Niğde Üniversitesinin Yerel Ekonomiye Katkısı: Enflasyon Etkisi Altında Bir Hesaplama

Can MAVRUK, Ali İlker TEKİNARSLAN, Murat GÜRÜN, Ayşe AYİŞ AKKURT
5.344 1.385

Öz


Niğde Üniversitesinin yerel ekonomik ve istihdam katkısı 2012-2013 mali yılı için Keynesyen Çarpım Yöntemi, Ryan Kısa Yol Yöntemi ve Anket Yöntemi kullanılarak hesaplanmıştır. Bu çalışmada üniversitenin bütçe harcamaları verileri, ulusal veri kaynakları, anket verileri kullanılmıştır. Keynesyen gelir çarpanı 1.22 hesaplanmıştır. Ekonomik katkının hesaplanmasında önemli bir etken olan marjinal tüketim eğilimi E-Views programı kullanılarak kısa dönem modelinde 0,70 hesaplanmıştır. Bu makale üniversitenin toplam yerel gelir ve istihdam kapasitesini hesaplayarak üniversitenin bulunduğu yer için ekonomik önemini gösterir ve üniversitenin yerel ekonomi için itici bir güç olduğu sonucuna varır. Bu makale yerel ekonomik katkıyı daha iyi hesaplamak için gerçek enflasyon etkisini içeren bir hesaplama yöntemi sağlar. Bu yöntem diğer etkilere ek olarak enflasyon etkisi altında gelir çarpanının küçük olduğunu gösterir. Üniversitenin toplam yerel katkısı sırasıyla gayrisafi yerel gelirlerin toplamı, doğrudan ve dolaylı katkıların toplamının gelir çarpanı ile çarpımı bağlı yerel harcamaların istihdam çarpanı ile çarpımı sonucu hesaplanmış ve sonuçlar karşılaştırılmıştır.

Anahtar kelimeler


Ekonomik katkı; üniversite harcamaları; enflasyon etkisi

Tam metin:

PDF

Referanslar


CEYHAN, M.Said and GÜNEY, Gül (2011), “Bartın Üniversitesi’nin Bartın İli’nin Ekonomik Gelişimine 20 Yıllık Projeksiyonda Katkılarının Değerlendirilmesi”, Fırat Üniversitesi Sosyal Bilimler Dergisi 21(2): 183-207

ERKEKOĞLU, H. (2000), “Bölge Üniversitelerinin Yerel Ekonomiye Katkıları: Sivas Cumhuriyet Üniversitesi Örneği”, Erciyes Üniversitesi İ.İ.B.F. Dergisi, 16, 203– 2

GARRIDO-YSERTE, Ruben and GALLO-RIVERA, Maria Teresa (2007), “The Impact of The University Upon Local Economy: Three Methods To Estimate Demand-Side Effects”, University of Alcala, Madrid, Spain

GÖRKEMLİ, HN (2009), “Selçuk Üniversitesi’nin Konya kent ekonomisine etkileri”. Selçuk Üniversitesi Sosyal Bilimler Enstitüsü Dergisi 22: 169-186

JIANN, Huah Cheng (2005), “Seasonal Adjustment of Time Series”, Economic Accounts Division, Singapore Department of Statistics, Statistics Singapore Newsletter: 1114

JOHANSEN, Tom and ARANO, Kathleen (2011), “The Economic Impact of Fort Hays State University on the State of Kansas: Fiscal Year 2010”.

LANTZ, Van A., BRANDER, John R.G., YIGEZU,Yigezu A. (2002), “The Economic Impact of the University of New Brunswick: Estimations and comparisons with other Canadian Universities”.

OHME, Allison M (2003), “The Economic Impact of a University on its Community and State:Examining Trends Four Years Later”, Institutional Research, University of Delaware

ÖZBAY, Özden (2013), “Üniversite ve Sosyal Çevresi: Niğde Üniversitesi ve Niğde”, Ankara Üniversitesi Sosyal Bilimler Enstitüsü Dergisi 4(1), DOI:10.1501

ŞEN, Arzu (2011), “Local Income and Employment Impact of Universities: The Case of Izmir University of Economics”, Journal of Applied Economics and Business Research 1: 25-42

SIEGFRIED, John J and SANDERSON, Allan R. and MCHENRY, Peter (2006), “The Economic Impact of Colleges and Universities”, Working Paper No.06-W12, Vanderbilt University

SÜRMELİ, Fevzi (2008), “Anadolu Üniversitesinin Eskişehir’e Etkileri ve Şehrin Üniversiteyi Algılayışı”, Anadolu Üniversitesi, Eskişehir, 336s

TAVOLETTI, Ernesto (2007), “The Local and Regional Economic Role of The Universities: The case of the University of Cardiff”, WP n.9 DiSSE, Universit`a di Macerata

TORUN, İ. (2005), “Yöre Halkının Üniversiteye Bakışı ve Beklentileri: Niğde Üniversitesi Örneği”. Sakarya Üniversitesi Eğitim Fakültesi Dergisi 9, 156-177.

