Tuesday, May 31, 2011

Growing probability of positive US data surprises

US economic data has disappointed significantly during the past three months. The chart below shows this with the help of the Citi US economic surprise index (in yellow). This was also one important factor driving US Treasury yields lower (in green). As a result, growth expectation for the current quarter as well as for the whole year have been scaled back. Besides overly optimistic previous expectations, key reasons are being thought to have been higher commodity prices, most notably for energy, ongoing weakness in the US housing market as well as supply disruptions related to the Japanese earthquake and nuclear disaster. However, I am convinced that a technical factor - seasonal adjustments - has played an important role but goes unnoticed (see below). Looking ahead, reduced expectations coupled with slightly lower energy prices and especially a turn in seasonal factors suggest that the US economy should start to surprise again on the positive side soon.

US data has been surprising on the downside since early spring
Source: Bloomberg

I am convinced that seasonal adjustments have played an important role in recent negative data surprises. Seasonal factors assume a significant re-acceleration of the US economy during spring following a weaker winter period. Given that the economy continues to operate with a high level of spare capacity, the seasonal swings in the economy should be less pronounced than has historically been the case (companies will fire fewer workers than usual during the winter and summer months as they have less workers anyhow and with that they will hire fewer workers during spring and autumn). Furthermore, employment in highly seasonal sectors dropped sharply during the last recession (-2mln employees in construction, -2mln in manufacturing since end 2007) whereas it grew in non-seasonal sectors (+1.4mln in education and health services). This as well should render the economy less seasonal. However, as the chart below shows, the seasonal adjustments do not reflect this.
The chart below shows the seasonal adjustments used in the US employment report to adjust the payrolls number, in blue the seasonal adjustment over the past 12 months and in red the average seasonal adjustment over the previous 10 years. Given that on average, employment was slightly higher over the past decade than now (133mln vs. 131mln), and because the economy should exhibit less seasonality, the seasonal factors should have become lower. Instead, they have even become larger!
As a result, seasonally adjusted data should be weaker than the underlying trend during the seasonally strong periods of spring and autumn and the data should be stronger than the underlying trend during the seasonally weaker months in summer and winter.
Seasonal adjustments are reversing again as summer draws closer
Source: BLS, ResearchAhead

In turn, it is probably not a co-incidence that the Citi US economic surprise index topped out at the beginning of March. This is the time when data relating to February starts being released and as the chart shows, February is the start of the period with highly positive seasonal adjustment factors. However, we are now entering the seasonally weak June-August period. Togehter with reduced expectations, the probability is becoming substantial that US economic data starts to surprise positively again!


  1. Dear Daniel,
    I agree with you that US economy could surprise on the upside soon, but only to the fact that market players reduce their expectations and not by the fact of seasonal factors. As you look to sadj. surveys as Chicago PMI or Consumer Confidence just out today, they show a huge correction in overall sentment for US economy which weighs more than distortions in seasonal factors.
    Kind regards Ralf Umlauf

  2. Dear Ralf,
    thanks a lot for your comment. Clearly, the data over the past days were very weak and not purely down to seasonal factors. My basic view remains though, that current economic data is weaker than the underlying trend would suggest but the data should become somewhat stronger than the underlying trend over the next few weeks. Coupled with lowered expectations that should create an environment where data starts surprising positively again. However, given thata most data relating to June is only released late in the month/early in July, it might well take another two weeks before that starts to become apparent.

  3. They can not meet with their problems right here.They also can not match with their project at this time.So in this regard they been both help in this respect.

    Cool Computer Cases