Mr. SHEPHERD. We have used the information collected by the A. A. A., so far as we could, but there are not many places where it is collected with this particular thing in mind, and it is a lot of work, a lot of expense to try to get something that represents the true yield, and we have not depended on them as much as we have on other sources of information. If we had that information over a period of years checked up against the census and see to what extent it is reliable, or check it up against their indications, against actual differences in wheat, then we would know to what extent we could rely on it. We are rather slow to shift to new bases until we try it out.

Senator MCGILL. Do you use, in making up your figures, the census reports as taken by a source in counties and townships and so on?

Mr. SHEPHERD. Wherever we have them. They are taken at different times, different degrees of reliability. In some places they give simply the acreage seeded, in some places acres that they expect to harvest. In each case we use them just as far as they are available. That covers quite a bit of the wheat territory.

Senator MCGILL. I know in my State, Kansas, the assessment is taken as of March 1.

Mr. SHEPHERD. There is heavy abandonment after that date at times.

Senator MCGILL. Yes; it is not absolutely reliable, the estimates that we get then. Yesterday a Member of Congress said to me "I raised on my land 1,500 bushels of wheat last year, and we are going to undertake to issue a pool of insurance based on average yield. How does the Agricultural Department know that I raised 1,500 bushels last year, except for what I say?"

Mr. SHEPHERD. They do not.

Senator MCGILL. And they simply would not know, would they? Mr. SHEPHERD. No.

Senator MCGILL. And in that State, so a party told me, they do not have the assessors take any statements from the farmers as to the number of acres planted to any given crop.

Mr. SHEPHERD. In some States we have acres harvested and production. In Iowa, for example, which is our main State, not for wheat but it is very helpful for corn because it helps us sometimes there is more shrinkage or less shrinkage of what the farmer expects and what they get in the spring, but the results from the previous year help us to show the percentage of farmers in other States, the error in their fall estimates. But as to estimating individual yields, individual farms, we have very little to help in the insurance program, where could help, perhaps, on the past history, showing the relative risk perhaps by areas, perhaps by counties, but I don't think even by counties it would be very satisfactory.

Senator MCGILL. It would not be reliable for a county?

Mr. SHEPHERD. Well, for counties that are reasonably uniform, perhaps so, but you get into a county where it is part irrigated and part dry, part elevated, say in Colorado or California, what we would have for county average would not be very satisfactory for this purpose, as applying to individual farms.

Senator POPE. Do you have any further information or thought about the matter, Mr. Jones?

Mr. JONES. I think Mr. Shepherd has covered the ground fairly well. The matter of individual farm estimates, I think, is entirely outside of what our division can furnish. The estimates for coun

ties would be limited largely to an average for the county over a period of years, and that would be dependent in addition to the material for that county over a period of years, which would be none too positive. It would depend to a considerable extent upon the regional, the district averages.

The district averages would have to be broken down. You would have to compare them with whatever we might have on counties, in order to build a county figure. Such figures were built up during the A. A. A. work in most of the States for the important crops, and so far as we were able to determine afterward they checked out fairly well. So that something in the nature of a county average over a period of years can be approximated, but a county average for a particular year would not be very accurate.

And, as a matter of fact, we are not really equipped to attempt county estimates, except at the expenditure of considerable extra time and effort and money in order to arrive at it.

Our whole system was built up with the idea of estimating State performance, and it is only in the last few years, with the assistance of grants of funds from the A. A. A. for that special purpose, that we have been able to advance to the matter of county estimates.

Senator POPE. But you have found that the figures obtained by the A. A. A. checked pretty well with your estimates?

Mr. JONES. Yes. Well, the difficulty there is that those figures have not been worked up in such a form that we have been able to use them with full satisfaction at all. As Mr. Shepherd said, the A. A. A. had not collected them with the idea of their use for that purpose. They had not assembled the material in such form. It has been only through partial checks here and there that we were able

Senator MCGILL (interposing). You had hardly collected them over a long enough period of time?

Mr. JONES. That is true.

Senator MCGILL. Of course, there are some States where very little irrigating is done. They are dependent almost wholly upon the average rainfall. You would not be able to supply data with reference to those States by counties?

