Are small grade differences significant?

The answer would seem to be obvious. Taking a recent (couple
months old) issue of _Coin Values_ that's on the table behind me,
we can see that the (stated) value at that time of a 1916-D
Mercury dime in MS63 is $13,000, in MS64 is $15,000, and in MS65
$23,000. For the 1934 dime, in MS64 it's $45, in MS65 $50, and
in MS66 $75. For the 1952-D quarter, the value is $14 in MS64,
$50 in MS65, $250 in MS66, and $5,000 in MS67. For the 1881-S
Morgan, in MS64 $65, MS65 $175, and MS66 $450. Finally, for the
1927 double eagle, MS64 $900, MS65 $1,800, and MS66 $2,500. So
that would certainly sound significant, as a difference in one
grade can sometimes mean often mean a doubling or more in price,
in one instance a 20 times increase (2000%).

But there's another meaning of "significant" dealing with
statistics. Often if we have two measurable quantities we want
to know if they are significantly difference. Saying one is 64
and the other is 65 and noting "yep, different" is of course
invalid. First, all measurements contain errors or standard
deviations, and can be measured to some level of precision and
accuracy. So let's say we have 64.0(3) and 65.0(2). The
difference is now 1.0 (note before it was 1), but the difference
also has a standard deviation, which is 0.4 (making certain
simplifying assumptions). And whether 1.0(4) is a significant
difference will depend on your confidence level you are working
with. Sometimes it will be significant; often it won't be.

But coin grades are some sort of variable quantities, though with
upper and lower limits. (Of course, numerical grading was merely
a way of correlating the adjectival grades and prices of 1794
large cents in the late 1940's and to think that coin grades can
meaningfully be a continuum is absurd.) But we can, with that
extremely big caveat, note that coin grades are better understood
as being of the form 64(x) (or 64.0(x) if you incorrectly think
that we can have such a level of precision in grading). And it
would appear to be the case that x is at least "1," which would
appear to mean that grades are not so precise that one can always
call a certain coin, or maybe even _any_ coin, "a definite 64."
And intuitively this makes sense. Having grades "smeared" as
such would easily explain the crackout game, notes or assertions
of premium quality and low quality for the grade, and things like
that.

In fact, whatever x would be above, it is in fact quite likely
even greater than one! Recently American Numismatic Rarities
auctioned off the "Cardinal Collection" of early silver dollars.
This was a major high grade collection of such coins, one of the
nicest sets, and the most important sale of early dollars this
year (now that the Reiver sale has been postponed to 2006).
Important for our purposes the pieces are also highly pedigreed,
meaning that I can compare the coins present grades with grades
they had in older sales (whether raw or slabbed), for those
auctions where I have a copy of the catalog or where it can be
found online. In addition the grades given in the Bowers
encyclopedia (1993; herein as B-B) will be given. (In general I
do not list pieces where the grades listed therein are
interpretations from early 20th and late 19th century auction
catalogs).

What follows is a long discussion of such "grade pedigrees." In
some cases they will be rather consistent; in others that is not
quite so, to give an understatement. After that will be what I
think should be concluded from this.

Lot 2 was a Continental currency piece, N. 1-C listed as PCGS
MS64. It was from the first Ford sale (Stack's October 2003) as
lot 3, raw, listed as "near-choice brilliant uncirculated" or
translating from adjectival grades, MS62. Up two points.

Lot 3 was a B-615 Libertas American medal in silver (nice piece)
PCGS MS65. It was from the first Harry Bass sale (B&M 1999[?],
lot 2084) as PF64 raw. Up a point. The minting status of search
early medals is not clearly defined and so I don't attribute any
significance to that.

Lot 5 was the 1794 dollar, MS64 PCGS. It was from B&M's
"Somerset" sale (May 1993) as MS63 raw. B-B lists it as MS63,
and notes that after the sale it was slabbed as such by PCGS. Up
a point, and PCGS graded it two different ways.

