extracted text

PROFESSOR DR J . TIWi3l!,RGEN

DEN HAAG, 2? F e b r u a r y 1950
Sijzenlaan ^ 1 .

Dr R.J. Saulnier,
Nationalf*ëlireau of Economie Research
Incorporated,
Financial Research Program,
•Hillside', W. 25'+th Street and
Independence Avenue,
New York 63, N.f-.
Dear Dr Saulnier,
Thank you very much for sending me Mr Kisselgoff's study
on Instalment Credit. I think it is very interesting indeed, and
I have only a number of minor remarks to make.
At page 1 - 6 I do not quite understand the comparison made
between the percentage of families indebted having an income of
less than $5»000 with the percentage of 77»^ their aggregate
income formed of total national income. It seems to me that you
have either to compare a number of families in both cases or
aggregate income in both cases.
I have a general remark to make on all the graphs of the
type of chart 5. I wojild like strongly to recommend that the
various curves shown in such graphs be drawn at the same scale,
since only then it can be seen from the graphs what the relative
influence of the various components is.
Further I would suggest to add in the title the indication:
equation (3»3)> which facilitatrs the comparison with the tables.
As to page II - 2h 1 am not quite happy about the argument
in the foot note. The choice between (3»3) and (35"+) should be
based, in my opinion, on economic arguments and not on statistical
ones. If there would be strong arguments of an economic nature
for including T, the fact that equation (3A) shows a larger
margins of error in the regression coefficients should not withhold
you from taking(3>^)» We could only s?.y in that case that unfortunately by accident intercorrelation in that case prevents us from
measuring more exactly the regression coefficients. Now the last
argument given on page II - 2^. viz. that it can be hardly
reconciled with the findings in the other cases, seems acceptable
to me. One could therefore better leave out the other arguments
for choosing (3j3)«
At page II - 3^ I v/onder hov^r the reliability of the equation
(U-,8), represented in chart 9» has been calculated; from the graph
I deduce a high rate of intercorrelation between current income
and size of monthly payment, v/hich might badly influence the
reliability of their coefficients. This might be a case where a
bunch map analysis night be applied.
At page II - h9 I wonder whether you should not show the
results given there also in the form of charts.

Apart, however, fïbrrrthëöe' few' remarks I do think that Mr Kisselgcfi"!£
study is highly interesing and I want to congratulate, you and him with "'
accomplishment.
, :sg?|g|p^,,.^%^'.r.
Yours ve'4*y sincerely,

sMsii^i^-

i^^%

^^^•^r'^l^fw 1
'^'•^^fsfe^ifeiSÖI^
^^?^S
iS-'^'f^.^'H:'!:;' .0''-i

"3f/

a*i
.,*^.:.v!-—'

-t c'i"'^'ö"*^^'"'''"."A-nT4'Y,'." -•'1 A - * r " - -V ,-'

'^;£L

;i^M^j^P^-f

*rt*'0;rt'cl