Asymmetric exchange rate and oil price pass-through in motor fuel market: A microeconometric approach

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

The Journal of Economic Asymmetries journal homepage: www.elsevier.com/locate/jeca

Asymmetric exchange rate and oil price pass-through in motor fuel market: A microeconometric approach☆ ⁎

Mustafa Utku Özmen , Fatih Akçelik Central Bank of the Republic of Turkey, Research and Monetary Policy Department, Istiklal Caddesi, No:10, Ulus, 06050 Ankara, Turkey

A R T I C L E I N F O

ABSTRACT

JEL classification: D22 D43 E31

We analyze the response of retail motor fuel prices to oil price and exchange rate changes. Using a novel microdata approach considering price spells separately; we find evidence for passthrough asymmetry in Turkish motor fuel market based on the sign, source and size of cost shocks. Exchange rate (oil price) is the main factor fueling asymmetry in case of cost increases (decreases). The smaller the magnitude of positive cost shock the higher the pass-through is. Sign asymmetry is reversed during crisis. Market structure is suggested as the main explanation of the asymmetry, yet there are factors limiting the use of market power.

Keywords: Oil price Exchange rate Micro data Motor fuel Pass-through asymmetry Turkey

1. Introduction In most of the non-oil producing countries, motor fuel prices fluctuate in response to changes in international crude oil prices. Motor fuel products are also generally subject to high proportion of consumption tax, which makes final consumer prices much higher. Therefore, the share of motor fuel in total expenditures of the households is at non-negligible levels, especially in developing economies. Not surprisingly, price changes in motor fuels, more specifically price increases, receive public attention and cause discomfort at times. Such sort of public reaction is also evident in Turkey, where the price of motor fuel is one of the most expensive in the world.1 A general perception is that decreases in crude oil prices in domestic currency are not reflected on prices as much as increases. Therefore, the question of whether the response of retail motor fuel prices to changes in crude oil prices in domestic currency is asymmetric or not in Turkey still remains open. Moreover, Turkey provides a good example to analyze the impact of oil prices as being an oil importer emerging economy with a volatile exchange rate which also plays a key role in pricing. In this study, acknowledging the shortcomings of time series methods and considering the high frequency of price changes in motor fuel sector, we propose a microeconometric approach to better identify the pass-through from oil price and exchange rate, and to study the evidence of asymmetry. Our strategy builds on two assumptions: price setters use all the available information rationally and the price elasticity of demand in this sector is zero in the short run. Given these, we can treat each price spell (the period between two consecutive price changes) of motor fuels as separate micro observations and associate them with the cumulative change in the

☆ The views and opinions presented in this study belong to the authors and do not necessarily represent those of the Central Bank of the Republic of Turkey or its staff. ⁎ Corresponding author. E-mail addresses: [email protected], [email protected] (M. Utku Özmen), [email protected] (F. Akçelik). 1 According to Bloomberg, Turkey has 9th highest gasoline price among 61 countries. Also, Turkey is ranked 7th in terms of portion of a day's wage needed to buy a gallon of gasoline.

http://dx.doi.org/10.1016/j.jeca.2017.02.002 Received 3 October 2016; Received in revised form 16 February 2017; Accepted 16 February 2017 1703-4949/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Özmen, M.U., The Journal of Economic Asymmetries (2017), http://dx.doi.org/10.1016/j.jeca.2017.02.002

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

domestic price of crude oil over the same period of time. With this, we are able to achieve a clean identification by exactly matching the size of cost changes with the size of retail price changes. Another contribution of our study is on the grounds of the composition of domestic crude oil prices, which many studies do not take into account. For a small open economy, not only the international crude oil price but also the exchange rate changes frequently. Therefore, distinguishing the sources of change in crude oil prices in domestic currency may provide important information on the nature of the asymmetry and on the pricing strategy of the market. Micro data based identification approach enables us to investigate the asymmetry and differences in pricing behavior in more depth in fact. The asymmetric response to cost shocks need not only be in terms of the sign of shock. In addition, there might be asymmetry regarding the source and the size of the cost change in response to shocks. Therefore, in this study we do not restrict our analysis only to sign based asymmetry but we further consider other possible dimensions of response difference to cost shocks in order to capture a more complete view on the pricing dynamics of the motor fuel market. Also, for several reasons, this market provides a good working environment to study pricing behavior. Few large firms dominate the market where a homogeneous good is sold. This removes the quality bias between the goods to a large extent. Moreover, major cost items (international oil price and the exchange rate), which are the main determinants of short term price movements, are publicly known and easily followed. This sector is also a very good example of state-dependent pricing where firms adjust their prices in a cost-based manner. Finally, prices in this sector are very flexible compared to other goods and services in the economy. Hence, our approach not only enables us to investigate the sign-based asymmetry, but also to study the pricing behavior in general, considering other aspects of cost pass-through to consumer prices. Our study contributes to the literature by applying the proposed identification strategy and by distinguishing the sources of the change in domestic crude oil prices for an emerging market economy. Our results point to presence of asymmetry in motor fuel market from various aspects in Turkey. First, pass-through of changes in domestic price of crude oil is asymmetric on overall retail motor fuel prices. For tax free price of diesel, the increases in crude oil prices are reflected at a higher proportion than decreases, meanwhile, there is no significant asymmetry in gasoline. Second, pass-through of cost changes is higher if the source of the rise (decline) is the exchange rate (oil price) when domestic price of crude oil increases (decreases). Third, when we consider positive cost shocks, the higher the size of change in domestic currency value of oil prices, the lower the degree of pass-through is. Also, the asymmetry is reversed during crisis periods, where the pass-through of negative cost shocks is higher. Although we focus on Turkey, our results may also shed light on the discussions of asymmetric price response especially in developing countries where exchange rate plays an important role in price setting. This study not only presents evidence of asymmetry via a careful identification strategy, but also provides insights for the micro pricing behavior of a market that displays almost perfect price flexibility. The rest of the study is organized as follows: Section 2 presents major price setting theories and Section 3 reviews the related literature on the asymmetric oil price pass-through. Section 4 presents data, methodology, estimation strategy and the results. The discussion of the results is presented in Section 5. Section 6 concludes the study. 2. Theory Theories on optimal price setting abound. Given the presence of some sort of rigidities, (i.e. information costs, menu costs, consumer reaction) firms may not be able to instantly optimize and reset prices. Among the major theories, state-dependent pricing (SDP) theories stand out when the motor fuel market is considered. SDP models consider the cost of price updating or adjustment. In principle, the models reveal that the expected gains from changing the price should at least offset the adjustment cost incurred. Some early examples of SDP include Barro (1972), Sheshinski and Weiss (1977), Caplin and Spulber (1987) and Dotsey, King, and Wolman (1999). Costly price adjustments are mostly introduced through menu-cost models. Sheshinski and Weiss (1977) show that the optimal price setting strategy for a firm follows an (s,S) type under an environment where monopolistic firms set the price of a perishable good. Here, if the size of the cost shock remains in the (s,S) band, it is optimal not to change the price for the firm. The model is further extended by Sheshinski and Weiss (1983) and Danziger, (1983, 1984) among others. In a recent strand of literature, there are SDP models where the timing of the price change is also modeled along with the size of price change under a profit maximization setting. These models also extend the previous generation of models on the grounds of providing a reason for small-sized price changes which are observed empirically (i.e. Klenow & Kryvtsov, 2008). Examples of these studies include Golosov and Lucas (2007), Gertler and Leahy (2008), Midrigan (2011), Dotsey, King, and Wolman (2013) and Vavra (2014). 3. Overview of related literature In the literature, whether the pass-through of crude oil price change to retail price is asymmetric or not has extensively been studied. Yet, there is no consensus regarding asymmetry. While Borenstein, Cameron, and Gilbert (1997) find asymmetric response of retail gasoline prices to crude oil prices in the US, Bachmeier and Griffin (2003) show symmetric behavior of retail gasoline prices to crude oil price shocks by using daily data. Lewis (2011) develops a search model of asymmetric adjustment and shows that the fit of the model with the data is better than previous explanations in favor of asymmetric response. However, Radchenko (2005) argues that asymmetry in gasoline prices is not coherent with search theory. Chen, Finney, and Lai (2005) support asymmetric adjustment in retail gasoline prices in the US by using threshold co-integration methods. Dunis, Laws, and Evans (2006) also find asymmetric adjustment by applying threshold co-integration tests to NYMEX 2

