The Role of Imported Inputs in Pass-through Dynamics

Transcription

1. The Role of Imported Inputs in Pass-through Dynamics Dilara ERTU Ğ Pınar ÖZLÜ M. Utku ÖZMEN Çağlar YÜNCÜLER February 2020 Working Paper No: 20/03
2. © Central Bank of the Republic of Turkey 2020 Address: Central Bank of the Republic of Turkey Head Office Structural Economic Research Department Hacı Bayram Mh. İstiklal Caddesi No: 10 Ulus, 06050 Ankara, Turkey Phone: +90 312 507 80 04 Facsimile: +90 312 507 78 96 The views expressed in this working paper are those of the author(s) and do not necessarily represent the official views of the Central Bank of the Republic of Turkey.
3. The Role of Imported Inputs in Pass-through Dynamics1 Dilara Ertu ğ, Pınar Özlü, M. Utku Özmen and Çağlar Yüncüler Research and Monetary Policy Department Central Bank of the Republic of Turkey Abstract In this paper, we analyze the extent to which the use of imported inputs affects exchange rate and import price pass-through into domestic producer and consumer prices for services in Turkey. We first calculate the use of imported inputs on sectoral level by analyzing the input-output tables. Then, by taking the sectoral heterogeneity regarding the use of imported inputs into account, we estimate import price and exchange rate passthrough by utilizing import prices, producer prices (consumer prices for services) and output gap on sectoral basis. Our results point to a substantial heterogeneity across sectors in terms of exchange rate and import price pass-through. While the import price (in foreign currency) pass-through is in line with the share of imported input to a large extent, the pass-through of exchange rate shocks to domestic prices are generally higher than the share of imported inputs in costs inclusive of labor. Our findings also reveal that this excess exchange rate pass-through has strengthened over the recent period. Additional analyses carried out reveal that the high share of foreign currency debt is associated with higher exchange rate pass-through, suggesting that the management of foreign exchange liability might play a critical role to enhance the effectiveness of monetary policy and to create room for maneuver to fight against inflation by reducing the excess exchange rate pass-through. Keywords: Imported inputs; import price pass-through, exchange rate pass-through; FX liability JEL Codes: D57; E31; E52 1 The views expressed in this paper are only those of the authors and should not be interpreted as reflecting those of the Central Bank of the Republic of Turkey. Email addresses:Dilara.Ertug@tcmb.gov.tr; Pinar.Ozbay@tcmb.gov.tr, Utku.Ozmen@tcmb.gov.tr; Caglar.Yunculer@tcmb.gov.tr. 1
4. Non-technical Summary Over the last two decades , increased global trade and integration of local economies to the global value chains have led to a considerable rise in the share of imported inputs in domestic production. Imported input use has tended to grow over time both across countries and industries. As a byproduct, the elevated use of imported inputs has also served as a factor increasing the responsiveness of an economy to import price and exchange rate shocks. In this framework, this study investigates to what extent the use of imported inputs weigh on exchange rate and import price pass-through to domestic prices in Turkey. To better understand the dynamics, we build on a sectoral level analysis incorporating manufacturing and services sub-sectors. Our empirical results reveal that there is a considerable amount of heterogeneity across industrial sectors in terms of the size of the pass-through and that the pass-through of exchange rate shocks is in general higher than the pass-through of shocks to import price in USD. More importantly, we find that while the extent of import price pass through is in line with the share of imported inputs use across sectors, the exchange rate passthrough is higher than the share of imported inputs in most occasions. Results regarding the services point to several sectors with substantial level of exchange rate pass-through as well, where the share of imported input use is high. Although the use of imported inputs is the main source of exchange rate and import price pass-through, there are factors that can affect the extent to which foreign prices and exchange rates transmit into domestic producer prices. In our analysis, we highlight the foreign currency debt as a channel that might have driven the excess exchange rate pass-through (difference between the exchange rate pass-through and the share of imported inputs) observed in the recent period in Turkey. Our empirical results suggest that the excess exchange rate pass-through is associated with the high share of foreign currency debt in total assets, especially the short-term debt. Overall, our results suggest that the intensive use of imported inputs per se help explain the heterogeneity in the pass-through of import prices to a large extent. However, its relation with the exchange rate pass-through is not one-to-one, suggesting that other factors also weigh on exchange rate pass-through. Considering the possible impact of foreign currency debtedness, the management of foreign exchange liability turns out to be critical also for controlling inflation by reducing the excess exchange rate passthrough. 2
