The Persistency of Correlation between Currency Futures : A Macro Perspective

This paper examines the dynamic correlation between currency futures prices. Using the Dynamic Conditional Correlation model (Engle, 2002) this study utilizes time-varying correlations, focusing on the persistency of correlation of currency prices. The sample includes eight currency futures traded on the Chicago Mercantile Exchange from 1999 to 2008 and the U.S. dollar index future. The study finds that the Canadian dollar and the Australian dollar have the highest persistency while the Swiss franc and the Russian ruble have the lowest persistency. In addition, the study finds that the time-varying conditional correlation between currency futures and the U.S. dollar futures is influenced by a country’s macroeconomic conditions.


Introduction
According to the Triennial Central Bank Survey, conducted by the Bank of International Settlements, the forex market averages about $5.3 trillion per day (as of April 2013) (Note 1).Due to the important role this market plays in the world economy, the literature for the forex market has been growing rapidly.In this particular study we examine the dynamic correlation across currency futures prices to U.S. dollar index futures (Note 2), with a focus on the persistency of correlation between eight currency futures prices traded on the Chicago Mercantile Exchange: British pound, Brazilian real, Australian dollar, Canadian dollar, Japanese yen, Euro currency, Swiss franc, and Russian ruble.Using the Dynamic Conditional Correlation (DCC) model developed by Engle (2002), we incorporate time-varying correlations into our analysis.This study differentiates from previous studies in that it is the first to analyze the persistency of relation between currencies future prices.This paper is most related to Lien and Yang (2006), which investigates the effects of spot-futures spread on the risk and return structure in currency markets.Using a bivariate GARCH framework, the authors find evidence that spreads on the risk and return structure of spot and futures markets produce asymmetric effects.The implications of these asymmetric effects are examined, with special consideration given to the performance of futures hedging strategies.This study differentiates from Lien and Yang (2006), however, in that our focus is on the persistency of correlation between currency futures prices and that we instead use a DCC framework.The DCC model is similar to a bivariate GARCH in spirit, but the DCC places several restrictions on how the correlation can change (in essence it is a special case of a bivariate GARCH).
In addition, this paper is also motivated by Harvey and Huang (1991) and Han, Kling and Sell (1999).Both of these papers explore how macroeconomic variables impact the currency futures market.In Harvey and Huang (1991), the authors examine volatility patterns in the forex market.They surmise that increases in volatility are more often attributed to macroeconomic news than private information through trading.In contrast, Han, Kling and Sell (1999) look at day-of-the-week effects in the currency futures market.Evidence in this paper suggests that the day-of-the-week effect is impacted by private information from trading or market microstructures, not macroeconomics news.Our paper tries to build upon these two studies by examining how different macroeconomic conditions affect the currency futures market.More specifically, we examine how four specific macroeconomic variables impact the correlation between US dollar futures and currency futures.
The sample spans from 1999 to 2008.The study finds that the persistency of currency futures interactions varies substantially across different currencies with the Canadian dollar and the Australian dollar having the greatest persistency while the Russian ruble and Swiss franc have the weakest.Further, the study finds that the time-varying conditional correlation between currency futures and the U.S. dollar futures is influenced by a country's macroeconomic conditions.
The rest of the paper is organized as follows: Section 2 describes the data, Section 3 presents the methodology, Section 4 examines the empirical results, and Section 5 gives the conclusion.

