Iron ore price and the AUD exchange rate: A Markov approach

ABSTRACT

The present paper fitted the monthly data set into the Markov regime switching model to examine the relationship between the iron ore price and the Australian dollar (AUD) exchange rate. The study dichotomised the AUD into state 1 (depreciation) and state 2 (appreciation). The empirical results indicate evidence of an asymmetric relationship between an iron ore price change and the AUD exchange rate fluctuation based on states. The AUD appreciates with a fall in iron price and depreciates with a rise in iron ore price. The results contradict with the understanding of the commodity-currency theory. Additionally, iron ore price reduces the AUD state expected duration and the switching probability, but increases the AUD volatility. Based on the transition probability, the AUD has a higher chance of depreciating than appreciating. The statistical economic impact of the AUD currency is higher when appreciating than depreciating.

KEYWORDS:

• The AUD exchange rate
• iron ore price
• Markov-switching model

JEL CLASSIFICATIONS:

• E19
• E39
• F39

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that support the findings of this study are openly available in: Reserve Bank of Australia [https://www.rba.gov.au/], FRED Economic Data [https://fred.stlouisfed.org/categories], and OECD Data [https://data.oecd.org/].

ORCID

Clemence Gomwe http://orcid.org/0000-0002-3763-4758

Notes

1. Changes in economic policies, technological advancements, financial innovations, shift in global economic power, resource endowments, social and political unrest all over the world, shift in global commodity demand and supply, leaving currencies and commodities highly volatile, unpredictable and challenging to come up with a single solution to deal with their many associated economic issues.

2. The new phenomenon is defined by the regime-switching volatility, mean, cross-covariance and autocorrelation of the mentioned macroeconomic variables, and also the phenomenon statistically varies across regimes.

3. For instance, cheap iron ore exports from Brazil negatively affect iron ore exports from Australia and a fall in production from VALE positively affects Australia exports (https://thewest.com.au/business/mining/wa-iron-ore-miners-riding-high-on-brazil-supply-chaos).

4. Diebold, Lee, and Weinbach (1994) introduced the Markov time-varying transition probability, which considers the Markov switching to be governed by a set of observed variables, and Chang, Choi, and Park (2016), Kalliovirta, Meitz, and Saikkonen (2015) and Kim, Piger, and Startz (2008), coming up with an almost closely related approach based on the Autoregressive latent factor to govern the Markov switching.

5. ${\propto }_{\mathit{t}}$ - as the AUD Real Trade Weighted Index at time T, and was introduced in 1970 to measure the AUD exchange rate perfomance since it is a more informative measure of currency performance as compared to other measures (ABS 1998). ${\mathit{\lambda }}_{\mathit{t}}$ - as iron ore spot price in AUD/dry ton given time T, ${\mathit{\eta }}_{\mathit{t}}$ as the inflation rate [measured by the consumer price index (CPI)], given time T, and ${\mathit{\kappa }}_{\mathit{t}}$ representing real interest rate given time T. Note: All monthly data set is in log form.

6. The higher the variance value, the lesser the currency volatility, the reason being, as measured by the time in-between switching, if the measured period is small, it mirrors a higher currency switching frequency (volatility). As the estimated variance gets small and small, volatility will be rising as well, until a point when the swings flatten out (no switching period, which however is rare).

7. Matrix A represents the Original Markov Model; Matrix B as the Extended Markov Model excluding iron ore price change; and Matrix C as the Extended Markov model including iron ore price change.