5. 1 . Introduction Increased globalization has made most of the countries more vulnerable to external factors, i.e. exchange rate and import price shocks, via the trade channel (Auer and Mehrotra, 2014). Accordingly, quantifying the extent of the impact of such factors has become more important for better forecasts of macroeconomic indicators, e.g. inflation, growth, and an optimal economic policy design in emerging market economies, whose integration to global value chains has increased significantly over the last couple of decades. Globalization of production has increased the use of imported inputs causing local costs to be more sensitive to exchange rates. Imported input use has tended to grow over time both across countries and industries. This tendency has increased the predicted sensitivity of retail prices of imported goods and other tradable goods to exchange rates. Stylized facts suggest that import prices are more responsive to exchange rate shocks than domestic producer and consumer prices. Meanwhile, a few recent studies show that when manufacturing sectors tend to use more imported intermediate inputs and also export their products, the degree of import price pass-through to producer prices increases significantly.2 Indeed, in many emerging market economies the share of foreign value added in value added of exports has substantially increased (Figure A1 in the Appendix). A significant number of studies in the vast literature of pass-through dynamics focus on estimating the pass-through coefficient, which provides valuable information for the policy makers and forecasters on to what extent shocks would be transmitted to the prices. Yet, a less studied but equally important aspect of pass-through is the factors determining its extent. These studies, which mostly concentrate on exchange rate passthrough, have pointed out that factors such as the composition of trade, the degree of openness of the economy, the substitutability of imported goods, the size of the economy, distribution margins and the flexibility of mark-ups, the firms’ pricing strategy, inflation outlook, exchange rate volatility, current account deficit and the direction and size of exchange rate shocks may affect the extent of pass-through.3 2 See Kiyotaka and Hoang (2016); Campa and Goldberg (2008). 3 See Burstein, Eichenbaum and Rebelo (2002); Goldberg and Knetter (1997); McCarthy (2007), among others. 3
6. The channels that the external factors affect prices are important determinants of the extent of pass-through . In this regard, one may consider direct and indirect channels. The direct channel is through imported final consumer goods included in the consumption basket. In this case, increases in import prices are expected to fully pass-through to product prices to the extent of pricing to market strategy. The latter is through production costs arising from the use of imported inputs in domestic production. In this case, the extent of pass-through is expected to be proportional to the share of imported inputs in production costs. Meanwhile, consumption goods generally constitute a smaller portion of total imports in most countries. Particularly, the dominance of imported input channel is more pronounced in countries for which the share of intermediate goods constitutes a significant portion of total imports and the intensity of imported input use is nonnegligible. Turkey can be categorized among the countries, where the indirect channel is very influential on the pass-through into final prices, where the intermediate inputs have a high share, around 73 percent, in total imports (Figure 1). Recently, the examples of a growing body of this literature such as Campa and Goldberg (2008), Auer and Mehrotra (2014), Ahn et al. (2016) and Casas (2019) emphasize the costs arising from imported input use as a more important channel for the sensitivity of consumer price changes to exchange rate shocks compared to imported final consumer goods. For example, the manufacturing supply chain in Asia is closely integrated. Auer and Mehrotra (2014) argue that in closely integrated supply chains any shock to domestic production costs or exchange rates can easily pass-through to other economies in the supply chain affecting intermediate prices. This structure has potential implications for both headline inflation and is also a concern for policy makers. Increased interconnectedness is likely to lead to greater sensitivity of aggregate inflation to cost of imported inputs especially when the cost changes are larger in magnitude. Similarly a recent study on Colombia shows that exchange rate pass-through to prices tends to be larger for industries in which firms use a larger share of imported inputs. While this link is stronger in the case of exports, it is positive for import prices.4 4 See Casas (2019). 4
