Gold and silver have been used interchangeably for thousands of years, indicating a long-term connection between the two precious metals. However, there are a number of factors that could cause their prices to diverge.The popularity of commodities as an investment and a hedge against adverse financial or economic events may constitute a force that either creates a non-existent long-run relationship or strengthens a pre-existing long-run relationship.2 If this is the case, periods of extreme price movements, such as those observed during bubbles or financial crises, may provide additional insights as to what factors drive long-term relationships. Examples of such factors include industrial demand for silver and jewelry, dental demand, and central bank demand for gold.

The research questions posed by Escribano and Granger (1998) and the significant price shifts in commodities, particularly in the precious metals gold and silver, in recent years are the impetus for this study. Escribano and Granger (1998) looked at how prices of gold and silver changed between 1971 and 1994. They found that prices of gold and silver are co-integrated, especially after a bubble burst. Additionally, they suggest that the two markets might be splitting up. This study aims to fill a void in the existing body of research by investigating whether the occurrences following the publication of the Escribano and Granger study supported their hypothesis regarding the separation of the two markets. Additionally, we are aware of no conclusive responses to the second question regarding the function of bubbles.

There are a number of other studies that look into the relationship between gold and silver prices. However, the majority of these studies use data that makes it impossible to compare them to the Escribano and Granger study or the samples analyzed are too small to answer the authors' questions. Ciner (2004), for instance, looks at the years 1994 to 1998 and argues that there is evidence for a co-integrating relationship over time.
Using data from 1990 to 2009, Figuerola-Ferretti and Gonzalo (2010) discover that gold and silver are only co-integrated under conditions of high volatility and a weak US dollar.
Using the time period from 1978 to 2002, Lucey and Tully (2003) discover a long-term relationship that is stable. Adrangi, Chatrath, and David (2000) use intraday data between 1993 and 1995 with a focus on price discovery, while Liu and Chou (2003) use daily data from 1983 to 1993.

The above-mentioned list demonstrates that no paper revisits and expands Escribano and Granger's (1998) analysis to address the following concerns:

Are the markets for gold and silver distinct or co-integrated? Is there a pattern to the relationship?
And do financial bubbles and crises link to one another, thereby cointegrating silver and gold price?
The purpose of this study is not only to determine whether gold and silver have a long-term relationship but also to determine whether investors are to blame or whether fundamentals are to blame for the connection between the two prices. We use a relatively long sample period of gold and silver spot prices to contribute to the literature on financialization, which examines whether the use of commodities as investment increases their price co-movement with other assets, particularly stock indices (e.g., see Tang and Xiong, 2011). We are able to replicate Escribano and Granger's (1998) estimates and investigate the effects of a longer sample period on the findings and conclusions thanks to the sample period that covers more than 40 years, from 1970 to 2011. Because gold and silver prices were regulated prior to 1970, the sample period is the longest economically feasible time frame. We show that bubbles and financial crises do indeed create long-term relationships, which is an interesting feature of the longer sample period because the data contain more periods of extreme price behavior similar to the bubble period in the late 1970s. The price of silver and gold do not co-integrate during "normal" times. Gold drives the long-term relationship despite its lower volatility in comparison to silver, as demonstrated by the derived Granger causality tests and the Vector Error Correction Model used for the entire sample.