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

futures contract daily closing prices and unleaded gasoline prices. Grasso and Manera (2007) show asymmetric adjustment in retail prices at the distribution stage for five main European countries by applying asymmetric error correction models. Al-Gudhea, Dibooglu, and Kenc (2007) also provide evidence of asymmetric transmission between upstream and downstream prices and show that the asymmetry is more apparent for small shocks by applying threshold and momentum models of co-integration for the US by using daily data. The controversy over the asymmetric response of retail gasoline price to crude oil prices is evident in recent studies as well. For example, Balaguer and Ripollés (2012) find symmetric response of gasoline and diesel prices to international wholesale oil prices by using asymmetric error correction model and daily data for Spain. However, Schalck, Lamotte, Porcher, and Silvestre (2013) show asymmetric response of gasoline and diesel prices by using ARDL method and weekly data for France, another large European market. Several studies analyze the motor fuel market in Turkey. Alper and Torul (2009) find asymmetric response of retail gasoline price to world crude oil prices by using SVAR method and monthly data from 1991 to 2007. Authors argue that this asymmetric behavior is a result of the frequently changing tax rate. Berument, Sahin, and Sahin (2013) find asymmetric effects of domestic currency value of oil price and exchange rate to pre-tax retail motor fuel prices with error correction model, by using weekly data from 2005 to 2011. However, this effect is not the same as what is referred to as asymmetric response of motor fuel prices to crude oil prices in the literature. One aspect of the lack of consensus in literature may well be related to the use of different methodologies and to divergence in the frequency of price observations across studies. Regarding frequency, first, we see that monthly, weekly and daily data are used. Using monthly data for analyzing the asymmetry of the retail motor fuel price response against an oil price change is not sufficient as the fuel prices change very rapidly. Hence, most of the time more than one price change is recorded within a month, which makes the identification of the response difficult if the interim changes in prices are not observed. Although earlier studies focused mostly on monthly observations, recent studies increasingly employ more frequent data. While use of weekly data is also common in the literature, several studies point to the lack of robustness of results obtained with weekly data. Bachmeier and Griffin (2003) contrast results from daily and weekly data for the response of US retail gasoline prices to crude oil prices. They show that using daily data notably weakens the price asymmetry results shown by Borenstein et al. (1997). Moreover, Bettendorf, van der Geest, and Varkevisser (2003) form five weekly data sets with a distinct starting working day for daily retail gasoline and Rotterdam oil prices in Dutch market. They show that the price asymmetry evidence from weekly data changes across datasets. Balaguer and Ripollés (2012) also show that the results critically depend on the choice of working day for Spanish fuel market. These results provide evidence for the lack of robustness of results obtained from weekly data and emphasize the importance of using daily data in empirical studies. Overall, the main problem of using aggregate data (weekly or monthly) is the inevitable selection of a starting day and averaging over a period. However, these concerns are only related to the use of time series techniques. Using daily data offers several advantages. It is possible to observe the actual prices of oil and motor fuel rather than prices averaged over a week or month. But more importantly, by using daily data more information is available including exact duration of each price spell and exact size of each price change, where the price spell is defined as the interval in which the price remains unchanged. This information provides important insights on pricing behavior of firms as well. However, commonly used time series techniques in asymmetry analysis are not very tractable even in the presence of daily data, given that the identification is problematic since prices change very often. For instance, if the price of motor fuel increases in day 1 and decreases in day 2, then, it is not easy to track which price change in oil prices passed through to motor fuel prices in each day. More directly, before the pass-through of an oil price shock to motor fuel price is completed, it is most likely that a new shock will occur. Thus, measuring the impact of changes becomes a difficult task. In this regard, a new approach based on identification of each price spell changes the nature of the problem from a time series analysis to a micro framework where instead of time dimension, the major unit of measurement becomes individual price spells for motor fuels. Such an approach will also form a bridge between state-dependent pricing theory and real-life observations. 4. Empirical analysis 4.1. Data Different from the existing literature, in this study we propose a microeconometric approach to study asymmetry with the aim of achieving a clear identification and separation of shocks to crude oil prices in domestic currency. Therefore, we construct individual price spells calculated via daily observations. In the study, we use daily data from January 1, 2006 to February 14, 2014 for tax free retail unleaded 95 octane gasoline and daily data from July 1, 2008 to February 14, 2014 for tax free diesel prices. The data are composed of the prices of four leading distributor companies and is received from the EMRA.2 In terms of determinants, we collect the daily price of Brent oil in USD, the USD/TL exchange rate and the special consumption tax excised on motor fuels in Turkey. Brent crude oil price is received from Bloomberg, the USD/TL exchange rate is from the Central Bank of Turkey, and the special consumption tax data is from the Ministry of Finance. Finally, we calculate the domestic crude oil prices by multiplying the Brent price in USD and the USD/TL exchange rate. Once the daily dataset is available, we can calculate the duration of each individual motor fuel price spell. The distribution of domestic motor fuel price spells for Turkey is presented in Fig. 1. The distribution of price spells presents an interesting picture. The first point to note is that duration of domestic gasoline and 2

The daily prices are those effective in the European district of Istanbul, which is the most populated area in Turkey.