7. On the other hand , it should be underlined that the import content of both consumption and production in many countries has steadily increased. According to OECD figures, as of 2015, the share of foreign value added in final consumption ranges between 10 and 75 percent in major advanced and emerging economies; it is considerably higher in EU countries and it is less than 30 percent in Turkey (Figure A2 in the Appendix). Stylized facts have shown that a generalized fall in exchange rate passthrough has taken place despite an increase in the import content over the past decades.5 Swallow et al. (2016) argue that not only import prices but also other prices such as wages, distribution costs, retail markups and that of non-tradables respond to the exchange rate shocks. Thus, despite the fact that the increased import content affect exchange rate pass-through adversely through imported input channel, the response of other prices are largely responsible for the decline in overall pass-through to consumer prices. Against this background, in this study, we estimate and elaborate on the sources of import price pass-through into producer prices in Turkey on a sectoral basis. Here, we focus on producer prices rather than consumer prices for three reasons. First, there has been a well-documented literature in Turkey for pass-through of external shocks and producer price shocks to consumer prices, yet the pass-through to producer price itself is rather neglected. Second, we aim to contribute to the literature by extending our analysis to individual sectors rather than working on aggregate price data to show the heterogeneity in the pass-through dynamics across sectors. Third, we aim to identify the source of the heterogeneity in the degree of pass-through across sectors by relating it to the sectoral differences in the use of imported input in domestic production and foreign exchange liability. Our study departs from the likes in the overall pass-through literature in some aspects. First, unlike most of the studies, which treat exchange rate shock as the main source of import price shock in domestic currency, we distinguish the exchange rate and import prices in foreign currency as two different sources of change in import prices and analyze their impact on producer prices separately. Stylized facts have revealed the importance of change in import prices in foreign currency. For example, in the case that 5 See Swallow et al.(2016); Mihaljek and Klau (2008). 5
8. exchange rate and commodity prices move in opposite directions to offset the impact of each other on import prices in domestic currency , if import prices in foreign currency are not included in the analysis as a separate variable, then pass-through effects are estimated as non-existent. However, Yüncüler (2011), Kara and Öğünç (2012), Özmen and Topaloğlu (2017) have shown that the degree and speed of the pass-through to prices might be different for each factor and, thus, decomposing their effect may enrich the information content. Second, estimating pass-through coefficients for each sector is not our ultimate aim. Rather, we consider them as a means to analyze the factors leading to different pass-through effect across sectors. In this regard, we compare the estimated pass-through coefficients for the two components of import prices with the share of imported input use in costs inclusive of labor, which is calculated from Input-Output Tables for Turkey, and examine if they are proportional to each other. Later, we elaborate on the sources why the pass-through coefficients for exchange rate and import prices might differ for a specific sector. Such distinction provides different policy implications to mitigate the impact of pass-through. Third, we do not restrict our analysis to industrial sectors and extend the analysis to services sector, where production is known to be less import dependent compared to industrial sectors. Our aim in such extension is to show that even within services sector some sub-sectors are more prone to external factors due to their exposure to imported input use. We use Vector Auto Regression (VAR) models to empirically estimate the passthrough effects using monthly data covering the period between 2010 and 2017. We find that both exchange rate and import price pass-through effects on producer prices vary significantly across industrial and services sectors. As expected, on average, the degree of pass-through for both shocks is higher in industrial sectors compared to services sectors. More importantly, we find that there is a significant heterogeneity in estimated pass-through coefficients across sub-sectors within industrial and services sectors as well. We mainly relate the variation in estimated pass-through coefficients across sectors to the intensity of imported input use in the production process. We find that there is a significant positive linear relationship between the intensity of imported input use and pass-through coefficient estimates for both exchange rate and import price shocks. Another striking finding is that the exchange rate pass-through estimates are higher than import price pass-through for the majority of industrial sectors. While the import price 6
