Country ... USA
Politics 2012 ... Bayesian metrics

Economy, and Obama, Mired in Down Data

Burgess COMMENTARY
When I read this article, it helped to confirm for me one of the deisgn aims of the TrueValueMetrics system. I have always thought that a single datapoint was not very useful. On the other hand two datapoints started to tell the user something, and maybe something of significance. For example, the same subject, but at a different time starts to show trend over time. The same subject at the same time in different places shows the difference between two locations, something that can be very interesting and useful. There are a myriad of other examples.

A lot of datapoints improves the reliability of the information, but quite quickly a point is reached where more data does not produce more information, or better information.

At some point the interaction of different activities needs to be brought into the analysis ... and to some extent the Bayesian method helps to do this. While I am not skilled in advanced statistical mathematics, I have used something which might be described as a basic common sense simplification of the idea, and have frequently looked not only at the way trend lines of datapoints related one to another, but how moving averages of datapoints moved. This 'smoothed' the single points into something that was often far easier to interpret and then more useful for decision making.

An added advantage of my simplistic procedure was that it was far easier to explain the conclusions to critical decision makers who probably would not be able to follow the statistical explanation in a useful way. My simplification may not have been the best way to get the statistics right ... but it did work to get the decisions right.
Peter Burgess

Economy, and Obama, Mired in Down Data

Once we introduce our presidential forecasting model next week, which is based in part on current and projected economic data in addition to the polls, we’ll be able to estimate in real time how something like the poor May jobs report would imperil President Obama’s re-election chances.

The short version: this is a case where our model’s insight would match the conventional wisdom — the bad economic numbers over the past weeks are sapping Mr. Obama of the edge he might have had on Mitt Romney.

It would be one thing if there were just one poor set of jobs figures. Economic data, like political polls, can be subject to measurement error and other idiosyncrasies. One data point — however good or bad — doesn’t usually tell you all that much. The model accounts for this problem by looking at several different economic data series in addition to the jobs numbers.

However, almost every significant economic figure that has come out over the past week has come in below expectations, reversing a pattern from earlier in the year.

The stock market is also down sharply as investors assess their concerns over the American economy as well as the problems in Europe. There are also some troubling numbers from China, which had been one of the few bright spots in the global economy over the past few years.

Sometimes the internals of a jobs report look better than the headline numbers, but this isn’t really one of those cases. In particular, the government revised downward its estimates of job growth in March and April, essentially wiping out the small 69,000-job gain that was said to occur in May.

If you want to find a bright spot, you would instead have to look at the household survey, from which the unemployment rate is calculated. Although the unemployment rate ticked up by one-tenth of a point, the number of employed persons increased and the unfavorable change in the unemployment rate occurred because more people were seeking jobs.

Another favorable data point, of sorts, is oil prices, which have declined substantially because of fears of a decline in global demand.

Oil prices have become a strongly counter-cyclical indicator in recent years. Prices tend to increase when the economic outlook is brighter — potentially creating headaches for consumers and undermining plans by businesses for expansion. Prices decline as pessimism comes to dominate again, creating a little bit more slack in the economy.

Thus, energy prices have behaved as something of a thermostat on the economy. But it’s a thermostat that may be tuned to a too-cool-for-comfort temperature given the tepid overall economic performance of the past few years.

Indeed, the American economy as a whole has had a strong tendency to revert to the mean since we officially exited the recession in June 2009. Promising economic quarters have been followed by negative surprises. The reverse to some extent has also been true: the set of relatively good jobs numbers that were printed in January through March came after a difficult period in late 2011 when fears of recession were increasing.

In general, however, the lows have been lower than the highs have been high. At no point in the past three years has the economy shown sustained growth above the long-term trend, as it will eventually need to do to make up for the loss in productivity and jobs it suffered during the recession.

It is also the case, however, that the statistics have not been recessionary — and incumbent presidents, especially in their first term after their party takes over the White House, have typically received a lot of credit from voters, getting a pass more often than not when they have an economic story to sell.

Mr. Obama’s approval ratings have not yet declined discernibly after increasing earlier this year. His head-to-head polls against Mr. Romney have so far declined only mildly, and are better at the state level than in national tracking polls.

The concern for Mr. Obama, however, is not so much what might happen if the election were held today but what will happen in November. The economic news in recent weeks has been poor enough to suggest that if the economy finally breaks from its pattern of tepid growth in the second half of the year, it could be toward the downside.