Snowflakes, Snowballs, and Blizzards
3/31/2009
Other than hermits and isolated aboriginals, can anyone escape the endless flow of economic data? Even those who avoid all forms of media cannot escape it since the economy has replaced the weather as the conversational topic du jour.
This pervasive presence arises not from increased availability, but rather increased relevance. Just as we listen more intently to weather broadcasts as storms approach, so too do we attend to business statistics when economic conditions are turbulent and threatening.
Earlier this decade, when incomes were increasing, jobs were plentiful, and housing prices were soaring, few economic reports made the headlines (a sunny day is hardly front-page news). Instead, most economic reports were relegated to the business section.
Yet, today, the press is as quick to report gross domestic product (GDP) and labor market conditions as changes in crude oil inventories, the cost of insuring sovereign debt, or housing pre-sales. It is a blizzard of numbers, with some snowballs among the snowflakes.
How severe will the blizzard be? No one really knows, in part because the quantity of yesterday’s snow (or data) tells us little about tomorrow’s.
As an example: on the New York Federal Reserve Bank’s website are measures of the U.S. and other economies that consist of literally hundreds of charts. These are just a few indicators for which data are available.
Although each of these statistics is meaningful, clearly they are not equally significant (i.e., most are snowflakes). No matter what their significance may be, they are already on the ground. That is, they indicate what has happened, which provides only limited—if any—information regarding what will happen in the future. A heavy snowfall may imply slippery conditions, except if the temperature rises.
Even for indicators with general significance (i.e., snowballs), the implications for the future are not always readily apparent (which explains why economists often offer differing interpretations).
Individually, snowflakes give only limited insight into general conditions--even though they may be valuable indicators of a particular sector. For example, how much can be inferred from retail sales data for a given month, even when it is accompanied by statistics on the previous month or the prior year?
Assuming the data indicate that activity has changed relative to earlier periods, what does that mean? Is the difference due to seasonal factors or generalized price changes (e.g., inflation)? Are key components (e.g., gasoline sales) behaving differently than the sector as a whole?
Unless these things are readily apparent, or unless an understanding of them is important to your decision-making, let the snowflakes land and move on. What insight is to be derived from irrelevant information? If on the other hand the data are relevant to your business, it rises from snowflake to snowball status and an informed analysis of its context is in order.
This leaves two questions: How to distinguish snowballs from snowflakes, and regardless of the mix, what do they signal about the depth and duration of the current blizzard (recession)?
Distinguishing snowballs from snowflakes is pretty straightforward. First, identify which data sets best capture conditions affecting your business. Those are your snowballs.
Beyond those, there are a few obvious others. Since GDP reflects the general health of the economy, it bears watching. Because it is quarterly, however, it is subject to rear-view mirror limitations (i.e., it reflects what has happened).
For each quarter, the Bureau of Economic Analysis issues three GDP estimates—Advance, Preliminary, and Final—which, because they are based on increasingly complete data, become progressively more reliable.
The preliminary estimate for the fourth quarter of 2008 indicated that the national economy contracted at an annual rate of 6.2% (the advance estimate indicated a smaller decline). That qualifies as a very large snowball—one to consider in your business planning.
A second snowball is the monthly “employment situation” published by the Bureau of Labor Statistics (BLS). This publication reports the data on the national labor market, most notably, the number of employees and the number of unemployed.
Although widely publicized, there is confusion about the data, particularly with respect to the unemployment estimates. The monthly data are based on two separate surveys: one of households, the other of employers (businesses and governments). Since the results are released about two weeks after the interviews, these data are relatively timely (i.e., less prone to rear-view mirror limitations).
Estimates of the unemployed are derived from a survey of 60,000 households which is conducted by the Census Bureau and analyzed by the BLS. The estimate of unemployment is based on responses to specific questions about jobseeking activity. It does not take into account receipt of unemployment benefits. The survey also obtains information regarding those who are "marginally attached" to the labor force (i.e., discouraged jobseekers) and the unemployed.
In February 2009, U.S. employment dropped and unemployment rose for the 14th consecutive month. Indications are that this pattern will persist through 2009. Those two snowballs, and a lot of supporting snowflakes, provide insight into the depth and duration of the current blizzard (downturn): The economy will continue to shrink through much of the coming year, but the rate of contraction will slow as the combination of fiscal and monetary stimulus bolsters consumer demand and rebuilds balance sheets.
Patrick O'Keefe is J.H. Cohn's Director of Economic Research. He can be reached at pokeefe@jhcohn.com or 973-364-7724.