9. pass-through is on average proportional to the intensity of imported input use , exchange rate pass-through is beyond that threshold. This may be attributed to the presence of factors other than production costs, such as foreign exchange liabilities and investment costs, which affect the pricing behavior and the firm’s balance sheet. In fact, we run crosssection regressions with estimated pass-through coefficients for two different samples (pre and post 2010) and find that higher foreign exchange liability is associated with excess exchange rate pass-through, which is defined as the difference between the exchange rate pass-through and the share of imported inputs. Thus, we conclude that considering the pros and cons of the use of high levels of imported inputs and management of foreign exchange liability are critical to enhance the effectiveness of monetary policy and to create room for maneuver to fight against inflation.6 The study proceeds as follow. In the next section, we briefly introduce the production structure of the Turkish economy on a sectoral basis and explain the motivation of this study. In the third and fourth sections, data and the methodology are presented. In the fifth section, we share the results of the empirical analysis and discuss them in detail. The last section concludes the study. 2. Background on Turkish Economy In this section, we provide data on the production and trade structure of Turkish economy to highlight the importance of indirect pass-through channel, i.e. effect through production costs arising from the use of imported inputs. Further, we elaborate on the trajectory of exchange rate and import prices in foreign currency to give more insight on the characteristics of the external shocks that Turkish economy encounters. From a historical perspective, intermediate goods, which also include energy items, constitute most of the total imported goods in Turkey (Figure 1). On average, while they account for 73 percent of all imported goods, the share of consumption goods is as low as 12 percent. The rest of the imported goods are capital goods. This hints that imported input use in production may be at significant levels, while the share of directly imported consumption goods in consumer basket is relatively low. See Erduman et al. (2019); Akgündüz and Fendoğlu (2019); Özcan and Sevinç (2019) for the use of imported intermediate goods in both production and export sectors. 6 7
10. Direct Import Content Source : TURKSTAT. 30 25 20 15 10 5 Total Consumption Public Consumption 0 Private Consumption 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Indirect Import Content Exports of Goods and Services Intermediate Goods Capital Goods Consumption Goods Figure 2. Import Content of Final Demand Components in Turkey (%) Total Investment Figure 1. The Composition of Imported Goods in Turkey (%) Source: Authors’ own calculation from 2012 I-O Tables. No official data is provided on the share of directly imported consumption goods in consumer basket in Turkey. Yet, we construct a proxy and provide a supportive argument for the higher influence of indirect pass-through channel on prices utilizing data from 2012 Input-Output (I-O) Tables. An I-O table describes the sale and purchase relationships between producers and consumers within an economy (see Table A1 in Appendix). In this regard, it is possible to measure the direct and indirect import content of each final demand component. We define direct import content of a demand component as the ratio of the value of the directly imported goods to the total value of the demand component. On the other hand, indirect import content of a demand component is the ratio of the value of imported input in domestic production of the demand component to its total value. As shown in Figure 2, in Turkey, the indirect import content constitutes a higher share for consumption and exports, whereas the opposite is true for total investments probably due to machinery and equipment investments. This hints that similar to many countries, consumer basket consists more of domestic goods. Another way to illustrate how imported goods affect the prices through cost channel is to study the relationship between producer and import price changes. We would expect a high correlation between the two, if imported input use is a significant portion of domestic production. As shown in Figures 3 and 4, the strong relationship between the changes in producer prices and import prices denominated in Turkish lira confirms the importance of cost shocks through imported inputs for the domestic prices. The correlation of annual changes is 0.81, while correlation of monthly changes is as high 8
