Directional analysis of consumers’ forecasts of inflation in a small open economy: evidence from South Korea

ABSTRACT

We evaluate the directional accuracy of consumers’ forecasts of inflation in predicting the movement of the actual CPI in a small open economy. In order to do so, we use a method developed by Pesaran and Timmermann (2009), based on South Korean data. By illustrating an application of the new market-timing test, we show that consumers’ expectations of inflation are not a useful predictor of the CPI in South Korea. Our findings suggest that the directional accuracy of consumers’ 1-year-ahead forecasts of inflation is not affected by the inflation targeting of the Bank of Korea. Our findings also suggest that consumers’ 1-year-ahead forecasts of inflation are scattered away from the Bank of Korea’s inflation target.

KEYWORDS:

• Directional analysis
• consumers’ inflation forecasts
• CPI
• small open economy
• survey data

JEL CLASSIFICATION:

• C53
• E17
• E31
• E37
• E52

ORCID

Yoichi Tsuchiya http://orcid.org/0000-0003-2485-3959

Notes

¹ The respondents to the survey are professional money managers in their areas.

² Forecasters working for investment banks, commercial banks, consultancies and in universities.

³ The data can be found on the BOK homepage, http://ecos.bok.or.kr/flex/EasySearch_e.jsp.

⁴ The surveys are usually conducted on the tenth of each month, which precedes the consumer survey and publication of BOK’s forecast.

⁵ See Ash, Smyth, and Heravi (1998) and Henriksson and Merton (1981) for details.

⁶ The PT test focuses on predictive failure. The null hypothesis of predictive failure is not, in general, equivalent to the hypothesis that the predicted and actual values of the variable are independent. However, in 2 × 2 cases as in this study, they are equivalent.

⁷ Pesaran and Timmermann’s (2009) concern is the dependence among serially correlated multicategory variables, which includes the case of two serially correlated binary sequences as a special case. They developed two versions of the test: one based on the trace canonical correlation, and the other on the maximum canonical correlation. We focus on the former because the two versions perform similarly in the two-dimensional case. See Pesaran and Timmermann (2009) and Chou and Chu (2011) for details.

⁸ We examined the directional accuracy of consumers’ forecasts whose publication months are equivalent to the consensus forecasts because the overlapping forecast horizons of the consumers’ forecasts might bias the main results, and to compare directional accuracy directly between consumers and private forecasters. Although not reported in this article, the consumers’ forecasts are not useful and our findings are robust.

⁹ Note that those forecasts made before are included in the subperiod before 2007 and those made after are included in the subperiod after 2007. The same rule also applies to the division of forecasts into groups hereafter.

¹⁰ We do not report these results in this article because they do not change the conclusion. However, the results are available upon request from the authors.

¹¹ The Bank of Korea announced changes in the target range on 9 January 2003, 8 January 2004, 26 November 2009 and 11 October 2012, while the inflation targeting at 3% has not been changed regardless of the target variable and the sample periods.