to PPIDenotes 12 month moving average of the documented pass-through
Fig. 2. Oil price pass-through to PPI. Denotes the oil price pass-through to PPI. Denotes 12 month moving average of the documented pass-through.
71D.D
edeoğlu,H.K
aya/Econom
icModelling
43(2014)
67–74
-.05
.00
.05
.10
.15
.20
.25
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
CPI Inflation
-.05
.00
.05
.10
.15
.20
.25
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
PPI Inflation
% %
1994:04
2001:04
1994:04
2001:04
Fig. 3. CPI inflation and PPI inflation.
.1
.2
.3
.4
.5
.6
.16
.20
.24
.28
.32
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
import/GDP(1) import/GDP(2)
Imp
ort/
GD
P(2
)
Imp
ort/
GD
P(1
)
Fig. 5. Import/GDPa.
0
40
80
120
160
60
80
100
120
140
160
180
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Oil prices Import price index
Oil
pric
es
Impo
rt p
rice
inde
x
Fig. 4. Oil price and import unit value index.
140
150
160
170
180
190
200
210
220
230
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Fig. 6. Unit labor cost index (1985 = 100).
0
10
20
30
40
50
60
70
80
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
%
Fig. 7. Real interest rate.
0.8
1.2
1.6
2.0
72 D. Dedeoğlu, H. Kaya / Economic Modelling 43 (2014) 67–74
cost is obtained from TISK (2001). While there are huge fluctuations inthe labor cost during the period, in the 2000s it remained unchanged.This indicates that, for firms in the 2000s when oil prices were high,there was no significant increase in the labor cost. The same is alsoevident for the cost of capital. We use real interest rate as a proxy forthe cost of capital.10 From Fig. 7 we can see that the average level ofreal interest rates in the 1990s is higher than the average level in the2000s. Finally, real oil price for Turkey11 (Fig. 8) has been increasing
10 Interest rate series over the period of 1990–2012 is collected from the IMF. Wecalculate the real rate using the Fisher equation. The calculation of real rate is as follows:[(1+ moneymarket rate) / (1+ inflation)]− 1. Since the expected inflation is not avail-able during this period, we use current inflation rate as a proxy.11 Real oil price is calculated by using the following formula: Real Oil Price =Nominal Oil Price × Nominal Exchange rate × (CPIUS/CPITR). CPI for United States isobtained from OECD.
0.0
0.4
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Fig. 8. Real oil price.
.01
.02
.03
.04
.05
.06
25 50 75 100 125
@MOVAV(C1,12) C1
.02
.03
.04
.05
.06
.07
.08
25 50 75 100 125
@MOVAV(C2,12) C2
.02
.03
.04
.05
.06
.07
.08
25 50 75 100 125
@MOVAV(C3,12) C3
.02
.03
.04
.05
.06
.07
.08
25 50 75 100 125
@MOVAV(C4,12) C4
.02
.03
.04
.05
.06
.07
.08
25 50 75 100 125
@MOVAV(C5,12) C5
.02
.03
.04
.05
.06
.07
.08
25 50 75 100 125
@MOVAV(C6,12) C6
.02
.03
.04
.05
.06
.07
.08
.09
25 50 75 100 125
@MOVAV(C7,12) C7
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C8,12) C8
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C9,12) C9
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125
@MOVAV(C10,12) C10
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125
@MOVAV(C11,12) C11
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125
@MOVAV(C12,12) C12
.00
.02
.04
.06
.08
.10
.12
25 50 75 100 125
@MOVAV(C13,12) C13
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C14,12) C14
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C15,12) C15
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C16,12) C16
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C17,12) C17
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C18,12) C18
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C19,12) C19
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C20,12) C20
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C21,12) C21
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C22,12) C22
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C23,12) C23
.01
.02
.03
.04
.05
.06
.07
.08
.09
.10
25 50 75 100 125
@MOVAV(C24,12) C24
Denotes the oil price pass-through to PPI
Denotes 12 month moving average of the documented pass-through
Fig. 9. The gap between oil-price pass-through to CPI and PPI.
73D.D
edeoğlu,H.K
aya/Econom
icModelling
43(2014)
67–74
74 D. Dedeoğlu, H. Kaya / Economic Modelling 43 (2014) 67–74
over time. Taking the increases in real oil price, import prices and theimport/GDP ratio alongwith the decreases in the costs of labor and cap-ital into consideration, we argue that oil became a more important costitem over time for firms in Turkey. Hence, as discussed in Shioji andUchino (2010), when oil price becomes a more important cost item,producers may decide to respond more to oil price changes.
We should also note that, in order to clarify this argument, examina-tion of the input–output table is necessary. The latest input–outputtable available in Turkey is for the year 2002when oil prices just startedto increase. Accordingly, to see the impact of increase in oil prices oncost structure clearly an updated input–output table is required.
6. Conclusion and policy implications
This study examines the evolution of the OPPT to domestic prices inTurkey. We find that the OPPT to domestic prices has increased overtime. It is also evident that the gap between the OPPT to CPI and PPIhas increased over time. As discussed in Shioji and Uchino (2010), theincreases in the impact of oil prices may be due to the change in therelative price of oil, in that it becomes a more important cost item forfirms. Thus, firms become more responsive to its price changes.