Lot 6 is easy, a 1795 B-3, PCGS AU55. From B&M's Flanagan sale
(2001) lot 4204 as the same. Yay consistency!

Lot 8 was a 1795 B-4 PCGS AU58. Also ex Flanagan sale (4205) as
MS60 raw. This coin was in B-B as MS60+. Down a point or more.

Lot 9 was a 1795 B-7 PCGS MS64. This was earlier in Heritage's
January 2002 sale (lot 7165) as NGC MS64, and before that in
their September 1999 sale (6478) as the same. Before that and
most importantly it was in the Eliasberg sale (B&M 1997) as lot
2169, there as MS63 raw. Up one point, but at least the two
major services agree about that.

Lot 11 was a 1795 B-2 NGC MS65. This coin is listed in B-B also
as MS65. No change

Lot 12 was a 1795 B-1 PCGS MS62. This is another piece from the
Flanagan sale (4208), there listed as NGC MS63. Before that it
was ex Eliasberg: 2171, there as AU58 raw. Down a point, or
maybe up 3 points. Or maybe up 4?

Lot 13 was a 1795 B-12 PCGS AU55. This specimen was in
Heritage's July 2003 sale (8117) as PCGS AU53 PCGS. Earlier it
was in Superior's May 1990 sale (450) as EF40. Before that it
was in B&M's May 1989 sale (161), also as EF40. Finally, it was
in B&M's "Saunders" sale (1987, lot 372) as AU55. B-B has it
listed as EF45. So, that is down 4 points, than holding steady,
then up a point, up two more point, and up yet another to bring
it back to where it was in 1987! PCGS has given it at least two
different grades.

Lot 14 was a 1795 B-5 PCGS MS62. This was in the Benson III sale
(Goldberg) as lot 691, there listed as PCGS MS61. Up a point,
and PCGS apparently isn't sure what it is.

Lot 16 was a 1795 B-15 PCGS MS62. In the Flanagan sale (4218) it
was also listed as such. Consistency can be nice sometimes.

Lot 18 was a 1796 B-2 NGC AU58. This coin was listed twice in B-B
as AU55 and EF40. Up one or five points, or something.

Lot 19 was a 1796 B-5 PCGS MS62. In B&M's March 1989 sale (1951)
this coins was listed as AU55 raw. Up four points.

Lot 20 was a 1797 B-3 PCGS MS62. In Heritage's August 1998 sale
(8028) this was MS60 raw. Up two points.

Lot 23 was a 1798 B-2 PCGS MS62. In the Flanagan sale (4230)
this was the same. In the Eliasberg sale (2180) this coin was
listed as AU58 raw, though noted as possibly uncirculated. Up 3
points at most.

Lot 25 was a 1798 B-32 PCGS AU58. In Goldberg's September 2002
sale (501) this piece was PCGS AU55. Up a point, and PCGS still
disagreed with itself.

Lot 28 was a 1798 B-17 PCGS AU53. In Heritage's 1995 ANA sale
(6511) this piece was AU53 NGC. They agree!

Lot 31 was a 1798 B-26 PCGS MS61. In the second Pittman sale
(1667) this piece was called "almost uncirculated" raw, which
translating from David Aker's adjectival grading means AU50 as
there were no modifiers on the grade. Up 5 points.

Lot 32 was a 1798 B-9 PCGS AU58. In Heritage's September 2002
sale (7777) this specimen was listed as AU53 PCGS. Up two
points, as PCGS doesn't even get one point precision on this
specimen.

Lot 33 was a 1798 B-24 PCGS MS64. In Heritage's September 2002
sale (7790) this piece was NGC MS64. They agree again!

Lot 34 was a 1798 B-8 NGC AU53. In David Lawrence's November
2004 sale of the Richmond collection (lot 1450) this piece was
the same grade. No change.

Lot 36 was a 1799 B-1 PCGS MS63. In Stack's session of Auction
'84 (lot 1185) this piece was listed as "brilliant uncirculated"
(eg MS60 raw). Up 3 points.

