Initial unemployment claims

[CARLHAYDEN]

United States Department of Labor  - Initial unemployment claims

In the week ending July 3:

The advance figure for seasonally adjusted initial claims was 454,000, a decrease of 21,000 from the previous week's revised figure of 475,000, (but)

The advance number of actual initial claims  unadjusted, totaled 463,560 in the week ending July 3, an increase of 22,560 from the previous week.

[angus]

what is seasonal adjustment?  I've always been curious about that.  How do you get t0 454000 from 463500?  Is there a formula saying that since a X people generally want to be employed in the summer, compared to Y in the other months, then we need to multiply the actual jobless rate by Y/X to compare?  Something like that?

[phk]

what is seasonal adjustment?  I've always been curious about that.  How do you get t0 454000 from 463500?  Is there a formula saying that since a X people generally want to be employed in the summer, compared to Y in the other months, then we need to multiply the actual jobless rate by Y/X to compare?  Something like that?

With respect to a hypothetical time series the underlying seasonality is either additive or multiplicative. Like sales increase by 40% in November is multiplicative or an extra 50 units are sold in December is additive.

You have to employ a seasonal smoothing parameter δ.

Additive model:
Forecast = St + It-p

Multiplicative model:
Forecast = St*It-p

St stands for the (simple) exponentially smoothed value of the series at time t, and It-p stands for the smoothed seasonal factor at time t minus p (the length of the season). This is to control for seasonality whether it be multiplicative or additive.

S if one was to generate an forecast you can enhance it by adding or multiplying the simple smoothed value by the predicted seasonal component. This seasonal component is derived analogous to the St value from simple exponential smoothing as:

Additive model:
It = It-p + δ*(1-α)*et

Multiplicative model:
It = It-p + δ*(1-α)*et/St

Here is a good example of seasonality in a time series for unemployment.
http://www.google.com/publicdata?ds=usunemployment&met=unemployment_rate&idim=county:CN060190&dl=en&hl=en&q=fresno+county+unemployment+rate

[angus]

Thanks.  That was a detailed and interesting answer.  I guess I always understood the point of the smoothing, i.e., eliminating the effect of changes that normally occur at about the same time, and in about the same magnitude, every year.  I just didn't know how it was done.

The link is interesting as well.  Seems that there's always a peak at February/March, and a valley at August/September, at least for Fresno.  When I looked at Black Hawk County, Iowa, the peak is at January and the valley is in May or June.  And the differences between peaks and valleys in Black Hawk aren't nearly as great as they are in the Fresno example.  But the trend is basically the same.  I suppose that agro and construction are the primary factors for the cycles, in both cases.  I guess you'd want to smooth these data for ease of comparison.  So we can look at Black Hawk County This Year versus Last Year (or this month versus last month, or whatever), without letting the warts and pimples distract us.  

So basically, the delta parameter is something like the ratio I'd described above, except that I hadn't considered that there might also be an additive possibility to consider.  My feeling is that for unemployment the better model is probably multiplicative, since seasonal fluctuations are probably caused by predictable factors.

[Cubby]

I've always assumed when they say initial unemployment claims and the state/national rates are seasonally adjusted, it meant for agricultural jobs. About 2-4 months after the monthly numbers are released, they are sometimes changed (adjusted). This seems dishonest to me.

The minimum wage for agricultural workers has always been lower than for all other workers. For such a small percentage of the workforce it sure manages to stand out.

[phk]

I've always assumed when they say initial unemployment claims and the state/national rates are seasonally adjusted, it meant for agricultural jobs. About 2-4 months after the monthly numbers are released, they are sometimes changed (adjusted). This seems dishonest to me.

The minimum wage for agricultural workers has always been lower than for all other workers. For such a small percentage of the workforce it sure manages to stand out.

If you are talking about the example I gave for Fresno. Ag is actually a big employer, close to 1 in 3 jobs depend on it.

[Cubby]

I've always assumed when they say initial unemployment claims and the state/national rates are seasonally adjusted, it meant for agricultural jobs. About 2-4 months after the monthly numbers are released, they are sometimes changed (adjusted). This seems dishonest to me.

The minimum wage for agricultural workers has always been lower than for all other workers. For such a small percentage of the workforce it sure manages to stand out.

If you are talking about the example I gave for Fresno. Ag is actually a big employer, close to 1 in 3 jobs depend on it.

And in Michigan perhaps 1 in 3 jobs depend on the Auto Industry but you don't see the Government  adjusting the numbers to account for seasonal variations in Autos. My point is that of all the industries in this country, Agriculture has always been treated completely different than all other sectors.

[phk]

I've always assumed when they say initial unemployment claims and the state/national rates are seasonally adjusted, it meant for agricultural jobs. About 2-4 months after the monthly numbers are released, they are sometimes changed (adjusted). This seems dishonest to me.

The minimum wage for agricultural workers has always been lower than for all other workers. For such a small percentage of the workforce it sure manages to stand out.

If you are talking about the example I gave for Fresno. Ag is actually a big employer, close to 1 in 3 jobs depend on it.

And in Michigan perhaps 1 in 3 jobs depend on the Auto Industry but you don't see the Government  adjusting the numbers to account for seasonal variations in Autos. My point is that of all the industries in this country, Agriculture has always been treated completely different than all other sectors.

I think with respect to Autos. You can always manufacture them, their production does not relate to weather, unless there is something I do not know.

But for agriculture certain crops can only, very strict on the only, be harvested in a particular part of the year.

[phk]

http://www.google.com/publicdata?ds=usunemployment&ctype=l&strail=false&nselm=h&met_y=unemployment_rate&scale_y=lin&ind_y=false&rdim=state&idim=county:CN261630&tdim=true&tstart=631152000000&tunit=M&tlen=244&hl=en&dl=en

[CARLHAYDEN]

I've always assumed when they say initial unemployment claims and the state/national rates are seasonally adjusted, it meant for agricultural jobs. About 2-4 months after the monthly numbers are released, they are sometimes changed (adjusted). This seems dishonest to me.

The minimum wage for agricultural workers has always been lower than for all other workers. For such a small percentage of the workforce it sure manages to stand out.

If you are talking about the example I gave for Fresno. Ag is actually a big employer, close to 1 in 3 jobs depend on it.

And in Michigan perhaps 1 in 3 jobs depend on the Auto Industry but you don't see the Government  adjusting the numbers to account for seasonal variations in Autos. My point is that of all the industries in this country, Agriculture has always been treated completely different than all other sectors.

I think with respect to Autos. You can always manufacture them, their production does not relate to weather, unless there is something I do not know.

But for agriculture certain crops can only, very strict on the only, be harvested in a particular part of the year.

First, if you check I believe you will find that the data relates to non-agricultural employments.

Second, there historically is a break in auto manufuacturing in the summer when the factories retool for the new models.