Turkey is a heavily oil dependent country and imports high amountsof oil based products. It has a substantial current account deficit and aconsiderable part of this deficit is comprised of energy imports, appar-ently making Turkey vulnerable to both oil price and exchange rateshocks. Turkey's growth is increasing, causing an increasing need forenergy. This increasing need shackles it with respect to the properimplementation of energy saving policies. In addition, it is worthwhileto note that Turkey has adopted an inflation targeting monetary policy.Taking these factors, alongwith the empirical results, into considerationreveals that the OPPT is very important for Turkey in energy, fiscal andmonetary policy designation issues. In such an environment appropri-ate policy recommendations for Turkey should include the following:First, the proportion of oil in the cost structure of firms should be re-duced. Second, to decrease the OPPT without hampering economicgrowth, energy efficiency policies rather than energy saving policiesand policies encouraging renewable energy usage should be employed.Since the Kyoto Protocol has become binding for Turkey, energy effi-ciency and renewable energy usage issues have become more impor-tant. Third, although it is not a major energy producer due to itsgeographical and geopolitical location, Turkey acts as a bridge andan outlet for transporting energy from Russia, the Caspian Sea region
and the Middle East to world markets. This strategic position can betranslated into energy cost cutbacks in return for Turkey's contributionsto energy security issues.
References
Basci, E., Özel, Ö., Sarikaya, C., 2008. The monetary transmission mechanism in Turkey:new developments. BIS Papers chapters In: Bank for International Settlements(Ed.), Transmission Mechanisms for Monetary Policy in Emerging Market Economiesvol. 35. Bank for International Settlements, pp. 475–499.
Bernanke, B.S., Gertler, M., Watson, M., 1997. Systematic monetary policy and the effectsof oil price shocks. Brook. Pap. Econ. Act. 1, 91–142.
Berument, H., 2002. Inflationary effect of crude oil prices in Turkey. DepartmentalWorking Papers 0203. Bilkent University, Department of Economics.
Blanchard, O.J., Gali, j, 2007. The macroeconomic effects of oil shocks: why are the 2000sso different from the 1970s? NBER Working Paper Series.
Blinder, A.S., Rudd, J.B., 2008. The supply-shock explanation of the Great Stagflationrevisited. NBER Working Paper no. 14563 (December).
Brown, S.P.A., Oppedahl, D.B., Yücel, M.K., 1995. Oil prices and inflation. Working Papers95-10. Federal Reserve Bank of Dallas.
Celasun, O., 1998. The 1994 currency crisis in Turkey. World Bank Policy ResearchWorking Paper, No:1913.
Chen, S.S., 2009. Oil price pass-through into inflation. Energy Econ. 31, 126–133.De Gregorio, J., Landerretche, O., Neilson, C., 2007. Another pass-through bites the dust?
Oil prices and inflation. Working Papers Central Bank of Chile 417. Central Bank ofChile.
Diboğlu, S., Kibritçioğlu, A., 2004. Inflation, output growth, and stabilization in Turkey,1980–2002. J. Econ. Bus. 56 (1), 43–61.
EMRA, 2012. Turkish Energy Market: An Investor's Guide, www.epdk.gov.tr.Hooker, M.A., 2002. Are oil shocks inflationary? Asymmetric and nonlinear specifications
versus changes in regime. J. Money Credit Bank. 34, 540–561.Jongwanich, J., Park, D., 2011. Inflation in developing Asia: pass-through from global food
and oil price shocks. Asian-Pacific Econ. Lit. 25, 79–92.Kara, H., Öğünç, F., 2008. Inflation targeting and exchange rate pass-through: the Turkish
experience. Emerging Markets Finance Trade 44 (6), 52–66.Kibritçioğlu, A., Kibritçioğlu, B., 1999. Ham petrol ve akaryakıt ürünü fiyat artışlarının
Türkiye'deki enflasyonist etkileri. Working Paper Series, No. 21. Turkish RepublicUndersecreteriat of Treasury.
LeBlanc, M., Chinn, M.D., 2004. Does high oil prices presage inflation? The evidence fromG-5 countries. Bus. Econ. 2, 38–48.
Leigh, D., Rossi, M., 2002. Exchange rate pass-through in Turkey. Working Paper 02/204.International Monetary Fund, Washington, DC.
McCarthy, J., 1999. Pass-through of exchange rates and import prices to domestic inflationin some industrialized economies. BIS Working Paper, No: 79. Bank for InternationalSettlements, Basel.
Shioji, E., Uchino, T., 2010. Pass-through of oil prices to Japanese domestic prices. NBERWorking Paper No. 15888.
Taylor, John B., 2000. Low inflation, pass-through, and the pricing power of firms. Eur.Econ. Rev. 44, 1389–1408.
TISK, 2001. 2011 yılı çalışma istatistikleri ve işgücü maliyetinin tisk araştirma servisincedeğerlendirilmesi. www.tisk.org.tr.
Valcarcel, V.J., Wohar, M.E., 2013. Changes in the oil price-inflation pass-through. J. Econ.Bus. 68, 24–42.
