Farrell: Earnings Surprises (MER/MSFT vs. JPM/IBM)

Just when it looked safe to get back into the water, sharks were sighted.

Merrill Lynch showed a bigger loss than expected (a much bigger loss), Microsoft was indeed soft and Google could have done better. On the other hand, JP Morgan surprised us as did United Technologies and IBM . So maybe you pay your money and you take your chances.

My best guess is unless Citi comes in Friday morning with a positive surprise, we will use Merrill and Microsoft -- but especially Merrill -- as an excuse to take some profits from this wonderful two-day rally.

Oil prices cooperated on Thursday, falling to below $130. The unemployment claims we follow so closely were up less than expected and the all-important four-week moving average was at 376,000 -- which still indicates sluggish growth.

Because of the strength of our exports and the consumer spending generated by the tax rebates, it looks like second quarter GDP will easily exceed 2 percent. Some optimistic folk are thinking as much as 3 percent.

There is some Congressional talk about another rebate, but I hope they lie down and let that feeling pass. Rebates artificially raise sales and artificiality is a Washington specialty -- but if we want to stimulate the economy via deficit spending let's spend money on infrastructure.

Mark Zandi of Moodys.com figures that "within the first year on enactment, increased infrastructure spending and aid to states generates roughly four times as much economic growth" as a bunch of other alternatives. But that "within the first year" time line puts you out past the election, so why bother if you won't get votes..?

Housing starts were startlingly good, but it was due to a quirk in the NYC building code which caused plans for multi-family home construction to be moved forward. Without that the number would have been poor. Actually, that is okay, in that we need to reduce the inventory overhang before we can be clear of the housing debacle -- and we may as well accept that.