3

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

.06 0

.02

.04

Density

.08

.1

M. Utku Özmen, F. Akçelik

0

10

20 Duration (days)

30

40

Fig. 1. Distribution of Domestic Motor Fuel Price Durations (in days). Notes: The horizontal axis shows the number of days a price remains unchanged. The vertical axis shows the density of each frequency. Only prices remaining less than 40 days are plotted. These observations account for 96% of the observations.

diesel prices is very low. Median duration of a price is 9 days and mode of the distribution duration is only 7 days. Second, almost one fourth of domestic motor fuel prices survived only four days or less. Motor fuel prices, therefore, differ from other consumer goods and services in terms of flexibility.3 Such a high frequency of price changes also sheds light on the pricing behavior of the motor fuel market in Turkey. Erdogdu (2014) provides evidence for the inelasticity of motor fuel demand to price changes in Turkey. Given these and the oligopolistic nature of the market, cost-push factors are dominant when motor fuel prices are determined.4 That is, state-dependent pricing is dominant in this market. Thus, we may focus on domestic oil price (oil prices in domestic currency) changes as the main determinant of domestic motor fuel price changes. Fig. 2 presents the histograms of the size of motor fuel price changes and the size of corresponding cost shocks that led the tax-free motor fuel prices to change. As can be seen from Fig. 2, cost shocks, in terms of changes in oil price in domestic currency, occur in both directions and the distribution of shocks are almost symmetric (left panel). Meanwhile, the distribution of domestic motor fuel price changes is more condensed than that of the changes in domestic currency value of oil prices (right panel). Comparing two panels in Fig. 2, it is easily observed that there is no one-to-one pass-through from oil prices in local currency to final consumption price of motor fuel in domestic market. Also, extreme price changes in domestic motor fuels are less frequent when compared to changes in domestic currency value of Brent oil price. From another perspective, even though a large number of cost shocks in the vicinity of zero percent are observed, the corresponding domestic motor fuel price changes are rarely near zero. Looking only at the right panel of Fig. 2 may lead one to consider the presence of menu costs. Reconciling both figures suggests that firms indeed respond to small-sized shocks as well, even at larger magnitudes, possibly to compensate for the menu costs. Finally, the descriptive statistics are shown in Table 1. We see that, for instance for the case of gasoline, there are more upward (510) price changes than downward (411) changes. Also, the median absolute size of increases is higher than that of price decreases. The results also point to different behavior of firms. For instance, the number of increases in the price of gasoline is higher for Firm 2, whereas, Firm 3 registered more price increases compared to others in case of diesel. The mean and median price durations also differ somewhat across firms. Before going into estimation methodology, a final remark needs to be made regarding the tax policy on motor fuels. There are two types of tax excised on domestic motor fuels in Turkey: Special Consumption Tax (SCT) and Value Added Tax (VAT). SCT changes from time to time but the VAT rate is constant at 18% over the sample period. Taxes play an important role in the final sales price. Therefore, even though we consider the tax free prices, the ongoing special consumption tax rate may also influence tax free prices.5 4.2. Empirical methodology Empirical methodology builds on the nature of data and on state-dependent pricing theory. Focusing on high frequency of price changes, we employ a strategy that treats each motor fuel price spell individually. We take each episode of change in the tax free price of gasoline and diesel as separate micro observations and associate these changes with cumulative percent change in domestic currency price of crude oil. Specifically, for instance, if the price of gasoline which is set on March 10th is changed on March 15th, then this would be recorded in our data as a 5-day-cumulative change in gasoline price matched with the corresponding 5-daycumulative change in cost. This is equivalent to the idea that firms adjust prices when the cost shocks cumulate up to a point. 3 Özmen and Sevinç (2016) show that the average duration of consumer prices is around two months in Turkey. Considering that, we can say that motor fuel prices change much more frequently than overall basket of goods and services. 4 According to EMRA (2013), four leading distribution companies have 78.4% market share of gasoline and 66.5% market share of diesel in Turkey. The market share of each of four companies is above 10% for gasoline and these four companies have more than two thirds of whole motor fuel market sales. 5 According to EMRA (2013), total taxes constitute around 60% of gasoline price and around 53% of diesel price in 2013 in Turkey.

4

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

Fig. 2. Distribution of Cost Shocks and Price Changes (percent). a. Distribution of Domestic Oil Price Changes. Notes: The horizontal axis shows percent change of oil price in domestic currency over the period between two consecutive price changes of motor fuels. The vertical axis shows the density of each frequency. Only changes over the range of 20% are plotted. b. Distribution of Motor Fuel Price Changes. Notes: The horizontal axis shows percent change in tax-free motor fuel prices. The vertical axis shows the density of each frequency. Only changes over the range of 20% are plotted. Table 1 Descriptive statistics of domestic motor fuel prices. Duration of Price (in days)

Price Changes Median (percent)

# of observation

Firm

Mean

Median

# of obs.

Increase

Decrease

Increase

Decrease

Gasoline

Total #1 #2 #3 #4

12.84 14.01 11.80 13.47 12.31

9 9 8 9 9

921 210 251 220 240

3.69 3.88 3.50 3.65 3.82

−3.04 −3.30 −3.07 −3.00 −3.00

510 116 141 123 130

411 94 110 97 110

Diesel

Total #1 #2 #3 #4

12.96 13.47 13.50 12.07 12.91

9 9 9 8 9

633 152 152 170 159

3.31 3.35 3.40 3.10 3.29

−3.13 −3.18 −3.21 −3.11 −3.11

335 77 79 95 84

298 75 73 75 75

5

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

4.2.1. Sign asymmetry In order to detect the possible asymmetric response of motor fuel prices to changes in domestic currency oil price, we first consider the following specification based on micro observations:

