Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Guidance on matrix inversion for OLS in mata


From   Thomas Jacobs <thomasjacobs@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Guidance on matrix inversion for OLS in mata
Date   Tue, 27 Apr 2010 12:12:46 -0500

Antoine,

Thanks again.  That worked like a charm.  For the future reference of
anyone else going down this path the final solution was

invsym(cross(X,X))*cross(X,Y)

without any primes on the matrices and X already has the column of 1s
included.  This produces identical values to regress even if both X
and Y have missing values.

Tom

On Tue, Apr 27, 2010 at 11:34 AM, Thomas Jacobs <thomasjacobs@gmail.com> wrote:
> Antoine,
>
> Thanks for the reply.  I caught my first error but appreciate the
> suggestion as it addresses the missing values issue automatically.
>
> Tom
>
> On Tue, Apr 27, 2010 at 11:29 AM, Antoine Terracol
> <Antoine.Terracol@univ-paris1.fr> wrote:
>> Hi Thomas,
>>
>> have a look at cross() in the mata help files.
>>
>> invsym(cross(X,X)) produces (X'X)^-1
>>
>>
>> Antoine
>>
>> On 27/04/2010 18:17, Thomas Jacobs wrote:
>>>
>>> Austin,
>>>
>>> Thanks for the reply.  I guess I have no choice but to do as you
>>> suggest or else rewrite the program entirely in Stata.  I don't have
>>> the days it always takes me to decipher how to do the stata mata data
>>> transfer for this project at present.  Every time I sit down to do
>>> something like this I always feel it a roll of the dice.  Do I go the
>>> mata route and discover what should have been done in stata or go the
>>> stata route and discover the opposite!
>>>
>>> Here is a short excerpt of the problem.  I am simply trying to compute
>>> x'x inverse for a 100 trading day time series of index returns and use
>>> it to solve for betahat for a similar time series of firm returns
>>> where the latter often has one or more missing values.
>>>
>>> Here are the initial steps I take in mata and will use an example with
>>> no missing values:
>>>
>>> X=J(100,2,1)
>>> X[.,2]=LnBAATreasSprd[800::899]
>>> : X
>>>                     1              2
>>>       +-------------------------------+
>>>     1 |             1    -.005838091  |
>>>     2 |             1   -.0063325767  |
>>>     3 |             1    .0100309728  |
>>>     4 |             1    .0040479107  |
>>>     5 |             1     .000136131  |
>>>     6 |             1      .00040839  |
>>>     7 |             1   -.0148075884  |
>>>     8 |             1   -.0107393684  |
>>>     9 |             1    -.002842935  |
>>>    10 |             1    .0066983248  |
>>>    11 |             1   -.0093153818  |
>>>    12 |             1   -.0502348617  |
>>>    13 |             1   -.0178726967  |
>>>    14 |             1    .0145704737  |
>>>    15 |             1   -.0073333359  |
>>>    16 |             1   -.0029384219  |
>>>    17 |             1   -.0095219389  |
>>>    18 |             1   -.0128216762  |
>>>    19 |             1   -.0150571838  |
>>>    20 |             1     .005833914  |
>>>    21 |             1    .0069765295  |
>>>    22 |             1    .0038266596  |
>>>    23 |             1   -.0067447214  |
>>>    24 |             1   -.0082368711  |
>>>    25 |             1    .0079292208  |
>>>    26 |             1   -.0185827948  |
>>>    27 |             1   -.0081839114  |
>>>    28 |             1    .0103758555  |
>>>    29 |             1    -.007115229  |
>>>    30 |             1   -.0043146866  |
>>>    31 |             1    .0007379653  |
>>>    32 |             1   -.0010016384  |
>>>    33 |             1    .0068858997  |
>>>    34 |             1   -.0044625248  |
>>>    35 |             1    .0031520503  |
>>>    36 |             1   -.0554844476  |
>>>    37 |             1    .0113020828  |
>>>    38 |             1    .0031746107  |
>>>    39 |             1    .0117888227  |
>>>    40 |             1   -.0010267079  |
>>>    41 |             1    .0095236665  |
>>>    42 |             1    .0261100251  |
>>>    43 |             1   -.0126527818  |
>>>    44 |             1   -.0114778923  |
>>>    45 |             1    .0066052717  |
>>>    46 |             1   -.0084775724  |
>>>    47 |             1   -.0032715374  |
>>>    48 |             1   -.0015591006  |
>>>    49 |             1    .0142634539  |
>>>    50 |             1    .0003712231  |
>>>    51 |             1   -.0077179475  |