11. as 0 .65. Similarly, Ahn et al. (2016) state that the correlation between monthly changes of total producer price index and import price index is 0.63 in Korea. Figure 3. Domestic Producer Price and Import Price Indices (Annual % Change) Figure 4. Domestic Producer Price and Import Price Indices (Monthly % Change) Domestic Producer Price Index (right axis) Domestic Producer Price Index (right axis) Import Prices (TL) 50 40 Import Prices (TL) 20 20 15 15 4 3 30 10 10 20 5 2 5 10 5 -1 -5 -5 -10 -2 -3 01.05 09.05 05.06 01.07 09.07 05.08 01.09 09.09 05.10 01.11 09.11 05.12 01.13 09.13 05.14 01.15 09.15 05.16 01.17 09.17 0 01.05 09.05 05.06 01.07 09.07 05.08 01.09 09.09 05.10 01.11 09.11 05.12 01.13 09.13 05.14 01.15 09.15 05.16 01.17 09.17 -20 0 0 0 -10 1 Source: TURKSTAT. Source: TURKSTAT. In fact, I-O tables allow for detecting the sectors in which the domestic production is dependent on imported inputs. Figures 5 and 6 present the intensity of imported input use in domestic production for each of the industrial and services sectors in 2002 and 2012, comparatively. Here, we define the intensity of imported input use as the ratio of the value of imported inputs to the sum of total intermediate input consumption and compensation of employees. Figure 5. The Intensity of Imported Input Use in Domestic Production of Industrial Sectors in Turkey (%) 2002 2012 Mining and quarrying Non-metallic minerals Food, beverages and tobacco Textile, apparel and leather Furniture Wood Fabricated metals Printing Other transport Machinery and equipment Pharmaceuticals Electricity Electrical equipment Motor vehicles Rubber and Plastic Paper Base metals Computer, electronic and optical Chemicals Refined petroleum 80 70 60 50 40 30 20 10 0 Notes: The intensity of imported input use in domestic production (import dependency ratio) is calculated by dividing the value of imported inputs in the domestic use table among the I-O tables for 2012 by the sum of total intermediate consumption and compensation of employees. 9
12. Figure 6 . The Intensity of Imported Input Use in Domestic Production of Services Sectors in Turkey (%) 2002 2012 Education Financial services Recreation and culture Telecommunication Postal and courier services Sporting, amusement and recreation services Legal and accounting services Human health Accommodation and food Travel services Land transport Insurance Other services Wholesale and retail trade Water transport Air transport Repair services 20 18 16 14 12 10 8 6 4 2 0 Notes: The intensity of imported input use is calculated by dividing the value of imported inputs in the domestic use table among the I-O tables for 2012 by the sum of total intermediate consumption and compensation of employees. Figures 5 and 6 show that as of 2012, the share of imported input in total production costs inclusive of labor costs is on average 25 and 7 percent for industrial and services sector, respectively. In addition, the intensity of imported input use shows significant variation across subsectors. In the industrial sector, the highest imported input intensity belongs to production of refined petroleum and coke with 72 percent, whereas the lowest reading is as low as 12 percent for mining and quarrying. In addition, sectors such as base metals, chemicals, motor vehicles and electrical equipment, which account for significant portion of Turkish exports, have higher than the average imported input intensity in industry. On the other hand, in the services sector, transport, wholesale and retail trade and repair services sectors use the most imported inputs in proportion to their production costs, while sectors such as education, financial services, recreation and culture have very low imported input use. The rises in abovementioned leading exporting sectors are also striking. Therefore, this heterogeneous outlook in imported input use feeds our motivation to pursue a pass-through analysis on sectoral basis and account for this heterogeneity in understanding the extent of both exchange rate and import price pass-through into producer prices. Last but not least, we share information about the course of external factors, of which we will be analyzing the impact on domestic prices. In this context, Figure 7 shows the path import prices in TL has followed. It is seen that Turkish economy has been exposed to rises in import prices in domestic currency steadily. However, the rise in 10
13. import prices seems to be more resilient after 2010 , as shown in grey shaded area. This means that pass-through dynamics of external factors have put more pressure on prices in Turkey after 2010. In order to understand the reasoning behind the rises, it would be better to disentangle the import prices to its two core components: exchange rate and import prices in foreign currency. The separate analysis of components, which is presented in Figure 8, provides important information about the dynamics and sources of import price rises in domestic currency. In the period after 2010 the changes in import prices in USD have been more moderate, while there has been gradual weakening of the Turkish lira, especially with the heightened uncertainty about global monetary policies after May 2013 