∆gasi = α ∆brent _tli + β∆brent _tliD1i + γ ∆scti + εi

(1)

where, i refers to individual price spell; Δgas is tax free motor fuel price change; ∆brent_tl is the cumulative change in the domestic currency value of oil price; D1 is dummy variable taking the value of 1 if ∆brent_tl > 0 (a positive cost shock) and 0 otherwise (a negative cost shock);α is the pass-through parameter for the case where the domestic value of oil price decreases; ∆sct is cumulative change in special consumption taxes on gasoline or diesel price (to control for tax effects) and εi is the error term. β represents the degree of asymmetry in the response of domestic motor fuel prices to positive domestic currency oil price shocks in comparison to negative cost shocks. Thus, at this stage, we define the sign asymmetry as the difference in the size of the pass-through in cases of positive and negative cost shocks. 4.2.2. Source asymmetry As discussed previously, an important point to consider for oil importing small open economies is that the domestic currency price of oil is composed of two main components. Hence, the response of domestic motor fuel prices to cost shocks may differ according to the source of the shock, either coming from international oil price or from the exchange rate. Next, we consider extended specifications for domestic currency oil price increases and decreases. In order to identify whether separate effects of oil price and exchange rate are asymmetric or not in the case of cost increases (∆brent_tli > 0) we consider:

∆gasi = α1∆brent _tli + β1∆brent _tl D2i + β2∆brent _tl D3i + γ ∆scti + εi i

i

(2)

where, D2 is a dummy variable taking the value of 1 if ∆brent_usdi > 0 and ∆usdi < 0, and 0 otherwise; D3 is a dummy variable taking the value of 1 if ∆brent_usdi < 0 and ∆usdi > 0, 0 otherwise; where ∆brent_usd is the cumulative percent change in Brent oil price in USD and ∆usd is the cumulative percent change in USD/TL exchange rate (an increase refers to depreciation of TL). Thus, with this specification, we decompose the episodes of positive cost shocks (increase in domestic currency oil price) into three categories: Here,α1 measures the pass-through parameter for the case where both the oil price and the exchange rate increase, which is the base category in this specification. Meanwhile β1( β2 ) captures the additional asymmetry effect in the case of increasing oil price (exchange rate) and declining exchange rate (oil price). Similarly, we consider another specification to identify whether separate effects of Brent oil price and exchange rate are asymmetric or not when domestic value of oil price decreases (∆brent_tli < 0):

∆gasi = α2 ∆brent _tli + β3∆brent _tl D4i + β4∆brent _tl D5i + γ ∆scti + εi i

i

(3)

Here, D4 is a dummy variable taking the value of 1 if ∆brent_usdi < 0 and ∆usdi > 0, 0 otherwise; D5 is dummy a variable taking the value of 1 if ∆brent_usdi > 0 and ∆usdi < 0, 0 otherwise. Similarly, here, α2 is the pass-through parameter for the case where both oil price and exchange rate decrease, which is the base category in this specification. β3( β4 ) captures the additional asymmetry effect in the case of decreasing oil price (exchange rate) and increasing exchange rate (oil price). 4.2.3. Size asymmetry The pass-through from domestic value of oil price may be in nonlinear form. We consider an alternative specification to identify whether the response of domestic motor fuel prices to oil price in TL differentiate with the magnitude of cost change when domestic value of oil price increases (∆brent_tli > 0):

∆gasi = α ∆brent _tli + β∆brent _tli ^2+ γ ∆scti + εi

(4)

where, ∆brent_tli ^2 is the squared cumulative change in domestic value of Brent oil price. If α is significantly positive and β is significantly negative, it can be interpreted that as the magnitude of cost shock increases, the level of pass-through decreases. This, then, may indicate that there is pass-through asymmetry depending on the size of cost shock. 4.3. Results 4.3.1. Sign asymmetry We first consider Specification 1 in order to detect the pass-through asymmetry between increase and decrease in domestic Brent oil prices. Results are reported in Table 2. Accordingly, a statistically significant asymmetry is detected for the motor fuel prices in Turkey over the sample period. While 42% (α ) of negative cost shocks are passed-through the domestic motor fuel prices, 49% (α + β ) of positive cost shocks are passed-through. This pass-through asymmetry is more strongly evident for diesel. The size of asymmetry is almost 10% points. Meanwhile, although positive, the asymmetry coefficient is not statistically significant for the case of gasoline over the sample period. 4.3.2. Source asymmetry As a novelty of our estimation strategy, we are able to separate the sources of shock, from the international oil price or from the 6

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

Table 2 Pass-through asymmetry of cost shocks to domestic motor fuel prices by type of motor fuel (Specification 1) a. Coefficients

Motor Fuel (Total)

Gasoline

Diesel

β

0.0714** (0.0350) 0.422*** (0.0247) −0.608*** (0.0582) 1554 0.391

0.0569 (0.0495) 0.449*** (0.0355) −0.787*** (0.0995) 921 0.408

0.0926** (0.0426) 0.383*** (0.0248) −0.485*** (0.0536) 633 0.375

α

γ Observations R-squared

a The cost shock is defined as the percent change in the value of oil price in domestic currency. β is the degree of asymmetry in the response of domestic motor fuel prices to positive domestic currency oil price shocks in comparison to negative cost shocks. α is the pass-through parameter for the case where the domestic value of Brent oil price decreases. γ is the coefficient of special consumption tax. Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

Table 3 Descriptive statistics of changes in domestic value of Brent oil price (percent) a. Sign of change

Cases

Mean

Standard Deviation

# of obs.

Increase in domestic currency Brent oil price

Oil Price ↑, USD/TL↑ Oil Price ↑, USD/TL↓ Oil Price ↓ , USD/TL↑

5.52 4.23 2.57

3.49 3.66 2.91

384 357 97

Decrease in domestic currency Brent oil price

Oil Price ↓, USD/TL↓ Oil Price ↓, USD/TL↑ Oil Price ↑, USD/TL↓ Total

−5.30 −3.74 −1.36 0.53

4.04 3.98 1.51 5.85

360 276 72 1554

a The table lists all the possible cases for overall cost increases and decreases where the effect is decomposed into international oil price and exchange rate changes. An increase in USD/TL refers to depreciation of Turkish lira.

