Gold And Silver: Underpriced Based On Historical Data?

| About: SPDR Gold (GLD)

Summary

Historical data can be used to form multi-linear regression models that take into account fundamental variables to predict a price output.

These models determine that gold and silver may be undervalued relative to what historical fundamental data might suggest.

Is that any indication that gold and silver are a buy in the current market?

Thesis

I use structural models (i.e., four different multi-linear regression models) to obtain pricing estimates for both gold (NYSEARCA:GLD) and silver (NYSEARCA:SLV). I obtain median estimates of gold prices going from $1,440-$1,800 and silver prices running from $20-$24 on a per ounce basis. Even with gold and silver prices hovering around $1,200 and $17.20 currently, this should not be taken as an implicit recommendation to buy one or both metals. In my view, both markets are currently light in the way of fundamental, macro catalysts that could lead prices higher.

Overview

In previous posts on gold and silver, I illustrated the process of using structural linear regression models to potentially price these two metals based on current economic data. The basic procedure entails adding in variables that statistically influence the price, and using the resultant linear equation to input up-to-date macro data to find a price output.

For both gold and silver, the most influential variables consist of US inflation, interest rates, budget deficit/surplus, money supply, and current account. US real GDP and total world reserves are of lesser significance. While gold and silver are largely lumped in with commodities, they don't typically trade on the traditional elements of supply and demand. They are closer to a form of alternative currencies and move up and down based on macroeconomic sentiment, particularly as it pertains to the US. The US economy is the largest in the world and consumes 24%-25% of all global activity, accounting for $18.9 trillion of around $77.8 trillion in total world GDP. And historically, precious metals became more sensitive to macroeconomic influences after the Nixon administration abolished the convertibility of US dollars into gold in 1971.

Going long gold is one of the best methods of expressing the thesis of an upcoming deterioration in economic conditions or uncertainty/volatility entering the markets. Hence, gold tends to increase in price as a wealth preservation mechanism during times of stagnation or turmoil, and vice versa when the market is viewing the US (or world) market more positively. Supply and demand related factors have some degree of influence, but are secondary and generally show no level of statistical significance over longer timeframes.

Since the models are based off quarterly data (going back to Q1 1968), I updated these from the end of the last quarter and re-ran them after adding the latest data. Since the end of Q3 2016, US inflation has increased (to 1.6% as measured by the price consumption expenditures price index (PCEPI)), and the M2 money supply expanded by $80 billion (+0.6%).

These events are traditionally bullish for gold and silver. However, since the November US election the general market sentiment has been buoyant on the US economy. Investors have positively interpreted the Trump administration's fiscal objectives to lower corporate tax rates (to the 15%-20% range), invest in infrastructure projects to increase productivity, and repatriate overseas cash at a discounted tax rate. These are all expansionary policies, which makes riskier assets such as stocks more appealing relative to safe-haven assets such as gold and US Treasuries (NYSEARCA:TLT)(NYSEARCA:IEF). As a result, gold and silver are down 5% and 7% since election day, respectively. The 10-year Treasury yield has increased 70 bps as a consequence of enhanced risk-taking appetite, and the S&P 500 (NYSEARCA:SPY) has advanced 6.3%.

Results

I use four different regression models - ordinary least squares ("OLS"), which is the most common approach to linear regression modeling, t-distribution, quantile regression, and log-normal. I tend to place the most emphasis on the t-distribution model, as it often does the best at representing the distribution of the data. Financial variables (such as price movements in the financial markets) often are better modeled by "fatter-tailed" distributions, meaning that the t-distribution (which has this characteristic) better accounts for the presence of outliers. OLS regression, which is based on the normal distribution (which is inherently thin-tailed), tends to understate the influence of outliers, making it less appealing, particularly in risk assessment models.

Based on the latest macroeconomic inputs statistically relevant to the price of gold (PCEPI inflation, federal funds rate, current account, federal deficit, M2 money supply), the four models produce median outputs running from about $1,440-$1,800.

(Source: author)

The log-normal model has a very large confidence interval due to the nature of the distribution and is inherently less reliable. But the remaining three suggest that based on the accumulation of historical data, today's macroeconomic situation might suggest a gold price about 25% higher.

The same level of undervaluation could also be shown with silver, if one again excludes the log-normal model:

(Source: author)

Median prices run from about $20-$24 per ounce, with a 95% confidence interval running from approximately $16.80-$29.40 among those three models.

And although both gold and silver are trading cheaply according to these models, this is not a necessarily a recommendation to buy either of these metals. Undervaluation or overvaluation itself is a poor reason on its own. Additionally, making a call based on one method of analysis and without a catalyst on the horizon would be injudicious. There are numerous other methods by which the value of gold can be estimated, and no decision should be made naively on the basis of one model. Moreover, even if regression models can point to highly statistically significant elements than can influence the output variable, they are not always appropriate for predictive purposes (although I do believe the gold and silver models used here can do a reasonable job of such based on internal diagnostics, such as the R-squared reading among others).

Before entering back into these markets, I would require some sort of evidence that inflation is picking up without accompanying growth or conditions are weakening in some part of the global economy that could influence the flow of funds toward safe assets such as precious metals. Until then, I'm remaining on the sideline of both markets.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

About this article:

Expand
Author payment: $35 + $0.01/page view. Authors of PRO articles receive a minimum guaranteed payment of $150-500.
Want to share your opinion on this article? Add a comment.
Disagree with this article? .
To report a factual error in this article, click here