Lot 38 was a 1799 B-15 PCGS MS64. In Stack's January 2002 sale
this piece (lot 1520) was called "choice brilliant uncirculated"
with superlatives, aka raw MS63 or finer. In Stack's session of
Auction '85: (1751) it was also called such, as Stack's does not
like to second guess themselves. In B-B this piece is MS60. Up
a point at most, or maybe four.

Lot 40 was a 1799 B-19 ANACS MS62. This coin was also graded as
such in ANR's January 2004 sale (141). In Heritage's September
2002 sale (7833) it was called PCGS AU55. Before that it was in
the third Norweb sale (1988) as lot 3760, there called AU50 and
raw. This single coin is listed thricely in B-B, graded as MS63,
AU58, and AU50! So, staring in 1988, it is up none, three, or
seven, then up two or down one or four, and then up four.

Lot 41 was a 1799 B-23 PCGS MS64. In Stack's October 2003 sale
(2724) this coin was listed as "brilliant uncirculated, virtually
choice" or MS62, maybe 63, raw. It is listed in B-B as MS63. So
perhaps down a point, and then up a point or two.

Lot 42 was a 1799 B-12 PCGS AU55. In ANR's January 2005 sale
(670) it was also that, as it was in Heritage's November 2004
sale (7566). Consistent!

Lot 46 was an 1800 B-10 PCGS MS64. In the Flanagan sale (4269)
it was also listed as this, while back in the Amon Carter sale
(lot 227, Stack's January 1984) this coin was called "brilliant
uncirculated" eg MS60 raw. It is also listed as MS60 in B-B. So
it started off with no change, then went up four points and
stayed there.

Lot 49 was an 1800 B-13 NGC AU58. In Stack's January 2005 sale
(2889) this piece was listed as "brilliant uncirculated" aka MS60
raw. Down a point for once.

Lot 50 was an 1800 B-14 PCGS MS64. In Heritage's January 2002
sale (7235) this coin was graded NGC MS65. Another piece down a
point!

Lot 52 was an 1800 B-17 NGC MS63. This coin was also listed as
the same grade and service in ANR's December 2003 sale (811). In
B-B it was listed as AU55. Up 5 points, then held steady.

Lot 53 was an 1800 B-17 PCGS MS61. In ANR's December 2003 sale
(812) this coin was graded as NGC AU58. Up two points, even
though AU58 and MS61 is an odd collection of grade as one is a
really nice AU whose only "problem" is just a touch of wear
whereas MS61 is a rather ugly uncirculated coin, not very
attractive.

Lot 55 was an 1801 B-4 PCGS MS64. In the Flanagan sale (4284)
this coin was called MS63 PCGS. Earlier in the Eliasberg sale
(2194) it was listed raw as MS62. Up a point, and then another.

Lot 56 was an 1802 B-4 PCGS AU58. In Heritage's April 2001 sale
(6570) it was again AU58, but graded by NGC. In Heritage's
August 1998 sale (8088) it was also AU58, and raw. Quite
consistent.

Lot 58 was an 1802 B-6 PCGS MS65. In the Flanagan sale (4293)
this coin was called MS64 PCGS. This coin is listed twice in
B-B, both times as MS63. Up a point, and then another.

Finally, lot 60 was an 1803 B-4 PCGS MS63. In the Flanagan sale
(4298) this specimen was MS63 but NGC. In the Eliasberg sale
(2197) this coin was raw and called MS62. Up a point, but at
least the services agree.

So out of these coins, 21 of them have been upgraded from their
original grades, three have been downgraded, eight have been
graded consistently, whereas three have just been all over the
place. Alternatively, taking the maximal grade differences for
these coins, eight have no difference, nine have a one point
difference, seven have two point differences, two have three
point differences, five have four point differences, three have
five point differences, and one has a six point difference. This
is an average difference of two grades per coin!