exchange rate. The descriptive statistics regarding such a classification of cost changes are reported in Table 3. In the following, we estimate Specification 2 and 3 for motor fuels in order to test whether the source of cost shock introduces a pass-through asymmetry. Estimation results of Specification 2 and 3 are reported in Tables 4 and 5, respectively. Accordingly, when positive cost shocks are considered (domestic value of Brent oil price increases), exchange rate is the source of asymmetry. As seen in Table 4, in the case of both oil price and exchange rate increasing, 46% (α ) of the shock is transmitted into motor fuel prices. Meanwhile, in the case of international oil price declining but exchange rate increasing, then 66% (α+β2 ) of the positive cost shock is transmitted. For the case of negative cost shocks (domestic value of Brent oil price decreases), instead, the results in Table 5 show that Brent oil is the source of asymmetry. Here, when both international oil price and exchange rate falls, 36% of the cost reduction is reflected onto motor fuel prices. Whereas, when the oil price is decreasing, despite a rise in exchange rate, a much higher proportion (54%) of the cost reduction is reflected onto motor fuel prices. 4.3.3. Size asymmetry Next, we consider the relationship between the size of shock and the level of pass-through. One may assume that firms may not Table 4 Pass-through asymmetry by the source of shock (in case of ∆brenttl >0 ) (Specification 2) a. Coefficients

Motor Fuel

β1

0.0568 (0.0579) 0.201*** (0.0597) 0.456*** (0.0275) −0.272*** (0.0972) 843 0.350

β2

α1

γ Observations R-squared

a α1 is the pass-through parameter for the case where both oil price and exchange rate increase. β1 is the asymmetry effect in the case of increasing oil price and declining exchange rate. β2 is the asymmetry effect in the case of increasing exchange rate and decreasing oil price. γ is the coefficient of special consumption tax. Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

7

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

Table 5 Pass-through asymmetry by the source of shock (in case of ∆brenttl <0 ) (Specification 3) a. Coefficients

Motor Fuel

β3

0.182*** (0.0411) 0.135 (0.108) 0.357*** (0.0222) −0.745*** (0.0592) 711 0.480

β4

α2

γ Observations R-squared

a α2 is the pass-through parameter for the case where both oil price and exchange rate decline. β3 is the asymmetry effect in the case of declining oil price and increasing exchange rate. β4 is the asymmetry effect in the case of decreasing exchange rate and increasing oil price. γ is the coefficient of special consumption tax. Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

Table 6 Nonlinear pass-through by the size of cost shock (Specification 4)a. Dependent variable: ∆gas

Motor Fuel

β

−0.0230*** (0.00724) 0.716*** (0.0662) −0.301*** (0.0940) 843 0.364

α

γ Observations R-squared

a β is the coefficient of the square of cumulative change in domestic value of Brent oil price. α is the coefficient of cumulative change in domestic value of Brent oil price. γ is the coefficient of special consumption tax. Robust standard errors are in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

be able to reflect the cost shock to motor fuel prices with the same mark-up independent of the size of the shock. In order to detect evidence of nonlinear pass-through asymmetry depending on the size of cost shock, we estimate Specification 4 for the case of cost increases, i.e. a rise in domestic currency value of oil price. The results are reported in Table 6. We see that as the magnitude of the cost shock increases, the degree of pass-through to domestic motor fuel price decreases (since α is significantly positive and β is significantly negative). According to the coefficients in Table 6, for instance, when domestic currency value of oil rises by 1%, 5% and 10%, corresponding motor fuel price pass-through rates are 69%, 60% and 49% on average, respectively. We may analyze further the asymmetry in pass-through of different sized cost shocks. Table 7 presents the pass-through of cost shocks where shocks are grouped according to their magnitude. For the positive cost shocks which are of the magnitude of 2% or less, the pass-through coefficient is 1.08. Meanwhile, the coefficient comes down as the size of shock increases (drops to 0.38 for shocks greater than 10%). However, such a marked difference is not evident for cost decreases. Finally, we consider whether the level of pass-through differs with the state of general economic activity. Our sample period includes the effects of economic slowdown over 2008. The Table 8 presents the estimation results of Specification 1 for different sample periods, i.e. before, during and after the crisis. When we consider the observations belonging to periods before and after the Table 7 Pass-through by the size of cost shock a. α

Size of Cost Shock (Percent, Interval) Positive Cost Shocks

(0,2] (2,5] (5,10] > 10

1.08 0.69 0.51 0.38

Negative Cost Shocks

[−2,0) [−5,−2) [−10,−5) < −10

0.39 0.60 0.45 0.35

a εi where, α is The cost shock is defined as the percent change in the value of oil price in domestic currency. Our specification is: ∆gasi = α ∆brent _tli + γ ∆scti + the pass-through parameter for corresponding sample of prices changes, where the cost shocks fall into that specific interval. Thus, each row corresponds to estimation results for a different sample. All coefficients are statistically significant at 1%.

8

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

Table 8 Pass-through by the state of the economy (Specification 1) a. Coefficients

Before Crisis

During Crisis

After Crisis

β

0.149** (0.0749) 0.374*** (0.0524) −1.021*** (0.0175) 257 0.429

−0.144* (0.0742) 0.398*** (0.0434)) −0.940*** (0.00284) 266 0.385

0.0991** (0.0388) 0.478*** (0.0295) −0.593*** (0.0579) 1031 0.408

α

γ Observations R-squared

a The cost shock is defined as the percent change in the value of oil price in domestic currency. β is the degree of asymmetry in the response of domestic motor fuel prices to positive domestic currency oil price shocks in comparison to negative cost shocks. α is the pass-through parameter for the case where the domestic value of Brent oil price decreases. γ is the coefficient of special consumption tax. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Crisis period is defined as the period from 2008Q2 to 2009Q1.