when the US Federal Reserve signaled a tapering of its asset purchases. On the contrary, the source of import price rise in pre-financial crisis period was the robust rise in commodity prices, where Turkish lira followed a stable course. This distinction in the sources of change in import prices between the sub-periods is the major reason why we should take into consideration the sources of shocks in pass-through analyses. In fact, previous studies for Turkey by Yüncüler (2011), Kara and Öğünç (2012), Özmen and Topaloğlu (2017) stress the importance of this decomposition to enrich the information content of pass-through, as the persistency and volatility of different shocks affect the estimated coefficients (Taylor, 2000). Figure 7. Import Price Index in TL terms (2010=100) Figure 8. The Components of Import Price Index in TL Terms Import Price Index (USD) USD/TL Exchange Rate (left axis) 250 4 140 230 210 190 130 3 120 110 170 150 2 100 130 110 90 90 1 80 70 70 Source: TURKSTAT. 0 60 01.03 01.04 01.05 01.06 01.07 01.08 01.09 01.10 01.11 01.12 01.13 01.14 01.15 01.16 01.17 01.03 01.04 01.05 01.06 01.07 01.08 01.09 01.10 01.11 01.12 01.13 01.14 01.15 01.16 01.17 50 Source: TURKSTAT, CBRT. 11
14. 3 . Data In this section, we present the details of dataset on sectoral basis. In the construction of dataset, we utilized the Input-Output (I-O) tables of 2002 and 2012, as well as foreign trade index, domestic producer price index (D-PPI), consumer price index (CPI), industrial production index (IPI) and turnover index using Turkish Statistical Institute’s (TURKSTAT) database. Exchange rate is the monthly average value of nominal TL/USD. We construct data on a sectoral basis. The selection is based on the sectoral classification of I-O tables. Normally, I-O tables contain 64 different sub-sectors under agriculture, industry, services and construction sectors. However, due to data limitations we restrict our analysis to only industrial and services sectors. We also have to eliminate some industrial and services sectors due to lack of price, production or turnover data. Thus, we run the empirical analysis with 37 sectors in which 20 of them belongs to industrial sectors7, 8. We use monthly data for the period between January 2005 and December 2017, and run regressions for two different sub-samples as 2005-2009 and 2010-2017 for the industrial sector. The reason is that we see a notable change in the dynamics of the external factors during the period of 2010-2017 compared to 2005-2009 (Figure 8). In order to focus on the most recent pass-through dynamics, we consider regressions between 2010 and 2017 as baseline analysis. The application of the same empirical analysis for 2005-2009 time period allows us to make robustness checks and to further investigate whether the determinants of the degree of pass-through based on imported input use has changed over time. We should note that regression analysis for two different time periods is only done for industrial sector as lack of monthly turnover data before 2009 prevents us from repeating the same analysis for services sector. 7 Industrial sector includes mining and quarrying, refined petroleum, chemicals, base metals, computer, electronic and optical, paper, rubber and plastic, motor vehicles, electrical equipment, electricity, pharmaceuticals, machinery and equipment, other transport, printing, fabricated metals, wood, furniture, textile, apparel and leather, food, beverages and tobacco, other non-metallic minerals. 8 Services sector includes repair services, travel services, financial services, education, wholesale and retail trade, repair, land transport, legal and accounting services, accommodation and food, water transport, telecommunication, air transport, human health, recreation and culture, postal and courier services, sporting, amusement and recreation services, insurance, other services. 12
15. Construction of sectoral import price index Unlike many other studies in the literature, which use aggregate import price indices in pass-through coefficient estimates, we construct import price indices in US dollar terms at sectoral level to better reflect the sectoral cost pressures arising from imported input use. This is particularly important as the composition of the use of imported goods changes across sectors. To be specific, according to 2012 I-O tables, refined petroleum sector’s imports consist mainly of crude oil. Therefore, using aggregate import price index instead of crude oil price index would be inappropriate to measure the imported input cost pressures that refined petroleum sector is exposed to. This is particulary important when relative prices change significantly. Therefore we calculate an effective import price index for each sector, where the price of imported goods are weighted according to the shares given in Import Use Table, which shows the decomposition of the use of imported input in the production process of each sector. The mathematical representation of the effective import price index for each sector is as follows: