The UAE UNiversity, The American University of Kurdistan

Transcription

1 MPRA Munch Personal RePEc Archve A MS-Excel Module o Transform an Inegraed Varable no Cumulave Paral Sums for Negave and Posve Componens wh and whou Deermnsc Trend Pars. Abdulnasser Haem-J and Alan Musafa The UAE UNvers, The Amercan Unvers of Kurdsan Sepember 6 Onlne a hps://mpra.ub.un-muenchen.de/7383/ MPRA Paper No. 7383, posed Sepember 6 7:3 UTC

2 A MS-Excel Module o Transform an Inegraed Varable no Cumulave Paral Sums for Negave and Posve Componens wh and whou Deermnsc Trend Pars. Abdulnasser Haem-J and Alan Musafa The UAE Unvers and he Amercan Unvers of Kurdsan E-mals: (AHaem@uaeu.ac.ae) and (Alan.Musafa@auk.edu.krd) Absrac Our am s o descrbe how a sofware componen called TDICPS can be used. TDICPS s a MS- Excel module developed n VBA (vsual bascs for applcaons) b he auhors ha ransforms an negraed varable no cumulave paral sums for posve and negave componens along wh graphs for a poenal sample sze of more han one mllon observaons. Several opons are avalable. The varable mgh have boh drf and rend, onl drf or no deermnsc rend pars. We demonsrae sep b sep how he sock prce ndex of he US marke can be ransformed no paral componens for posve and negave changes. An oher varable can also be ransformed n a smlar wa. The ransformed daa can be used for mplemenng he asmmerc causal ess as developed b Haem-J (). I can also be used for esmang he asmmerc generalzed mpulse response funcons and he asmmerc varance decomposons as nroduced b Haem-J (4). Oher opons are also possble. The MS-Excel code s avalable b e-mal from he auhors. JEL Classfcaons: G5, G, C3 Kewords: VBA, Sofware Componen, Asmmer, Causal, Impulses, The US.

3 . Inroducon I s wdel agreed b praconers ha allowng for asmmerc properes n emprcal nvesgaons accords well wh real. There s a number of reasons behnd hs concluson. For example, s well-known n he fnancal leraure ha people reac more o bad news compared o good news. Thus, he mpac of a negave change mgh be dfferen n he absolue erms han he mpac of a posve change. Anoher reason for poenal asmmerc effecs s he fac ha markes wh mperfec nformaon exs as s demonsraed b semnal work of Akerlof (97), Spencer (973) and Sglz (974). An addonal reason for asmmerc affecs s he naural lmaon on he prces. B ha means ha prces of dfferen asses or commodes can ncrease poenall whou an lmaon. However, here s a lm for he decrease of prces. In he end he prce of an valuable asse or commod canno become less han zero n he real world characerzed b raonal. Thus, a gven level of prce ncrease mus have a dfferen consequence han a prce decrease b he same amoun n he absolue erms. Oher facs from markes also confrm he exsence of asmmerc effecs. For example, b he ncrease of ol prces, fuel prces ncrease mmedael. Conversel, when ol prces decrease he fuel prces do no adjus mmedael. Based on observaon we can clearl see ha he fuel prce adjusmen for negave changes akes me and he adjusmen s usuall paral confrmng an asmmerc srucure. There are also addonal reasons for asmmerc mpacs movaed b moral or legal consrans. The economerc mehodolog has been rng o develop dfferen mehods ha can ake no accoun asmmerc effecs va hresholds, ndcaor varables or Markov regme swchng models. These mehods are avalable whn several well-known economerc packages. However, here are recen developmens n he leraure ha suggess ransformng he underlng daa no paral cumulave sums for negave and posve changes. The ransformed daa can hen be used for mplemenng asmmerc causal ess. I can also be used for esmang asmmerc mpulse response funcons. Snce hese mehods are new he are usuall no easl avalable o common praconers snce ha approprae sofware componens are mssng. For hs reason we have developed a MS-Excel module developed n VBA (vsual bascs for applcaons) b ha ransforms an negraed varable no cumulave paral sums for posve and negave

4 componens along wh graphs for a poenal sample sze of more han one mllon observaons. The goal of hs paper s o demonsrae how hs sofware componen can be used. Afer hs nroducon he res of he paper s srucured as follows. Secon represens he mahemacal formulas ha are needed for ransformng an negraed varable no paral cumulave sums. Secons 3 descrbes how our module,.e. TDICPS, can be used for hs purpose va a sep b sep demonsraon of a real me seres varable. The las secon provdes conclusons.. Mahemacal Equaons Consder an negraed varable of he frs degree, denoed b, wh drf and rend ha s generaed b he followng process: a b, () where =,, T. The varable ε s a whe nose dsurbance erm. The denoaons a and b, represen consans o be esmaed. I can be shown ha he soluon o equaon () s as he followng b usng he recursve mehod: a b, () Noe ha represens he nal value. Posve and negave changes for he underlng varable can be obaned as he followng: 3

5 4, b a (3) and. b a (4) Noe ha he posve and negave shocks are defned as follows:, max and, mn. For deals and proofs of hese resuls see Haem-J (, 4) and Haem-J and El-Khab (6). I should be menoned ha he module esmaes he parameers a and b b usng he ordnar leas squares (OLS) mehod. Tha s, he followng soluons are ulzed for esmang he underlng parameers: b a (5) and T T b (6) Where T T and T T.