crisis, the evidence of positive asymmetry is more pronounced. However, during the crisis period (2008Q2-2009Q1), negative asymmetry is observed: cost decreases are reflected more than the cost increases to final consumer prices. 5. Discussion In this section, we shed more light on the empirical findings of the previous section and discuss the results. With our flexible and novel estimation strategy we are able to detect asymmetry in the reflection of cost shocks onto consumer motor fuel prices. Our findings on asymmetry can be summarized under respective categories: sign asymmetry (in terms of the sign of the cost shock), source asymmetry (differing pass-through rates depending on the origin of the cost shock), size asymmetry (pass-through depending on the size of shock) as well as state asymmetry (pass-through depending on the state of economy). Main finding of the study points to asymmetry in the pass-through of cost shocks in terms of the sign. Over the sample period in consideration, increases in domestic currency oil price are reflected on motor fuel prices with a higher proportion than decreases in the domestic currency oil price in Turkey. In the literature, the most common theoretical explanations for asymmetry in motor fuel market are collusion of firms (i.e. Borenstein et al., 1997) and consumer search behavior (Cabral & Fishman, 2012; Lewis, 2011; Yang & Ye, 2008). Among the studies in empirical literature, Johnson (2002) and Lewis (2011) provide evidence for search-based explanation of asymmetry. On the contrary, other studies provide evidence for the relevance of market power being effective in response asymmetry, i.e. Verlinda (2008) and Deltas (2008). Also, Adams (1997) claims that search cost is less effective in gasoline, as an example of homogenous items, compared to in-store items. In a recent study, Brewer, Nelson, and Overstreet (2014) also find that the retailers in motor fuel market only make reasonable positive profits in months of cost reduction, providing additional evidence for market power. In the context of our study we argue that the market power argument is also valid for Turkey. In items basis, our results suggest that there is no pass-through asymmetry in gasoline, but pass-through asymmetry exists in diesel in Turkey. The reason behind this difference may be attributed to the structural change of motor fuel demand in the Turkish market. With this structural change, a rapid transition of consumers from cars operating with gasoline to cars operating with diesel and LPG has taken place. According to TurkStat, the market share of automobiles operating with diesel, which was 9.5% in 2006, increased to 27.9% in 2014.6 Meanwhile, in this period, the market share of automobiles operating with gasoline decreased from 62.5% to 30.2%.7 Thus, even though firms respond asymmetrically in the overall motor fuel market, it seems that they are able to act so only in the sector where they face an increasing demand (diesel). Hence, considering the oligopolistic nature of the market in Turkey and the difference in terms of response asymmetry between gasoline and diesel, we may conclude that the evidence from Turkey also supports the market power argument being relevant for explaining asymmetric pass-through of positive and negative cost changes. Second finding emerging from our analysis is that the pass-through of cost shocks also differs with the source of the change -which may be referred as source asymmetry. That is, in the domain of upward cost changes (when domestic currency value of oil price increases), the pass-through is higher in cases where only exchange rate increases (depreciation) in comparison to other cases. On the contrary, in the domain of downward cost changes (when domestic currency value of oil price decreases), the pass-through is higher in cases where only the oil price (Brent oil price in USD) decreases, in comparison to other cases. Consequently, we may conclude that pass-through asymmetry is exacerbated if the source of change is the exchange rate for upward cost shocks and if it is the international oil price for downward cost shocks. This finding may be linked to two explanations: First, economy wide cost shocks may be reflected more onto consumer prices than sector specific cost shocks. Beck, Hubrich, and Marcellino (2011) suggest that region or country level shocks explain a higher portion of the variation in prices than sector level shocks. Also, Ashenfelter, Ashmore, Baker, and McKernan (1998) and Gron and Swenson (2000) show that the pass-through of industry wide shocks are considerably higher than firm specific shocks. It would not be misleading if we form an analogy between sector/firm specific and economy/sector specific shocks. In our case, a change in the exchange rate is an economy wide phenomenon, meanwhile, the change 6 7

Value for 2014 refers to May 2014. Also, the market share of automobiles operating with LPG increased from 24.8% in 2006 to 41.4% in 2014.

9

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

in international oil prices is only a motor fuel sector specific phenomenon, at least in the short run. Hence, the pass-through of economy wide shock being higher is intuitive in that sense. Second, consumer behavior is thought to play an important role in such a response asymmetry. On theoretical grounds, consumer anger at price increases (Rotemberg, 2005) and consumers’ beliefs about the fairness of the price (Rotemberg, 2011) may affect the decisions of the firms. From this perspective, we argue that when the cost increase stems from the exchange rate, then, firms in motor fuel market are able to pass-through a higher proportion of the cost shock to consumers. On the contrary, when the cost increase is due to a rise in international oil prices, which is a sector specific shock, firms cannot pass-through the shock as much as in the previous case. The idea behind is that when exchange rate increases (depreciation of the local currency), many prices in the economy are adjusted and the motor fuel sector does not receive a special public attention. However, when only oil price increases, the sector is subject to public attention and criticism as prices are adjusted only there. Economy wide shock may be considered a fair reason to adjust prices, yet a sector specific shock may not be considered as fair by consumers. Third finding relates to shock size asymmetry. As the magnitude of cost shock increases, in case of positive cost shock, degree of pass-through decreases. In other words, there is a nonlinear pass-through by the magnitude of cost increases, as outlined in Table 7. However, no such a marked difference is observed when different thresholds of the size of cost decreases are considered. Combining these results and the evidence from Fig. 2, we see that for smaller positive cost shocks, there is perfectly complete pass-through. However, for large cost increases and for cost decreases, the pass-through is incomplete. An evaluation of the above discussion suggests that even if the market power of the firms in the motor fuel retail sector enables them to asymmetrically pass cost increases and decreases, there are also factors restraining their market power, depending on the source and the size of the cost shock. The fair price perception and consumer anger, along with focused public attention may be considered as major factors against the full use of market power, given the evidence that firms can pass-through less of the cost increases if the rise is fueled by oil prices or if the size of the cost increase is large. Fourth, our findings reveal that the sign of pass-through asymmetry also differs with the state of the general economic activity. The presence of the significant slowdown in the economic activity over 2008 enables us to identify periods referring to different states of the economy. Our findings show that during the crisis period (2008Q2-2009Q1), the asymmetry turns out to be negative: cost decreases are reflected more than the cost increases to final consumer prices. This suggests that the state of economic activity may well be another factor limiting the market power of the firms in motor fuel market. Finally, our results point to incomplete pass-through of tax changes onto consumer prices. Even though tax-free motor fuel price is analyzed, when the tax changes are controlled for, the coefficient of special consumption tax is significantly negative in all specifications. This result can be interpreted as firms not fully reflecting the tax changes immediately to domestic motor fuel prices, another possible sign of reduced market power of the firms. Before we conclude, a discussion of the findings of the recent literature on pass-through asymmetry in other countries similar to Turkey may help put our results into a more general context. There are several studies, analyzing domestic motor fuel markets and focusing on whether the response of prices to changes in the cost of crude oil the symmetric or not, conducted for several net oil importer small open economies. Most of these studies analyze the asymmetry over various aspects in addition to the classical question of whether positive cost shocks pass-through to domestic prices more than cost decreases. Few examples come from Eastern European countries. Leszkiewicz-Kędzior and Welfe (2014) study the nature of price adjustments in the Polish fuel market. They show that there is asymmetric price response in Poland at different stages of the market, i.e. European wholesale, Polish wholesale and the domestic retail market. Especially in the retail market, asymmetry is observed in the speed of price adjustment to the long-run equilibrium and also in the size and duration of the price response. Thus, retailers raise the prices concurrently amid a cost increase; while they adjust prices less and over a longer period of time in case of wholesale cost decreases. Svacina (2013) provides evidence for asymmetry from gasoline market in Czech Republic, treating crude oil price (Brent) and exchange rate separately. The results of the study reveal that positive crude oil price changes are passed-through faster than negative crude oil price changes and that retail gasoline prices respond faster to crude oil price increases than changes in the exchange rate (depreciation). Also, author reports asymmetric response regarding the size of the shock. Gasoline prices respond faster to low exchange rate changes than high exchange rate changes. Polemis (2012), investigating the gasoline market in Greece, finds that the effects of price increases are larger than the impact of price decreases. This study also includes the exchange rate into the analysis. In the wholesale market, the spot price of the gasoline is found to be responding to exchange rate depreciation (an increase in the cost of imported oil), but not to exchange rate appreciation (a reduction in the cost of imported oil). In a recent study, Bagnai and Ospina (2016) consider the response of domestic gasoline markets to shocks to imported oil price and to exchange rates separately, focusing on European countries. They show that in many countries, including Greece, Spain and Italy, exchange rate depreciations are reflected more than exchange rate appreciations. Thus, firms are able to pass cost increases more on consumers. Meanwhile, they show that the asymmetry is reversed when the response of gasoline prices to crude oil prices is considered. This finding is somewhat similar to our findings which show that the pass-through of negative cost shocks is higher when the driver of the change is the fall in crude oil price. Several studies provide evidence from Asian oil-importing countries. Pal and Mitra (2016) analyze the oil products market in India, focusing on asymmetry both in sign and size of cost changes. In terms of the magnitude, authors consider crude oil price changes in different quantiles and show that the prices of oil products respond more to changes in the higher quantiles than to changes in lower quantiles. Consequently, the pass-through to prices of oil products is higher in case of a cost increase than in case of a cost reduction. Meanwhile, Salas (2002) studies the asymmetric response of retail gasoline prices to changes in crude cost in domestic currency in the Philippines. The author finds that there is a significant asymmetry regarding price adjustment in the face of crude cost changes. The cumulative adjustment of prices following a cost increase is higher than that of cost decreases. The study 10

