December and the Dollar

Cognitive dissonance is when one holds contradictory beliefs simultaneously. This is the condition investors are in when they believe in regular patterns in the markets and the efficient market theory.

‘Tis the Season

The largely unregulated foreign exchange market has an average daily turnover last officially estimated (BIS 2007) at $3.2 trillion. As the year winds down, many investors and analysts, who otherwise believe foreign exchange is the most efficient market in the world, want to believe there is a seasonal pattern that allows one to profitably sell the dollar in the month of December.

It is understandable why there may be seasonality for some agricultural products. Commercial agrarian societies could be very cyclical as the agriculture cycle drives the credit cycle. Even for modern service oriented economies, seasonality can play a big role in economic activity and behaviors. As we approach the holiday season this is obvious. Economists recognize this and often try to make seasonal adjustments to the macroeconomic data, including for example the U.S. non-farm payrolls and trade figures.

In the capital markets there seems to be some seasonality as well. For example, investors are familiar with some version of the January effect in the stock market, which holds that the stock market tends to do particularly well in either the first week or the entire month, or be a sign of the general direction for the next 11 months.

Before the Japanese fiscal year-end on March 31st and the half year-end on September 30th, Japanese institutional investors often sell foreign assets and repatriate the funds. The evidence for this appears stronger in the weekly portfolio flow data the Ministry of Finance publishes than in the foreign exchange market itself.

Perhaps owing to similar considerations, investors may do the same thing on a broader scale as the end of the year draws near. There may indeed by a winding down of activity, but it is not clear why that should have an asymmetrical negative impact on the dollar. Moreover, until recently a couple large financial institutions had fiscal years that ended November 30th. The governments and many other businesses have a fiscal year that differs from the calendar year.

Evidence

There is little substitute for actually looking at the data. We use Bloomberg data for the 1990-2008 data. We begin with the euro-dollar exchange rate. Prior to the advent of the euro in January 1999, we used a synthetic calculation for the euro.

Over the entire period, the dollar has fallen in eight of the nineteen Decembers against the euro. That does not strike one as very compelling evidence. If it were simply random, one would expect the dollar to fall in nine or ten Decembers. Surely if you flipped a coin nineteen times and eleven times were heads, you would not conclude that it was something wrong with the coin. There is no reason to reject what is called the null hypothesis, that there is no relationship between events.

The evidence for negative seasonality for the dollar against the yen is even more elusive. On the contrary the dollar has risen against the yen in December more than it has fallen. The data set shows the dollar fell in only eleven of the nineteen Decembers. This too is within the range of outcomes one would suspect is random.

Nor is negative seasonality more evident in the Federal Reserve’s broad trade weighted measure of the dollar. Over the nineteen year period, the dollar has risen in ten of the Decembers.

Another Way to Skin the Cat

Putting the dollar’s performance in the month of December within the context of the entire year suggests an alternative hypothesis than seasonality. There is a mild tendency for the dollar to move in the month of December in the direction that it moves for the whole year.

In thirteen of the past nineteen years, the dollar has moved against the euro in the month of December in the direction of the year’s trend. Against the yen, the dollar has spent twelve of nineteen Decembers moving in the same direction as it has for the entire year. December has been consistent with the direction of the year’s trend for the broad trade weighted dollar in thirteen of the past nineteen years.

The hypothesis that the dollar does in December what it did for the entire year is a better fit with the data than the idea that the dollar falls in December. However, turning it into a successful trading rule is more difficult. Consider that in the 2005-2008 period the month of December was not a trend month for the dollar against the euro at all.

Does this mean that December has become a trend reversal month? If a coin turns up heads four times in a row, is the best explanation that it is a rigged exercise? Small sample sizes can produce quirky results. For the 2001-2004 period, December was a trend month for the dollar. This was also the case for the entire 1992-1999 period.

For the record the euro finished November at $1.5005 according to Bloomberg and is more than 7% higher against this year. For the trend pattern to hold the euro would need to finish this month above there or sell off more than 7% in the last three weeks of the year to finish below $1.3971.

Versus the yen, the dollar has moved against the underlying trend the past two Decembers, but the two before that (i.e., December 2005 and December 2006) the dollar moved with the trend. For the record, the dollar closed last month at JPY86.41, according to Bloomberg. Year-to-date, the dollar has fallen about 2.7% against the yen. For the trend pattern to hold the dollar has to finish the year below JPY86.41 or rally more than 2.7% to close above JPY90.64.

Geneva Convention for Statistics

It seems true, as some wag put it, significantly tortured statistics will confess to anything. The analysis presented here is really cursory. Nineteen data points may not be sufficient to draw hard conclusions. The data used are price representations not necessarily actual traded prices. There may be other patterns and relationships on some deeper or more subtle level than explored here.

Nevertheless, selling the dollar just because it is December does not seem like a particularly compelling strategy. The demand for peeps maybe a function of seasonality, but it is more difficult to conceptualize why the demand for dollars would be driven by seasonality. And if such a pattern did exist, why wouldn’t it have been arbitraged away like many say has happened to the January effect in the stock market. Nor should investors take too seriously the measures of the dollar’s average move in December as it clearly obscures a great deal of variance. After all the average temperature of a person whose feet are in a block of ice while their hair is on fire is just fine, thank you.