Mr. JONES. Not right off.

Mr. SHEPHERD. We have better figures relative to yields by counties in a reasonably uniform area, like the Panhandle of Texas, for instance. Our yield figures might be fairly helpful, but as to estimated acreage in individual years, that varies so much between farms that we do not have much on acres for individual counties.

Senator McGILL. It is rather difficult for me to understand how we are going to arrive at farmers' average yields for the last 5 years, for instance, on a given farm, in order to base his policy of insurance. I was in hopes that we might get some data on that.

Senator POPE. I think, as stated by Mr. Green, the A. A. A. figures are best suited for that purpose.

Senator McGILL. And that is about all there is that is suited for that purpose, I suppose?

Senator POPE. And the figures that are obtained in two or three States where they have some pretty accurate data in that respect, North Dakota and Kansas, I believe, have pretty accurate figures obtained by the States. That is available.

Senator MCGILL. Those figures obtained by the States, so far as Kansas is concerned, that is an estimate from statements given by

the farmer as of March 1, and it is not taken for the purpose of making assessments and is not necessarily accurate. It is more or less of an opinion given by each individual farmer.

Mr. SHEPHERD. Primary plantings.

Senator MCGILL. And as to how much yield, and so forth.

Mr. JONES. Only a few States get information on production. Practically all of them are confined to acreage now. Formerly they did collect production figures, but most of them do not do that now-at least, some of them did formerly. Iowa does collect some production figures. That is the outstanding example of the State that does. Most of the others no longer make the effort.

Senator MCGILL. They just show the number of acres planted to a given crop.

Mr. JONES. That, of course, is the essential thing. Yield is much more readily determined, that is the average, than acreage. Acreage is a very difficult thing; therefore they concentrate on acreage.

Senator POPE. I see Mr. Orr has come in. If you will come forward, Mr. Orr, perhaps the committee will desire to ask you some questions. I believe you work particularly with wheat in the Crop Reporting Board?

STATEMENT OF J. L. ORR, CROP REPORTING BOARD, DEPARTMENT OF AGRICULTURE

Mr. ORR. Yes; Mr. Chairman.

Senator POPE. These gentlemen have just told us generally of their method of arriving at the crop estimate. Would you have something to say about your methods particularly as to wheat and how you arrive at your estimates on wheat?

Mr. ORR. The methods do not differ generally from other crops. In the case of wheat we have more checking information; that is, we collect, so far as possible, data on railroad movement and receipts of wheat at elevators and that gives us a better measure of year-toyear changes in wheat production.

Senator POPE. And you rely very largely upon your correspondents in the various sections to furnish you the information you have? Mr. ORR. That is correct, so far as our current estimates go. When we come to revising our estimates for last year, or checking up on them, then we collect this supplementary information on disposition, which will show us if we are very much out of line with the facts.

Senator MCGILL. You have, for instance, a report as to the number of acres planted to wheat in a State?

Mr. ORR. Yes.

Senator MCGILL. Then when that wheat is harvested and threshed and taken to the elevators, you get those figures and determine the yield per acre? Is that the way you do it?

Mr. ORR. Well, yes; that has a bearing on yield per acre. Of course, there are two factors entering into production, acreage and yield per acre. In a good many States we have a pretty good check on acreage through assessors' reports or things of that kind. In those States we can determine the yield per acre quite accurately, because we have got our check information on production.

Senator MCGILL. Your check information on production? Do you get that from correspondents, and if you get it from correspondents,

where do they get their information from? Don't they get it from the elevators and mills?

Mr. ORR. Yes; we get it directly from railroads and elevators in most cases.

Senator MCGILL. The amount shipped and the amount stored or milled?

Mr. ORR. That is right, the amount received by elevators, the amount shipped out in carloads.

Senator MCGILL. Now, a State that does not supply information like that through an assessor, as to how many acres have been planted, how do you arrive at their average yield?

Mr. ORR. In a State like that we have no more information on wheat than we do on any other crop. That is, we have our census to tie to every 5 or 10 years, and our indications of changes in the increase in yield from our crop correspondents..