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

also analyzes the market structure by separating large and small firms in the retail gasoline market, and shows that smaller firms engage in a price-follower strategy. In a recent study, Chou and Tseng (2016), on the other hand, provide evidence for a negative asymmetry in the Taiwanese gasoline market. Authors reveal that gasoline prices adjust more quickly to cost decreases than cost increases. Also, they show that the main source of asymmetry is the exchange rate where the observed negative asymmetry is stronger for changes in the exchange rate than in the oil price. Authors argue that political economy concerns affect the pricing decisions of the sector. Balmaceda and Soruco (2008) examine the nature of the asymmetry in Chilean gasoline market, providing an example from Latin America. Authors show that the gasoline prices respond asymmetrically to cost changes: retail prices respond quicker to cost increases than cost decreases. They also distinguish between the type of gas stations, as being branded or unbranded, and show that the branded stations exert a higher asymmetry, which is argued to be as a result of higher market power. Overall, the asymmetric response of domestic retail motor fuel prices to cost changes is a widely reported outcome for oil importing small open economies, with few exceptions only. Also, the asymmetry is not only limited to the sign of the change, but also extends to magnitude and source of the shock, speed of adjustment, size and type of the retail firms, and the state of the economy. 6. Conclusion In this paper, we investigate the oil price and exchange rate pass-through asymmetry in retail motor fuel market in Turkey. Different from previous studies, we propose a new approach based on micro price observations and corresponding cost factors. Our results show that oil price increases pass-through more than decreases to motor fuel prices. However, the pass-through asymmetry is evident only for diesel, but not for gasoline. Our approach also enables us to analyze different aspects of pricing asymmetry. In terms of the source of shock, we show that asymmetry strengthens if the source is the exchange rate (oil price) in the presence of positive (negative) cost shock. Also, as the size of shock increases, the pass-through from oil prices decreases in case of positive cost shocks. In terms of the state of the economy, our empirical analysis indicates that the asymmetry is reversed during crisis periods. Overall, the asymmetry findings suggest that the oligopolistic nature of the motor fuel market in Turkey enables firm to exercise a certain level of market power. On the other side, other findings regarding source and size of shocks, and the state of the economy suggest that there are limitations for the use of market power in price setting. The results of the study offer several policy implications. The oligopolistic nature of the market stems from the existence of entry barriers and high start-up costs in oil products retail sector. To eliminate the asymmetric pricing behavior, market regulatory institutions may play an active role. The results obtained regarding the role of exchange rates might be generalized to other net oil importer small open economies which are financially integrated with the global economy and are prone to exchange rate fluctuations. This further emphasizes the important role of exchange rate in price setting in emerging market economies. References Adams, A. F. (1997). Search costs and price dispersion in a localized, homogeneous product market: Some empirical Evidence. Review of Industrial Organization, 12, 801–808. Al-Gudhea, S., Dibooglu, S., & Kenc, T. (2007). Do retail gasoline prices rise more readily than they fall? A threshold cointegration approach. Journal of Economics and Business, 59, 560–574. Alper, C. E., & Torul, O. (2009). Asymmetric adjustment of retail gasoline prices in Turkey to world crude oil price changes: The role of taxes. Economics Bulletin, 29, 775–787. Ashenfelter, O., Ashmore D., Baker J., and McKernan S. (1998). Identifying the Firm-Specific Cost Pass-Through Rate, Federal Trade Commission Working Paper No. 217. Bachmeier, L. J., & Griffin, J. M. (2003). New evidence on asymmetric gasoline price responses. The Review of Economics and Statistics, 85(3), 772–776. Bagnai, A., & Ospina, C. A. M. (2016). “Asymmetric asymmetries” in Eurozone markets gasoline pricing. The Journal of Economic Asymmetries, 13, 89–99. Balaguer, J., & Ripollés, J. (2012). Testing for price response asymmetries in the Spanish fuel market. New evidence from daily data. Energy Economics, 34, 2066–2071. Balmaceda, F., & Soruco, P. (2008). Asymmetric dynamic pricing in a local gasoline retail market. The Journal of Industrial Economics, 56(3), 629–653. Barro, R. J. (1972). A theory of monopolistic price adjustment. The Review of Economic Studies, 39, 17–26. Beck, G., Hubrich, K., & Marcellino, M. (2011). On the importance of sectoral and regional shocks for price-setting. CEPR DP8357. Berument, H., Sahin, A. and Sahin, S. (2013). Crude Oil Price and Exchange Rate Effects of Asymmetry on Petroleum Product Prices: Evidence from a Set of Northern Mediterranean Countries, Available at SSRN, 〈http://ssrn.com/abstract=2215722〉. Bettendorf, L., van der Geest, S. A., & Varkevisser, M. (2003). Price asymmetry in the Dutch retail gasoline market. Energy Economics, 26(6), 669–689. Borenstein, S., Cameron, A. C., & Gilbert, R. (1997). Do gasoline prices respond asymmetrically to crude oil price changes? Quarterly Journal of Economics, 112, 305–339. Brewer, J., Nelson, D. M., & Overstreet, G. (2014). The economic significance of gasoline wholesale price volatility to retailers. Energy Economics, 43, 274–283. Cabral, L., & Fishman, A. (2012). Business as usual: A consumer search theory of sticky prices and asymmetric price adjustment. International Journal of Industrial Organization, 30(4), 371–376. Caplin, A. S., & Spulber, D. F. (1987). Menu costs and the neutrality of money. The Quarterly Journal of Economics, 102, 703–726. Chen, L. H. C., Finney, M., & Lai, K. S. (2005). A threshold cointegration analysis of asymmetric price transmission from crude oil to gasoline prices. Economics Letters, 89(2), 233–239. Chou, K. W., & Tseng, Y. H. (2016). Oil prices, exchange rate, and the price asymmetry in the Taiwanese retail gasoline market. Economic Modelling, 52, 733–741. Danziger, L. (1983). Price Adjustments with Stochastic Inflation. International Economic Review, 24, 699–707. Danziger, L. (1984). Stochastic inflation and the optimal policy of price adjustment. Economic Inquiry, 22, 98–108. Deltas, G. (2008). Retail gasoline price dynamics and local market power. Journal of Industrial Economics, 56, 613–629. Dotsey, M., King, R. G., & Wolman, A. L. (1999). State-dependent pricing and the general equilibrium dynamics of money and output. The Quarterly Journal of Economics, 114, 655–690. Dotsey, M., King, R. G., & Wolman, A. L. (2013). Inflation and real activity with firm-level productivity shocks (Working Paper No. 13-35). Philadephia FED, Research Department. Dunis, C. L., Laws, J., & Evans, N. (2006). Modelling and trading the gasoline crack spread: A non-linear story. Derivatives Use, Trading & Regulation, 12, 126–145.

