Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# Re: st: Running Product Function

 From Nick Cox To statalist@hsphsun2.harvard.edu Subject Re: st: Running Product Function Date Sun, 9 Dec 2012 22:54:14 +0000

```I disagree. The geometric mean fits with the idea that log scale is a
natural scale for analysis, which is true of compound growth
processes. The harmonic mean doesn't and it is not defined either for
zeros.

Nick

On Sun, Dec 9, 2012 at 10:41 PM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
> So maybe I could try the harmonic mean as well.
>
> P.S. The geometric mean is excellent for my objectives:
> psychologically if there is a positive momentum it is plausible that
> more weight is given to the latter periods, while if there is a
> negative momentum more weight is given to earlier periods (where the
> market was high)
>
> On Mon, Dec 10, 2012 at 12:25 AM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>> Thanks Nick. That was very helpful.
>>
>> On Mon, Dec 10, 2012 at 12:14 AM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>> Note also that the geometric mean of this series is 25. This implies
>>> that while the arithmetic mean gives more weight to earlier periods,
>>> the geometric mean gives more weight to latter periods.:
>>>
>>>  ameans(reduct_per)if appt==2862
>>>
>>>     Variable |    Type        Obs        Mean       [95% Conf. Interval]
>>> -------------+----------------------------------------------------------
>>>   reduct_per | Arithmetic      37    8.108108        4.152295   12.06392
>>>              |  Geometric      12          25              25         25
>>>              |   Harmonic      12          25              25         25
>>> ------------------------------------------------------------------------
>>>
>>>
>>>
>>> On Mon, Dec 10, 2012 at 12:06 AM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>>> Eventually, I improved the system and put:the following commands:
>>>>
>>>> by appt: replace reduct_per=1 if reduct_per==0
>>>> by appt: gen y1=exp(sum(ln(reduct_per)))
>>>>
>>>> I got the following table, which is exactly what I need (I don't want
>>>> the geometric mean to be set to zero):
>>>>
>>>> . list reduct_per y3 if appt==2862
>>>>
>>>>         +---------------------+
>>>>         | reduct~r         y3 |
>>>>         |---------------------|
>>>>     30. |        1          1 |
>>>>     31. |        1          1 |
>>>>     32. |        1          1 |
>>>>     33. |        1          1 |
>>>>     34. |        1          1 |
>>>>         |---------------------|
>>>>     35. |        1          1 |
>>>>     36. |        1          1 |
>>>>     37. |        1          1 |
>>>>     38. |        1          1 |
>>>>     39. |        1          1 |
>>>>         |---------------------|
>>>>     40. |        1          1 |
>>>>     41. |        1          1 |
>>>>     42. |        1          1 |
>>>>     43. |        1          1 |
>>>>     44. |        1          1 |
>>>>         |---------------------|
>>>>     45. |        1          1 |
>>>>     46. |        1          1 |
>>>>     47. |        1          1 |
>>>>     48. |        1          1 |
>>>>     49. |        1          1 |
>>>>         |---------------------|
>>>>     50. |        1          1 |
>>>>     51. |        1          1 |
>>>>     52. |        1          1 |
>>>>     53. |        1          1 |
>>>>     54. |        1          1 |
>>>>         |---------------------|
>>>>     55. |       25         25 |
>>>>     56. |       25        625 |
>>>>     57. |       25      15625 |
>>>>     58. |       25     390625 |
>>>>     59. |       25    9765625 |
>>>>         |---------------------|
>>>>     60. |       25   2.44e+08 |
>>>>     61. |       25   6.10e+09 |
>>>>     62. |       25   1.53e+11 |
>>>>     63. |       25   3.81e+12 |
>>>>     64. |       25   9.54e+13 |
>>>>         |---------------------|
>>>>     65. |       25   2.38e+15 |
>>>>     66. |       25   5.96e+16 |
>>>>         +---------------------+
>>>>
>>>>
>>>> On Sun, Dec 9, 2012 at 11:56 PM, Nick Cox <njcoxstata@gmail.com> wrote:
>>>>> That is reasonable if and only if zero is in effect a code for missing
>>>>>
>>>>> (In terms of your earlier reference, -prod()- is a user-written -egen-
>>>>> function which must be installed from
>>>>>
>>>>> STB-51  dm71  . . . . . . . . . . . .  Calculating the product of observations
>>>>>         (help prod if installed)  . . . . . . . . . . . . . . . . . .  P. Ryan
>>>>>         9/99    pp.3--4; STB Reprints Vol 9, pp.45--48
>>>>>         extension to egen for producing the product of observations
>>>>>
>>>>> Please remember to explain _where_ you obtained user-written code.)