Data
The initial futures data consists of daily future prices for currency futures over the period January 1999 to December 2008.This data is collected from RC Research (www.Price-Data.com)The weighted U.S. dollar futures are used as a basis for comparison.The U.S. dollar index (USDX) (Note 3) is an index (or measure) of the value of the United States dollar relative to a basket of foreign currencies.The USDX futures contract has two features that influence its pricing and its use.First, the USDX index is a geometric average, rather than an arithmetic average, of the constituent currencies.Second, the foreign exchange (FX) rates in the USDX index (in U.S. dollars per foreign exchange rate) are in the denominator of the index, implying that a dollar appreciation leads to a higher index level.Both the geometric averaging and the use of quoting convention have implication for the use of the USDX futures contract in hedging a foreign exchange exposure.Eytan, Harpaz, and Krull (1988) point out, the divergence between the geometric and arithmetic averages depend on both the volatilities of the individual currencies and their co-movements (sometimes referred to as their "correlations").
The USDX futures contract began trading on November 20, 1985 on the Financial Instruments Exchange, a division of the New York Cotton Exchange, which is now part of the New York Board of Trade (NYBOT).The USDX index was originally a geometrically weighted average of ten different currencies, with each currency representing a country that was a major trading partner with the United States.With the introduction of the Euro, the USDX index became a geometrically weighted average of six currencies, which represent five major U.S. trading partners and the Euro.

Index Formula
The formula for the index level on date t is the product of the six currencies spot rates, each raised a power related to a currency-specific weight.The general formula for the index can be written as: where USD per foreign determined for the US futures con We first be while thei return bein statistics o ce matrix for a of time-varyin tes the conditio denotes the un ated, residual v mit a two-stage nd the estimate dard deviation on for the DCC r estimates a q expressed as:   call that the Eu more closely re ence that its st ollar are 13.6% beta parameter) and Canadian d r have the hig ch of the eight oted as rho) be dency to be n h pound, Cana r, one does obs    For both the Australian dollar and the Swiss franc industry production is statistically significant and positively related with the dependent variable.The Canadian dollar, on the other hand, is statistically significant and negatively related with rho in regards to industry production.As for inflation, with the exception of the Brazilian real and the Japanese yen, all currency futures are negatively related and statistically significant to rho.Only the Japanese yen is positively statistically significant for inflation.Lastly, in regards to the risk free rate and money growth, only the Canadian dollar and the Japanese yen are positively statistically significant for the risk free rate and only the Swiss franc is negatively statistically significant for money growth.

Conclusion
This study investigates the time-varying correlation between currency futures prices utilizing the DCC model, focusing on the persistency of correlation.The study finds that the Russian ruble and the Swiss franc have the weakest persistency while the Australian dollar and the Canadian dollar have the greater persistency.However, the relationships do vary somewhat over the sample period.In addition, the study finds that the time-varying conditional correlation between currency futures and the U.S. dollar futures is influenced by a country's macroeconomic conditions; specifically, industry production, inflation, the risk free rate and money growth.
In summary, this paper provides evidence on the persistency between different currency futures (British pound, Brazilian real, Australian dollar, Canadian dollar, Japanese yen, Euro currency, Swiss franc, and Russian ruble) and USDX futures and how macroeconomic growth variables impact that persistency.
Note 3. The short-coming of using the U.S. Currency Futures Index is that it is an unequally weighted index, so the currency that is weighted more heavily, such as Euro, will inherently move more closely with the index.
Fig relation estimates Swiss franc, and R

Table 1 .
and includes open, high, low, and close prices; as well as, volume and open interest.All daily future prices are in U.S. dollars.The currency futures included in this study are listed as follows: British pound, Brazilian real, Australian dollar, Canadian dollar, Japanese yen, Euro currency, Swiss franc, and Russian ruble.All eight currency futures are traded on the Chicago Mercantile Exchange (CME) and all currencies prices are coded the same way-the US$ price of per unit of currency.Table1provides a summary of the contract size, approximate margin, and minimal fluctuation of the 8 currency futures.Sample periods for currency futures traded in U.S.

Table 5 .
Time-varying correlations and country characteristics Note.This table reports parameter estimates of rho = a + b 1 × Industry Poduction + b 2 × Inflation + b 3 × Risk Free Rate + b 4 × Money Grown for British pound, Brazilian real, Australian dollar, Canadian dollar, Japanese yen, Euro currency, Swiss franc, and Russian ruble futures for the period from January 1999 to December 2008.Standard errors are reported in the parenthesis.