11

The Journal of Economic Asymmetries xxx (xxxx) xxx–xxx

M. Utku Özmen, F. Akçelik

Energy Market Regulatory Authority (EMRA) (2013). Annual Petroleum Market Sector Report. Erdogdu, E. (2014). Motor fuel prices in Turkey. Energy Policy, 69, 143–153. Gertler, M., & Leahy, J. (2008). A Phillips curve with an Ss Foundation. Journal of Political Economy, 116, 533–572. Golosov, M., & Lucas, R. E., Jr. (2007). Menu costs and Phillips curves. Journal of Political Economy, 115, 171–199. Grasso, M., & Manera, M. (2007). Asymmetric error correction models for the oil-gasoline price relationship. Energy Policy, 35(1), 156–177. Gron, A., & Swenson, D. (2000). Cost pass-through in the U.S automobile market. Review of Economics and Statistics, 82(2), 316–324. Johnson, R. N. (2002). Search costs, lags and prices at the pump. Review of Industrial Organization, 20, 33–50. Klenow, P. J., & Kryvtsov, O. (2008). State-dependent or time-dependent pricing: does it matter for recent U.S. inflation? The Quarterly Journal of Economics, 123, 863–904. Leszkiewicz-Kędzior, K., & Welfe, A. (2014). Asymmetric price adjustments in the fuel market. Central European Journal of Economic Modelling and Econometrics, 2, 105–127. Lewis, M. S. (2011). Asymmetric price adjustment and consumer search: An examination of the retail gasoline market. Journal of Economics & Management Strategy, 20(2), 409–449. Midrigan, V. (2011). Menu costs, multiproduct firms, and aggregate fluctuations. Econometrica, 79, 1139–1180. Özmen, M. U., & Sevinç, O. (2016). Price rigidity in Turkey: Evidence from micro data. Emerging Markets Finance and Trade, 52(4), 1029–1045. Pal, D., & Mitra, S. K. (2016). Asymmetric oil product pricing in India: Evidence from a multiple threshold nonlinear ARDL model. Economic Modelling, 59, 314–328. Polemis, M. L. (2012). Competition and price asymmetries in the Greek oil sector: An empirical analysis on gasoline market. Empirical Economics, 43, 789–817. Radchenko, S. (2005). Oil price volatility and the asymmetric response of gasoline prices to oil price increases and decreases. Energy Economics, 27(5), 708–730. Rotemberg, J. J. (2005). Customer anger at price increases, changes in the frequency of price adjustment and monetary policy. Journal of Monetary Economics, 52(4), 829–852. Rotemberg, J. J. (2011). Fair pricing. Journal of the European Economic Association, 9(5), 952–981. Salas, J. I. S. (2002). Asymmetric price adjustments in a deregulated gasoline market. Philippine Review of Economics, 39(1), 38–71. Schalck, C., Lamotte, O., Porcher, T., & Silvestre, S. (2013). Asymmetric gasoline price responses in France. Applied Economics Letters, 20(5), 457–461. Sheshinski, E., & Weiss, Y. (1977). Inflation and costs of price adjustment. The Review of Economic Studies, 44, 287–303. Sheshinski, E., & Weiss, Y. (1983). Optimum pricing policy under stochastic inflation. The Review of Economic Studies, 50, 513–529. Svacina, D. (2013). Stickiness of gasoline prices in the Czech Republic: Evaluation of different models of price adjustment (Bachelor thesis). Charles University in Prague. Vavra, J. (2014). Inflation Dynamics and time-varying volatility: New evidence and an Ss interpretation. Quarterly Journal of Economics, 129(1), 215–258. Verlinda, J. A. (2008). Do rockets rise faster and feathers fall slower in an atmosphere of local market power? Evidence from the retail gasoline market. Journal of Industrial Economics, 56, 581–612. Yang, H., & Ye, L. (2008). Search with learning: Understanding asymmetric price adjustments. RAND Journal of Economics, 39, 547–564.

12