Seasonal patterns are common-place and well established in many commodity markets.
The grain markets have a strong tendency to show rising prices from late spring into midsummer and subsequent recessions into late summer and early fall. Crude oil has a bias toward low prices in midsummer and a subsequent rally into late fall. This article discusses whether seasonal patterns exist in the long-term municipal bond market.
In this discussion, seasonality will mean the degree to which any pattern or tendency recurs from one 12-month period to another 12-month period. Data and analyses will be presented in an attempt to allow the reader to decide whether or not such a tendency exists. Next, a means to utilize the phenomenon is suggested, and finally, possible reasons for its existence are listed.
It may be easy to accept an explanation for the seasonality of the grain markets - drought, flood, diseases, uncertainty until harvest - or to believe an explanation for crude oil's seasonality - low commercial purchases in summer followed by heavy purchases in fall. It is another issue, an possibly an irrelevant one, to explain the forces behind any recurring pattern in the tax-exempt bond market. If a seasonal pattern can be detected here. it may be enough to appreciate the odds it suggests. The value of a seasonal is the value of odds and the extent to which future periods mimic the past.
The data for this analysis were provided by The Bond Buyer. The monthly high and low yields for the 20-Bond Index of 20-year general obligation bond yields was the longest continuous time series available. It dates back to 1915 and is used here.
It is best to define the use of a tool before trying to create it. Pose the question: "What good would seasonal tendencies be to anyone?" If a person knew the chances of profit and loss of a strategy of making, on balance, purchases in one or a number of months and, on balance, sales in another or other consecutive months every year, then the tool would be of value.
The tables in the accompanying chart show just that. In the first table, the numbers filling the grids are the number of times in the 78 years that a column month's yield exceeded a row month's yield for the immediately following 12-month period. The raw data from The Bond Buyer are high and low yields for each month, but here average yields for each month are used, because the chances of buying or selling at the average of any month are greater than at the highs or lows.
[TABULAR DATA OMITTED]
The second table in the chart shows probabilities. Each grid number is the percent of 78 years that the row month's yield exceeded the column month's yield. There are 78 years of 11 possibilities each year (comparison of a month to itself is taken out), for a total of 858. As a percent of 858, various total probabilities can be assigned for months of the reader's choice. These are listed as well.
[TABULAR DATA OMITTED]
The results show February has the best numbers for sales; it is a 530/858, or 62%, bet that its yield will be exceeded in some month in the immediately following 11 months. This is considerably above the next most likely month, January, with 487/858, or 57%. The month whose yield exceeds all other months most frequently is September. Its total was 477/858, or 56%. It is closely rivaled by June, July, and December.
A gambling casino can make very good money with 52% to 48% odds of winning and losing. Here we have 56% buying odds and 62% selling odds in a two-part process. As for a casino, these odds are useful only over numerous tries.
The above analysis shows that buying in September gives the best odds of buying the highest yields for the ensuing 12 months. It also shows that independent of this, selling in February gives the best odds of selling the lowest yields of the ensuing 12 months. When buying with no sell date in mind or selling with no repurchase date in mind, these are the best independent months to use.
This is not the same as consistently buying in one month and selling in another. When such a strategy is employed, choosing to buy municipal bonds in July and selling them the following February made money 67% of the time in the past 78 years. December-February is the runner-up, with 65%. In fact, it looks like a consistent 63% bet buying after midyear and selling the following February.
This system may or may not work in the next 78 years. Also, the reader has to decide how significant these odds are. Still, it is true that had an investor of municipal bonds waited until at least the nearest July to make intended purchases and used Februaries to make intended sales, odds of profit would have been increased in the last eight decades relative to a random-month program.
No matter how the facts are displayed, February is the best month to sell municipal bonds. The belief and reality that the future does not deviate significantly from the past is necessary for these odds to work.
Reasons if Needed
Reasons behind any municipal bond seasonal pattern may be unimportant - that they display odds of success should be sufficient. But for those who desire reasons, these may have something to offer:
* (1) Tax payment day is April 15 every year. Tax-exempt bond purchases may become more urgent as this realization hits people in February and the desire to shelter interest income becomes great, even though it won't help until next year.
* (2) By February, winter's end is in sight. Thoughts of spring, new growth, flowers, and crops all are upon us. This is a time for new beginnings, new plantings, and new investments.
* (3) By September, the annual November elections are being prepared on everything from local budgets (especially school board budgets) to judges and legislatures. Elections seriously affect investment decisions by heightening uncertainty. Human reaction is to take profits or just raise cash.
* (4) Fall is harvest, time to reap the products of one's labor and investments. September is the realization that harvest and eventually winter are coming, a time of death and endings. It is time to take profits or your money out of the markets.
Stephen Kowal is director of research and trading for Tapworks. Before that, he was a vice president and municipal bond trader at Marine Midland Bank.