>>>>>
>>>>> Nick
>>>>>
>>>>> On Sun, Dec 9, 2012 at 9:40 PM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>>>>> I don't have negative values, but I have zeros, in which case I can
>>>>>> replace them by one and then take the -ln()-
>>>>>>
>>>>>> On Sun, Dec 9, 2012 at 11:31 PM, Nick Cox <njcoxstata@gmail.com> wrote:
>>>>>>> Oddly enough I was thinking earlier today about how you would
>>>>>>> generalise this if any values were not positive.
>>>>>>>
>>>>>>> If any value is zero, then the product becomes zero; otherwise one
>>>>>>> would need to separate out products of -abs()- and -sign()-.
>>>>>>>
>>>>>>> Nick
>>>>>>>
>>>>>>> On Sun, Dec 9, 2012 at 9:19 PM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>>>>>>
>>>>>>>> Please ignore my previous e-mails regarding this question
>>>>>>>>
>>>>>>>> After a short additional search, I found a very nice (and well known)
>>>>>>>> trick proposed by Nick Cox to address the problem (which, from some
>>>>>>>> reason did not come to my mind):
>>>>>>>>
>>>>>>>> bysort group : gen prod = sum(ln(x))
>>>>>>>> by group : replace prod = exp(prod[_N])
>>>>>>>>
>>>>>>>> On Sun, Dec 9, 2012 at 11:05 PM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>>>>>>>> P.S. According to stata's help, the details of the author of the
>>>>>>>>> -prod()- function is:
>>>>>>>>>
>>>>>>>>> Philip Ryan
>>>>>>>>> Department of Public Health
>>>>>>>>> South Australia
>>>>>>>
>>>>>>>
>>>>>>> On Sun, Dec 9, 2012 at 10:53 PM, Yuval Arbel <yuval.arbel@gmail.com> wrote:
>>>>>>>
>>>>>>>>>> I appreciate very much your assistance in the following question:,
>>>>>>>>>>
>>>>>>>>>> I'm looking for an equivalent function for -gen y1=sum()- which will
>>>>>>>>>> calculate running product for each point in time
>>>>>>>>>>
>>>>>>>>>> In fact, what I would like to calculate is a running geometric mean
>>>>>>>>>>
>>>>>>>>>> Note also that -gen y2=prod()- does not work (i.e., stata does not
>>>>>>>>>> identify the function). Only -egen y2=prod()- works, but it generates
>>>>>>>>>> only one product for each panel, and this is not what I need.
>>>>>>>>>>
>>>>>>>>>> Finally, I tried the -amean- command,, but it simply gives summary
>>>>>>>>>> statistics of different means, and it is not a function..
>>>>> *
>>>>> *   For searches and help try:
>>>>> *   http://www.stata.com/help.cgi?search
>>>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>>>> *   http://www.ats.ucla.edu/stat/stata/
>>>>
>>>>
>>>>
>>>> --
>>>> Dr. Yuval Arbel
>>>> 4 Shaar Palmer Street,
>>>> Haifa 33031, Israel
>>>> e-mail1: yuval.arbel@carmel.ac.il
>>>> e-mail2: yuval.arbel@gmail.com
>>>
>>>
>>>
>>> --
>>> Dr. Yuval Arbel
>>> 4 Shaar Palmer Street,
>>> Haifa 33031, Israel
>>> e-mail1: yuval.arbel@carmel.ac.il
>>> e-mail2: yuval.arbel@gmail.com
>>
>>
>>
>> --
>> Dr. Yuval Arbel
>> 4 Shaar Palmer Street,
>> Haifa 33031, Israel
>> e-mail1: yuval.arbel@carmel.ac.il
>> e-mail2: yuval.arbel@gmail.com
>
>
>
> --
> Dr. Yuval Arbel
> 4 Shaar Palmer Street,
> Haifa 33031, Israel
> e-mail1: yuval.arbel@carmel.ac.il
> e-mail2: yuval.arbel@gmail.com
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/
```