This would seem to suggest that a given grade is really only
precise to within two grades. But if an MS65 and an MS66 is then
for all practical purposes interchangeable (and I am asserting
that they are) why should there be a large difference in price?
Certainly _some_ difference in price is to be expected; one may
be able to perceive slight differences in quality between two
similar graded coins, and this of course often happens. But
slight differences should not have a large effect on price, for
if you show the two coins to someone else, they may disagree as
to which is the better example. In fact, if one examines
assorted EAC censae of early date large cents, the ordering can
differ, along with the grades given.

But if minor differences in grade should not greatly affect
price, why do they actually do so? Most importantly we must
blame the acceptance of the Sheldon grading system. Remember
that all it was was a statement of a correlation of grade of
price of 1794 large cents, and beyond that states nothing more as
to how much better a "65" is over a "60;" that is what adjectival
grading is for. More importantly, having whole numbers between
the three original uncirculated grades (60, 65, and 70) allowed
later people to add new grades if it was perceived necessary.
And thus was born 63 and 67, and then the rest of them. As it
became popularly accepted that a higher grade meant more and more
money (especially when MS65 and above suddenly meant "investment
grade") the grade/price curve became more curved and steeper in
the higher grades.

Of course the grade/price curve can work both ways, and so a coin
bought as a 65 may suddenly upon selling may be a 64. But this
would mean that 20x times grade/price differentials could not be
perpetuated for long under such conditions. If your sellers
thought their 65s would eventually be 64s, why buy 65s and 20x
the price? But if the dealers decided to find "independent"
grades for their coins, by companies not set up by long time
collectors or well-published numismatics but rather by a set of
dealers specializing in "investment grade" coins, then the
grade/price scheme may be continued. (As always this should not
be construed by saying anything bad about the vast majority of
dealers.) And this is good for them, as typically, the more
expensive the coin, the greater the profit (on a dollar basis).
And besides, certainly investment grade dealers can tell between
"high for the grade" (eg coins they are selling) and "low for the
grade" coins, so and with such an extreme grade/price scheme, all
is well in investment grade coin dealer land.


--
Ed. Stoebenau
a #143

"Ed. Stoebenau" wrote in message
...

This would seem to suggest that a given grade is really only
precise to within two grades. But if an MS65 and an MS66 is then
for all practical purposes interchangeable (and I am asserting
that they are) why should there be a large difference in price?


MS66--Must have above average quality of strike and full original mint
luster, with no more than two or three minor but noticeable contact marks. A
few very light hairlines may show under magnification, or there may be one
or two light scuff marks showing on frosted surfaces or in the field. The
eye appeal must be above average and very pleasing for the date and mint.
Copper coins display full original or lightly toned color as designated.

What if there's no marks at all with zero luster? Maybe I think a
flat-looking finish is very pleasing? It says "lightly toned" for
copper--what is that? Some wild toned Lincolns are getting crazy money in
this grade. According to this a wildly toned copper coin should never be
above a MS66.

MS65--Shows an attractive high quality of luster and strike for the date and
mint. A few small scattered contact marks may be present, and one or two
small patches of hairlines may show under magnification. Noticeable light
scuff marks on the high points of the design. Overall quality is above
average and overall eye appeal is very pleasing. Copper coins have full
luster with original or darkened color as designated.

What if there is tons of luster with one very noticable large gash, no scuff
marks on the high points and no hairlines?

MS64--Has at least average luster and strike for the type. Several small
contact marks in groups, as well as one or two moderately heavy marks may be
present. One or two small patches of hairlines may show under low
magnification. Noticeable light scuff marks or defects might be seen within
the design or in the field. Attractive overall quality with a pleasing eye
appeal. Copper coins may be slightly dull. Color should be designated.

What if there's lots of very small marks all over, lots of luster and above
average strike?