6 3. The TDICPS To demonsrae how he TDICPS module works we make use of he sock prce ndex for he US marke as an example. Ths varable covers he perod of on monhl bass. The source s he Federal Reserve Economc Daa (FRED) daabase ha s gven onlne b he Federal Reserve Bank of S. Louse. The frs sep s o sar TDICPS n MS-Excel. Then, he followng wndow opens: Before usng he module we need o clck on Oka boom n order o confrm coprgh of he auhors. Then, ou can cop and pase our varable n column B. We donae hs column b. 5

7 However f ou prefer an oher le ou can change. The poenal sample sze can be more han one mllon observaons. The nex sep s o choose one of he opons ha s gven on he lef sde (marked b he ellow color). Tha s o choose from followng opons. If he frs opon s seleced hen here s no drf and rend (.e. a= and b=). If he second opon s seleced hen here s a drf bu no rend (.e. a and b=). If he hrd opon s chosen hen here s a drf and rend (.e. a and b ). B clckng on he chosen opon he module provdes he ransformed daa for ha opon. I s also possble o oban he resuls for all hree specfcaons f he Oupu For All Three Opons s seleced. In order o demonsrae an example, we selec opon hree and he module provdes he followng resuls: 6

8 Column A s he me ndex and column B s he orgnal daa. Column C s he ransformed daa for posve changes. Column D s he ransformed daa for negave changes. The module auomacall provdes graphs for columns B, C and D. For example n hs case we oban he followng graphs. 7

9 Seres (sample daa)+.5 (sample daa)

10 .6 (sample daa) (sample daa)- Noe ha represens orgnal daa for he US sock prce ndex. Paral cumulave sum for posve changes s denoed b +. Paral cumulave sum for negave changes s sgnfed b -. I should be menoned ha he ransformed daa can be used for mplemenng asmmerc causal ess as developed b Haem-J (). I can also be used for esmang he asmmerc generalzed mpulse response funcons as suggesed b Haem-J (4). Ths s sragh forward because afer ransformng he daa can be mpored o an consumer frendl economerc package such as EVews, Ras, Saa, PcGve or Mcrof among ohers. 9

11 4. Conclusons I s wdel agreed n he leraure ha s mporan o allow for poenal asmmerc mpacs when emprcal suded are conduced. Snce here s lack of avalabl for consumer frendl sofware componens for mplanng new developmens n he economerc mehodolog ha allows for asmmerc mpacs, we have creaed a module wren n VBA language. Alhough, s a common pracce o develop a proope of an applcaon usng a hgh level language and hen usng alernave envronmen for applcaon developmen o work on mprovng he performance of he execuon of ha algorhm. Ths sofware componen s eas o use and s accessble va MS-Excel. We have demonsraed here how he sock prce ndex of he US marke can be ransformed no posve and negave paral cumulave componens va hs module. Afer ransformng he underlng daa, he praconer can easl mplemen asmmerc causal ess va an auomac sofware ha provdes an opon for causal esng. I can also be used for mplemenng generalzed mpulse response funcons. For echncal descrpon of hese mehods see Haem-J (, 4) and Haem-J and El-Khab (6). I s also possble use he ransformed daa o es for hdden conegraon as suggesed b Granger and Yoon () or o es for hdden panel conegraon as suggesed b Haem-J (a). Oher possbles are esng for asmmerc panel causal (see Haem-J, b) or for mplemenng he asmmerc GARCH model as recommend b Haem-J (3). The TDICPS sofware componen s avalable from he auhors va e-mal. I s ver consumer frendl. Acknowledgemens The curren work has parall been funded b a research gran gven b he UAE Unvers (Gran # 3B8), whch s greal apprecaed. There s also anoher module ha s wren n Ocave b Haem-J and Musafa (6) whch can be used o ransform he daa. However, he Ocave module has onl one opon unlke he MS-Excel one ha we provde here, whch has hree opons.

12 References Akerlof G., (97) The marke for lemons: Qual unceran and he marke mechansm, Quarerl Journal of Economcs, 84, Granger, C.W., Yoon, G. () Hdden Conegraon. Deparmen of Economcs Workng Paper. Unvers of Calforna. San Dego. Haem-J A (a) Hdden panel conegraon, MPRA Paper 364, Unvers Lbrar of Munch, German. Haem-J A (b) Asmmerc Panel Causal Tess wh an Applcaon o he Impac of Fscal Polc on Economc Performance n Scandnava, MPRA Paper 5557, Unvers Lbrar of Munch, German. Haem-J A () Asmmerc Causal Tess wh an Applcaon, Emprcal Economcs, 43(), Haem-J A (3) A New Asmmerc GARCH Model: Tesng, Esmaon and Applcaon, MPRA Paper 457, Unvers Lbrar of Munch, German. Haem-J A (4) Asmmerc generalzed mpulse responses wh an applcaon n fnance, Economc Modellng, 36, 8-. Haem-J A. and El-Khab Y. (6) An exenson of he asmmerc causal ess for dealng wh deermnsc rend componens, Appled Economcs, 48, Haem-J A. and Musafa A. (6) TDICPS: OCTAVE Module o Transform an Inegraed Varable no Cumulave Paral Sums for Negave and Posve Componens wh Deermnsc Trend Pars, Sascal Sofware Componens, Nr. OCT, Boson College Deparmen of Economcs. Avalable on lne. (hps://deas.repec.org/c/boc/bocode/oc.hml). Spence M. (973) Job marke sgnallng, Quarerl Journal of Economcs, 87, Sglz J. (974) Incenves and Rsk Sharng n Sharecroppng, Revew of Economc Sudes, 4, 9-55.