Senator MCGILL. Are there very many States that do not have the information gathered by assessors as to the number of acres planted in each year?

Mr. ORR. Yes; the majority of States do not have.

Senator MCGILL. Wheat-producing States?

Mr. ORR. Yes. But in the main wheat-producing area a good many of the States do have. That is, in the Northern and Central States.

Senator MCGILL. In the Middle West they generally supply that information. I know that, but seemingly some of the States in this section of the country do not supply that sort of information.

Mr. ORR. That is true. We have no supplementary information. Senator MCGILL. At the same time they produce quite a quantity of wheat.

Mr. ORR. Yes.

Senator MCGILL. How are we going to go about it to arrive at the average yield from farms in these States where they do not keep any record as to the number of acres planted?

Mr. SHEPHERD. At least you could do this: After you secure estimates of the individual farms, check them up to see whether their total sales overran actual deliveries, for example. You would be able to see generally whether they were or were not overstating their output. That is about all we could supply, but that would be a valuable test.

Mr. JONES. In the Central States.

Senator MCGILL. That would not be applicable to these States where they have not kept any statistics.

Mr. SHEPHERD. It would not be applicable where the wheat does not move from the farms.

Mr. JONES. And in those States there is a reasonably dependable check in the census which obtains the acreage and the yield 1 year in 5. Senator MCGILL. I would think you could more readily handle the situation in those States, but I have in mind these States, some of them not far away, that have no information seemingly the census returns do not show anything with reference to the number of acres planted in a year.

Mr. JONES. We have supplementary information that we receive from crop correspondents concerning acres. We have one rather important objective method of determining acreage, which is by measuring along established routes each year. We will travel 5 or 10

thousand miles along a certain series of roads and measure the frontage in various crops along that road, or a series of roads, from year to year, and based upon the change shown in those measurements from year to year we get a very good indication of changes in crop acreage. Senator MCGILL. Does the Census Bureau here have statistics with reference to the acres planted?

Mr. JONES. Only once in 5 to 10 years.

Mr. ORR. Even that is acres harvested.

Mr. JONES. We have to allow for abandonment, on which we collect information, of course, to build up the final acreage.

Senator POPE. Are there any further questions? If not, thank you, gentlemen, for your information. We will call Mr. Rowe. Will you state your full name and position, Mr. Rowe?

STATEMENT OF WILLIAM H. ROWE, SENIOR AGRICULTURAL ECONOMIST, BUREAU OF AGRICULTURAL ECONOMICS

Mr. Rowe. My name is William H. Rowe, senior agricultural economist, Bureau of Agricultural Economics. My field of work is research in agricultural insurance and I have done most of the research on the crop-insurance work.

Senator POPE. Mr. Green advises me that you have some information that would be of value to the committee. Will you just proceed to give us what you think may be helpful?

Mr. RowE. I think the testimony of the men from the Crop Reporting Board, just preceding me, indicates some of the difficulties that we have run into in going ahead with our research work on crop insurance.

I feel that there is a misunderstanding as to the basis of the yield data used for the crop-insurance work or at least that the matter is not clear. There are two sets of data. One of them is the data that was obtained in connection with the A. A. A. program on wheat. The other is the data that the Crop Reporting Board has prepared, and which has been described to you.

In connection with the A. A. A. program, the farmers reported their yields of wheat. Those yields are for the 6 years, from 1930 to 1935, and are for individual farms. Those records are the basic material that we have used to calculate our rates.

We hope to be able to use the State and county yield data of the Crop Reporting Board to adjust those rates for a longer period.

Senator POPE. About what percentage of the farms are represented in the A. A. A. data?

Mr. RowE. The total for the country was, I believe, 78 percent of the average acreage.

Senator MCGILL. It varies in different States, does it not? Mr. RowE. Yes. In some States it is as high as 94 percent. Senator MCGILL. And in some States it is a high percentage and in others not so high?

Senator POPE. Of course, the data you now refer to are from those farms which participated in the A. A. A. program?

Mr. RowE. Yes. So that with regard to getting yields on individual farms, that problem has, to a certain extent, been met. Those farms that did not participate and those on which the operators have changed since 1935 will require some further work.