TREVINO, Reggina (2008), “PreMBA Analytical Primer”, 1st Ed., Palgrave McMillan

TUĞÇU, C.T. (2004),” Üniversitelerin Yerel Ekonomik Faaliyet Hacmine Katkıları: Nevşehir Örneği”

TUTAR, E. (2005), “Üniversitelerin Yerel Ekonomiye Katkısı: Niğde Örneği”. Ankara: Detay. APPENDIX A Linear regression model is X t = b + cY t where X t = EX t = Final Consumption Expenditure of Resident Households, Y t = GDP t = Real Gross Domestic Product, c = Marginal Propensity to Consume, b = Autonom Consumption Expenditure. Seasonally adjusted data in regression model and short run model both was taken quarterly in number of periods between 2002Q4 and 2012Q3. Because of seasonal variations in GDP and EX it might not be easy to detect a trend. Census X12 ARIMA in E-Views was used to deseasonalize (seasonally adjust) GDP and Final Consumption Expenditure of Resident Households (EX). To validate the automodel choice by X12 ARIMA, Ljung-Box(LB) statistics was conducted on the residuals. There was no significant autocorrelation among the residuals. Gross Local Output (GDP) and Final Consumption Expenditure of Resident Households (EX) both increases as time increases. Time series EX and GDP do not have a constant mean and variance. Therefore, EX and GDP are not stationary, but they have a constant and trend. Before ADF test is applied to make them stationary, it is important to read the graphs for the assumptions. The same can be discussed for EXSA and GDPSA. In our study to test the hypothesis whether EXSA and GDPSA have a unit root or not, ADF test is applied to both EXSA and GDPSA series at level. Results given on Table A1 shows at level ADF test values are greater than McKinnon critical values. Therefore, H hypothesis could not be rejected which implies both EXSA and GDPSA series were not stationary at level. To make EXSA and GDPSA stationary first difference was taken for each and ADF test values were found to be less than McKinnon critical values. Therefore, the first differenced series both DEXSA and DGDPSA became stationary.

Table A1. ADF test t-statistics, ( ):McKinnon 5% critical values Series Level 1st Difference EXSA 254127 (-2.910860) -760135 (-2.912631) GDPSA 719017 (-2.912631) -179232 (-2.912631) So EXSA~ I(1) and GDPSA ~ I(1). This means that variables EXSA and GDPSA both are integrated of order 1. Since they were integrated of the same order, we could run

Johansen Cointegration test with lag 1 to check for the number of cointegrating equations using Trace Statistic and Maximum Eigenvalue Statistic. Table A2 shows that both statistics indicated 1 cointegrating equation at the 5% level. The two variables EXSA and GDPSA were cointegrated and this suggested that they had a long run relationship. Table A2. Johansen Cointegration Test Lag 1 Results at 5% Hypothesis Statistic 5% critical values # of CE* equations p-value H : No CE* rejected H : At Most 1 CE* not rejected Trace 94809 87760 49471 84147 1 0.0031 0.1706 H : No CE* rejected

H : At Most 1 CE* not rejected Max-Eigen 07049 877601 26460 84147 1 0.0036 0.1706 *CE means Cointegration Equation The linear regression model was used to generate the residual series and then residuals were tested for stationarity using ADF test and residuals were stationary at level. Results are given on Table A3 below. Table A3. ADF Residual Test Results at 5% None Hypothesis t-statistic McKinnon Critical Value Level H : Residual is not stationary (rejected) -815368 -1.946654 Since residuals were stationary we could conclude that the variables in regression model had long run relationship and they were cointegrated. Regression model was estimated by OLS method and corrected to remove serial correlation (autocorrelation).

Regression Model: X t = 1646 + 0.70Y t + 0.36AR(1) X t = EXSA t and Y t =GDPSA t Constant coefficient was not significant. Long run coefficient of GDPSA, marginal propensity to consume, was highly significant. ADF results of the regression model are given below. Residuals of the regression model were not serially correlated (had no autocorrelation) by Breusch-Godfrey Serial Correlation LM Test, monoskedastic by Breusch-Pagan-Godfrey Heteroskedasticity Test and normally distributed by Jarque-Bera probability test. All implies that the regression model X t = 1646 + 0.70Y t + 0.36AR(1) is a good model. Table A4. ADF Results of the Regression Model Variable Coefficient p-value C 1642 0.3744 R 2 0.998221 GDPSA 0.702092 0.0000 Durbin-Watson 986084 AR(1) 0.361712 0.0248 Akaike 38016 F-statistic 0.0000 SHORT RUN MODEL

Correlogram and LB statistics were also used to check for autocorrelation. DEXSA and DGDPSA had no autocorrelation. Null Hypothesis H : DGDPSA is stationary was accepted. Since the two variables were cointegrated, using E-Views we could run Error Correction Model (ECM) to estimate marginal propensity to consume (c). ECM model: ∆X t = b + c∆Y t +d*u t-1 From Table 5 below, ECM short run model is ∆X t = 335 + 0.70∆Y t – 0.64*u t-1 Since the coefficient of error correction term u t-1 was negative (-1< u t-1 < 0) and significant, it validated long run equilibrium relationship between X t and Y t in our linear regression model. In the ECM model constant coefficient was not significant, but short run coefficient of ∆Y t which is marginal propensity to consume was highly significant. Marginal propensity to consume was estimated 0.702157 and approximated to be 0.70 in our study. Table A5. Short Run Model Variable Coefficient p-value C 32745 0.6037 R 2 0.655178 DGDPSA 0.702157 0.0000 Durbin-Watson 951602 U(-1) -0.638101 0.0006 Akaike 42431 F-statistic 0.0000