Wednesday, February 20, 2019

Unequal Exchange Rate Pass-Through across Income Groups


Unequal Exchange Rate Pass-Through across Income Groups


One sentence summary: There is evidence for redistributive effects of an exchange rate shock across income groups.



The corresponding paper by Hakan Yilmazkuday has been accepted for publication at Macroeconomic Dynamics


Abstract
Exchange rate pass-through (ERPT) into prices and into income loss are shown to be enough to calculate ERPT into welfare loss by using implications of a simple model. These ERPT measures are estimated at the good level by using a unique micro-price data set from Turkey, and they are combined with income-group specific expenditure shares at the good level to obtain aggregate-level ERPT measures for alternative income groups. An exchange rate shock resulting in a real depreciation of 1% is shown to decrease welfare by about 0.80% for the average-income consumer, while this estimate ranges between 0.73% and 0.83% for consumers in the lowest and highest income quintiles, respectively, suggesting evidence for redistributive effects of an exchange rate shock. Using micro prices has further resulted in showing that traded, nondurable, flexible-price, or income-elastic goods contribute more to ERPT into welfare loss for the average-income consumer, suggesting important policy implications for filtering out the noise in the measurement of aggregate-level prices.


Non-technical Summary
Open economies are subject to international shocks that are mostly reflected as movements in their exchange rates. The effects of such movements, the so-called exchange rate pass-through (ERPT) measures, highly depend on the extent of expenditure switching between domestic and foreign goods, which may take time to observe due to price stickiness. Especially when exchange rate movements are not fully reflected in prices (i.e., incomplete ERPT into prices), there are significant welfare effects due to price setting in the currency of consumers versus producers that should be taken into account by policy makers. Therefore, it is not only important to understand ERPT into prices but also into quantities and thus welfare, both in the short and long-run. This paper follows such an approach by estimating ERPT into prices, income and welfare for alternative horizons.

Although there is a large body of literature investigating/estimating ERPT measures into prices, the existing evidence on ERPT is mostly at the aggregate level, suppressing disaggregated-level details such as income redistributive effects of exchange rate movements among consumers. Despite several studies in the literature that have proposed such income redistributive effects in their theoretical frameworks, to our knowledge, there is no corresponding empirical evidence. The lack of empirical evidence is mostly because aggregate-level prices (e.g., consumer price index) are published only for the average-income consumer, while the investigation of redistributive effects requires the knowledge of prices faced by alternative income groups. This paper proposes a solution to this problem by using a good-level investigation. In particular, with the knowledge of the set of goods consumed by each income group, we use the corresponding micro prices to estimate ERPT measures at the good level, which are further combined with the expenditure shares of goods for each income group to estimate the aggregate welfare effects.

Estimating ERPT measures at the good level (as in this paper) is also useful for avoiding any aggregation bias, since estimations at the aggregate level suppress several micro-level details. These include micro-level distortions such as price stickiness, tradability of goods, degree of competition reflected in markups, transportation costs in different sectors, or the quality of goods. These micro-level details not only are important to understand the economic intuition behind ERPT into good-level prices but also can be used to identify the goods/sectors responsible for the effects of exchange rate movements at the aggregate level. By using a good-level approach, this paper not only considers these micro-level details by construction but also achieves further decomposition analyses showing the contribution of each good category to ERPT measures for each income group.

Finding the goods/sectors that are responsible for ERPT measures has important monetary policy implications as well, because, understanding changes in micro prices can offer more relevant information about the nature of inflation in countries such as Turkey, where good-level prices change more frequently compared to other countries. In particular, to have a healthy measure of inflation that can be used for optimal policy making, the noise in aggregate-level prices should be filtered out by using measures such as the trend or core inflation, and using disaggregate-level price data to determine the responsible goods/sectors (as in this paper) is one way to do it as suggested by several studies in the literature. These measures are also useful to increase the effectiveness of communicating monetary policy actions in an environment of frequently changing prices.

Regarding the estimation methodology, several empirical studies in the literature have considered single-equation frameworks that result in endogeneity bias. Also considering our discussion on micro-level details above, it is implied that an empirical investigation based on a system of equations at the good level is necessary to avoid both aggregation and endogeneity biases in the estimation of ERPT measures. This paper achieves such an unbiased estimation of ERPT by using a structural VAR model at the good level, where ERPT ratios are considered for the measurement of ERPT. Specifically, ERPT into prices (income) is measured as the cumulative response of prices (income) divided by the cumulative response of exchange rates, both following an exchange rate shock. Such an approach followed at the good level effectively addresses concerns related to both aggregation and endogeneity biases. Micro-price data, good-level expenditure shares for alternative income groups, together with data on income and exchange rates, are used from Turkey over the monthly period between 2004m1-2018m12. Once ERPT into prices and income are estimated, by using the implications of a simple model introduced in this paper, ERPT into welfare is calculated as ERPT into income minus ERPT into prices.

The results for the average-income consumer suggest that an exchange rate shock resulting in a 1% depreciation of the currency increases the aggregate price index by about 0.45%, reduces income by about 0.34%, and reduces welfare by about 0.80%. When the same investigation is achieved across alternative income groups, the welfare loss ranges between 0.73% and 0.83% for consumers in the first and last income quintiles, respectively, suggesting redistributive effects of an exchange rate change among consumers.

The good-level investigation in this paper also allows for the decomposition of this aggregate-level result into the contribution of each good category to the welfare of alternative income groups. In particular, among good categories, those that are traded, nondurable, flexible-price, or income-elastic contribute more to ERPT into welfare for the average-income consumer, and the contribution of durable and income-elastic goods gets higher with consumer income.



Among sectors, "Food and Non-Alcoholic Beverages" followed by "Communications" and "Transport" contribute the most to ERPT into welfare for the average-income consumer, although this decomposition differs significantly across income groups. Specifically, ERPT into welfare is mostly through "Food and Non-Alcoholic Beverages" and "Housing, Water, Electricity, Gas and Other Fuels" for the lowest-income consumers, while it is mostly through "Transport" and "Communications" for the highest-income consumers. Due to their higher contribution to ERPT measures, it is implied that these sectors should be paid more attention while measuring the trend/core inflation and thus conducting policy.