>>>    52 |             1    .0082479911  |
>>>    53 |             1    .0071813553  |
>>>    54 |             1    .0058756177  |
>>>    55 |             1   -.0082988022  |
>>>    56 |             1   -.0003692649  |
>>>    57 |             1   -.0046538925  |
>>>    58 |             1   -.0049419519  |
>>>    59 |             1    .0040403828  |
>>>    60 |             1   -.0099174296  |
>>>    61 |             1    .0231989622  |
>>>    62 |             1   -.0097859399  |
>>>    63 |             1   -.0028060512  |
>>>    64 |             1   -.0028140107  |
>>>    65 |             1    .0092088645  |
>>>    66 |             1    .0032609063  |
>>>    67 |             1    .0080019441  |
>>>    68 |             1    .0002604046  |
>>>    69 |             1    .0027563504  |
>>>    70 |             1   -.0134371053  |
>>>    71 |             1    .0051452187  |
>>>    72 |             1    .0188821666  |
>>>    73 |             1    .0129679879  |
>>>    74 |             1    .0097926175  |
>>>    75 |             1    .0015559109  |
>>>    76 |             1   -.0006521898  |
>>>    77 |             1    .0055551799  |
>>>    78 |             1    .0016954662  |
>>>    79 |             1              0  |
>>>    80 |             1   -.0011965502  |
>>>    81 |             1   -.0123975417  |
>>>    82 |             1    .0008581794  |
>>>    83 |             1    .0113398973  |
>>>    84 |             1   -.0044504236  |
>>>    85 |             1   -.0101748193  |
>>>    86 |             1    .0006073291  |
>>>    87 |             1    .0028294155  |
>>>    88 |             1   -.0016158026  |
>>>    89 |             1    -.009393123  |
>>>    90 |             1   -.0073215333  |
>>>    91 |             1    .0056367237  |
>>>    92 |             1   -.0044041635  |
>>>    93 |             1    .0046085101  |
>>>    94 |             1    .0162160564  |
>>>    95 |             1    .0047139199  |
>>>    96 |             1   -.0005505134  |
>>>    97 |             1    .0033981025  |
>>>    98 |             1     .034180887  |
>>>    99 |             1   -.0102245966  |
>>>   100 |             1   -.0060580331  |
>>>       +-------------------------------+
>>>
>>> and the inverse step with a partial excerpt (all values are zero save
>>> for three cells):
>>> : invsym(X*X')
>>> [symmetric]
>>>                     1              2              3              4
>>>          5
>>>
>>> +----------------------------------------------------------------------------
>>>     1 |             0
>>>     2 |             0              0
>>>     3 |             0              0              0
>>>     4 |             0              0              0              0
>>>     5 |             0              0              0              0
>>>          0
>>>     6 |             0              0              0              0
>>>          0
>>>     7 |             0              0              0              0
>>>          0
>>>     8 |             0              0              0              0
>>>          0
>>>     9 |             0              0              0              0
>>>          0
>>>    10 |             0              0              0              0
>>>          0
>>>    11 |             0              0              0              0
>>>          0
>>>    12 |             0              0              0              0
>>>          0
>>>    13 |             0              0              0              0
>>>          0
>>>    14 |             0              0              0              0
>>>          0
>>>    15 |             0              0              0              0
>>>          0
>>>    16 |             0              0              0              0
>>>          0
>>>    17 |             0              0              0              0
>>>          0
>>>    18 |             0              0              0              0
>>>          0
>>>    19 |             0              0              0              0
>>>          0
>>>    20 |             0              0              0              0
>>>          0
>>>    21 |             0              0              0              0
>>>          0
>>>    22 |             0              0              0              0
>>>          0
>>>    23 |             0              0              0              0
>>>          0
>>>    24 |             0              0              0              0
