In April, the unemployment rate made a surprise uptick, to 9.0%, even though the jobs number came inbetter than expected.

The reason: the household survey got it wrong, according to Deutsche Bank's Joe LaVorgna.

Each month's jobs number includes two surveys: one the household survey, the other, the establishment survey. The household survey is used to compute the unemployment rate, and surveys 60,000 households across the U.S.. The establishment survey checks in with around 400,000 businesses, and is the number reflected in the headline jobs number released with the Employment Situation report.

For the month of April, the establishment survey, which surveys 400,000 establishments, showed job growth, while the household survey, which only surveys 60,000 households, showed a net job loss.

LaVorgna says that the household survey in April ignored imminent improvements to come in the agricultural employment market.

Agricultural employment can be quite volatile, and the April decline could have been due to weather. According to the USDA, the weekly crop report indicated that only 13% of the corn crop was planted by May 1, which is the slowest pace since April-May 1995, a period when farming employment declined -285k. The report noted that other agricultural activity was also shut-in by inclement weather. This disruption should undoubtedly prove to be temporary and reverse in the coming months.

Further, LaVorgna says the household survey got the government employment number wrong (the BLS survey is better at this) and state and local job cuts should start to slow soon as revenues rise. One drag remains, with those that are self-employed now seeking full-time employment at firms, according to LaVorgna,

The result is that the unemployment rate should "fall hard" to 8.7% in May due to improvements in the household survey, according to Deutsche Bank's LaVorgna. To put that in context, it's only a tenth lower than March's 8.8% rate.