The combinations are endless really.
With Jeffersons there is supposed to be sharply struck creases on his
shoulder area to ever get a 66--if I don't see that I figure they are not a
66. Because it is average for most years of Jeffersons to not find many with
sharply defined jacket and collar creases. Then you have that full step
designation. I've seen that 66+ handed out on some where the middle steps
are all there but the side steps are just totally missing. So I say those
are not ever going to be a 66 with full steps because they may be attractive
strikes but they are not a high quality strike. IOW I figure a Jefferson
with ALL the details present might get a 66+ unless they got gashes the size
of Montana or they don't have much luster.


Certainly _some_ difference in price is to be expected; one may
be able to perceive slight differences in quality between two
similar graded coins, and this of course often happens. But
slight differences should not have a large effect on price, for
if you show the two coins to someone else, they may disagree as
to which is the better example. In fact, if one examines
assorted EAC censae of early date large cents, the ordering can
differ, along with the grades given.

But if minor differences in grade should not greatly affect
price, why do they actually do so?

A lot of prices go off of population reports. And we all now those are not
very accurate. Then you also can get people who bid up certain coins which
will mess up the real values. Dealers will see someone get outrageous prices
for something and figure it's time to raise their prices. That's what
happened with wild toned coins. Those never brought a whole lot of money in
the past and then all of the sudden people started to pay top dollar--that's
why you see so many coin doctors getting rich right now. When they started
putting more stock into overall eye appeal that opened up the floodgates.
Perceptions are 99% of the game. Someone at one time had paid 89 bucks for
my '62 mint set--it still has a price sticker on it for that amount so at
one time people went nuts over that for some reason. I paid something like
$25 for it a few years back.
'99 silver proof sets anyone?
Check out some prices for some high grade early-to-mid '90's Jefferson
nickels. Those were minted in the bazillions in fairly high grades but
hardly anyone thought those were worthy. All it takes is a few shots at
buying raw BU to get some pretty high grades with those but there is hardly
much interest in those with the public in general. Average Joe finds a wheat
cent in circulation and they think they really have something when an hour
earlier they spent an MS67 1992D Jefferson they had sitting in a jar for
years. Hardly anyone pays attention to that sort of thing. You see that all
the time here in the group--people ask all the time about something really
common as dirt but pay absolutely no attention to that grading system. You
never see someone come on and say they got a '98 Lincoln with no marks and
superior luster and they want to know what it's worth.


Most importantly we must
blame the acceptance of the Sheldon grading system. Remember
that all it was was a statement of a correlation of grade of
price of 1794 large cents, and beyond that states nothing more as
to how much better a "65" is over a "60;" that is what adjectival
grading is for. More importantly, having whole numbers between
the three original uncirculated grades (60, 65, and 70) allowed
later people to add new grades if it was perceived necessary.
And thus was born 63 and 67, and then the rest of them. As it
became popularly accepted that a higher grade meant more and more
money (especially when MS65 and above suddenly meant "investment
grade") the grade/price curve became more curved and steeper in
the higher grades.

Of course the grade/price curve can work both ways, and so a coin
bought as a 65 may suddenly upon selling may be a 64. But this
would mean that 20x times grade/price differentials could not be
perpetuated for long under such conditions. If your sellers
thought their 65s would eventually be 64s, why buy 65s and 20x
the price? But if the dealers decided to find "independent"
grades for their coins, by companies not set up by long time
collectors or well-published numismatics but rather by a set of
dealers specializing in "investment grade" coins, then the
grade/price scheme may be continued. (As always this should not
be construed by saying anything bad about the vast majority of
dealers.) And this is good for them, as typically, the more
expensive the coin, the greater the profit (on a dollar basis).
And besides, certainly investment grade dealers can tell between
"high for the grade" (eg coins they are selling) and "low for the
grade" coins, so and with such an extreme grade/price scheme, all
is well in investment grade coin dealer land.


--
Ed. Stoebenau
a #143

And the someone shows up with a thousand MS66 coins where there might have
been only 10 known of previously.