>>>          0
>>>    25 |             0              0              0              0
>>>          0
>>>    26 |             0              0              0              0
>>>          0
>>>    27 |             0              0              0              0
>>>          0
>>>    28 |             0              0              0              0
>>>          0
>>>    29 |             0              0              0              0
>>>          0
>>>    30 |             0              0              0              0
>>>          0
>>>    31 |             0              0              0              0
>>>          0
>>>    32 |             0              0              0              0
>>>          0
>>>    33 |             0              0              0              0
>>>          0
>>>    34 |             0              0              0              0
>>>          0
>>>    35 |             0              0              0              0
>>>          0
>>>    36 |             0              0              0              0
>>>          0
>>>    37 |             0              0              0              0
>>>          0
>>>    38 |             0              0              0              0
>>>          0
>>>    39 |             0              0              0              0
>>>          0
>>>    40 |             0              0              0              0
>>>          0
>>>
>>> qrinv gives something similar with a few more non-zero values while
>>> cholinv and luinv produce nulls.  pinv does work in this case.
>>> Suffice it to say I have no problem regressing a similar single firm
>>> return series on this index for the time period in question using
>>> regress.
>>>
>>> If I am missing something, please let me know.  Otherwise, I guess I
>>> need to go to stata or do a better job of reproducing regress!  Thanks
>>> again.
>>>
>>> Tom
>>> On Tue, Apr 27, 2010 at 10:23 AM, Austin Nichols
>>> <austinnichols@gmail.com>  wrote:
>>>>
>>>> Thomas Jacobs<thomasjacobs@gmail.com>:
>>>> Zero is not a problem, but you should expunge the missings first;
>>>> however you seem to be trying to rewrite -regress- as you go, which is
>>>> far from a good idea. Why *not* export your vectors to Stata and run
>>>> -regress- and let Stata handle the sample selection and matrix
>>>> inversion for you?
>>>>
>>>> Maybe if you give us a simple example with real numbers, the problem
>>>> will be clearer and you can get better guidance...
>>>>
>>>> On Tue, Apr 27, 2010 at 12:45 AM, Thomas Jacobs<thomasjacobs@gmail.com>
>>>>  wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> I am trying to perform a lengthy series of simulations to examine some
>>>>> event study methodologies.  I have moved to mata for the bulk of the
>>>>> work but find that for those cases where I wish to use a market model
>>>>> approach requiring an OLS regression to establish abnormal returns I
>>>>> am unable to generate an inverse for x'x in seeking to solve for beta
>>>>> hat.  I am typically working with vectors that have 1. missing values,
>>>>> 2. zero values, and 3. very small values close to zero (within a
>>>>> couple of decimal places such as -.01 or .005).  I have tried mata's
>>>>> cholinv, invsym, pinv, luinv, and qrinv (I realize that some of these
>>>>> are probably inappropriate for my problem but I am no expert) and
>>>>> generally get an inverse matrix of missing values or bizarre results
>>>>> like a single populated row.
>>>>>
>>>>> I would prefer not to go back and forth between stata and mata to use
>>>>> the stata regress function unless that is the only way to accomplish
>>>>> this effort.
>>>>>
>>>>> Can anyone offer general guidance on how to proceed here?  Thanks.
>>>>>
>>>>> Tom
>>>>
>>>> *
>>>> *   For searches and help try:
>>>> *   http://www.stata.com/help.cgi?search
>>>> *   http://www.stata.com/support/statalist/faq
>>>> *   http://www.ats.ucla.edu/stat/stata/
>>>
>>>
>>>
>>> --
>>> Thomas Jacobs
>>>
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/statalist/faq
>>> *   http://www.ats.ucla.edu/stat/stata/
>>>
>>
>> --
>> Ce message a ete verifie par MailScanner
>> pour des virus ou des polluriels et rien de
>> suspect n'a ete trouve.
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>
>
>
> --
> Thomas Jacobs
>



-- 
Thomas Jacobs

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index