The effect of VAT rate on inflation in Latvia: evidence from CPI microdata

Abstract

This article evaluates the inflation effect of recent value added tax (VAT) rate changes in Latvia by using consumer price index (CPI) microdata. Our findings suggest that the pass-through of the tax rate to consumer prices is strong in case of upward tax adjustments, especially when there are no demand restrictions, while the pass-through is weaker for tax reductions. The frequency of price changes peaks at the moment of VAT adjustment, which, however, is partially compensated by lower average size of price revisions. The level of pass-through exhibits a high degree of heterogeneity with higher pass-through for goods, especially food, and lower for services.

Keywords:

• value added tax
• inflation
• sample selection model
• CPI microdata
• Latvia

JEL Classification:

• C24
• D40
• E31
• H20

Acknowledgements

The authors want to thank Oskars Alksnis (Central Statistical Bureau of Latvia) for the provision of CPI microdata, and Martins Bitans, Daina Paula (Bank of Latvia) and anonymous referees for their valuable comments and recommendations. The views expressed in this publication are those of the authors, employees of the Monetary Policy Department of the Bank of Latvia. The authors assume responsibility for any errors and omissions.

Notes

¹ In May 2004, as Latvia became a member of the EU, the reduced value added tax (VAT) rate for several product groups, such as newspapers and periodicals, accommodation services, water, refuse and sewage collection and some medical products, decreased from 9% to 5% (there were two different reduced rates prior to accession). On the other hand, VAT rates increased from 0% to 5% for some products, such as books and cinema. From the beginning of 2005, the reduced VAT rate was applied to passenger transport by road and railway. Since 2007, the reduced VAT rate application has widened to include electricity and natural gas consumption by households previously exempted from VAT. For a short period (January 2007 to December 2008), the reduced VAT rate was also applied to solid fuels and hairdressing services. Heat energy is the only product group for which VAT rate increased during the period 2006 to 2007 (July 2006) from 0% to 5%. Table A1 in Appendix provides a more detailed description of nonstandard VAT rate changes for several products since 2003.

² More details on how frequency of price changes are calculated and how frequency approach differs from duration approach can be found in Aucremanne and Dhyne (2004) and Bils and Klenow (2004). The results for individual products were aggregated using product weights in the consumer price index (CPI) basket. As coverage of the CPI microdata is below 100%, the weights of existing products were adjusted to get the total sum of weights equal to 100%. The adjustment was done on the most detailed disaggregation level (usually three- or four-digit COICOP (Classification of Individual Consumption by Purpose) level) taking into account the consumption basket weights provided by the Central Statistical Bureau (CSB) of Latvia.

³ Benkovskis et al. (2010) report that the frequency of price changes was 28.7% and the duration was approximately 3.5 months in 2003 to 2009. Although Benkovskis et al. (2010) use the same CPI database, these results are not directly comparable with those reported in Table 2, as the current article ignores price imputations. The difference mainly comes from decreased frequency of price changes in clothing and footwear.

⁴ In the case of fuel, a more frequent price reduction was obviously driven by the decreasing world price of oil. Lowering of beer prices was most probably due to weather conditions, for the relatively cold summer diminished the demand for beer.

⁵ Gabriel and Reiff (2010) report a similar effect for VAT rate changes in January 2004, January 2007 and September 2006 in Hungary.

⁶ These figures are available upon request.

⁷ Hoffmann and Kurz-Kim (2006) use the average size of log changes while defining μ⁺ and μ^–, as they proxy German CPI by a geometric Laspeyres formula. Taking into account that Latvia’s CPI is calculated using an (arithmetic) Laspeyres formula, we define μ⁺ and μ^– as a simple growth rate. Note that μ^– is always negative.

⁸ An important difference from Benkovskis et al. (2012) is that the current article uses accumulated inflation calculated from Harmonized Index of Consumer Prices (HICP) at constant tax rate. The effect of changes in indirect tax rates is already accounted for by VAT change variables; therefore, the use of standard inflation will underestimate the tax rate pass-through.

⁹ To account for the possible asymmetries, we distinguish between the cases where the previous price adjustment is positive and the cases where it is negative (by using dummy variable, DW, for cases when previous price change is negative).

¹⁰ Year dummies can be interpreted as the effect of omitted macroeconomic conditions, e.g., the demand and supply factors.

¹¹ Estimates were not made for fuel products due to a very specific price setting nature. Full pass-through of tax shocks was assumed instead.

¹² For main characteristics of all individual products, see Benkovskis et al. (2010).

¹³ This is in line with the results obtained by Gabriel and Reiff (2010) for Hungary. They report that the inflation effect of 5 percentage point increase in VAT rate in September 2006 is estimated to be 2.13%, while the inflation effect of 5 percentage point decrease in VAT rate in January 2006 is only –1.08% (for affected products).