10.12.2012 Views

Challenges in the Era of Globalization - iaabd

Challenges in the Era of Globalization - iaabd

Challenges in the Era of Globalization - iaabd

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Challenges</strong> <strong>in</strong> <strong>the</strong> <strong>Era</strong> <strong>of</strong> <strong>Globalization</strong><br />

Edited by Emmanuel Obuah<br />

The application <strong>of</strong> Granger causality test to panel data is a recent occurrence. Panel data analysis provides<br />

more observations and thus <strong>in</strong>creases degree <strong>of</strong> freedom associated with econometric analysis. Accord<strong>in</strong>g<br />

to Hsiao (2001, 2003), Hsiao et al (2002) and Anderson and Hsiao (1982) panel data analysis gives a<br />

more accurate prediction <strong>of</strong> outcomes than time series analysis.<br />

With i = 1, 2, ….N; t= K+1, K+2, …, T and y, x are <strong>the</strong> two variables be<strong>in</strong>g tested.<br />

The error term, εit is assumed to satisfy <strong>the</strong> orthogonality conditions:<br />

In Equation (5), bi captures country-specific effects <strong>the</strong>reby allow<strong>in</strong>g <strong>the</strong> relationship between x and y to<br />

vary across <strong>the</strong> panel <strong>of</strong> data. However, <strong>the</strong> approach <strong>of</strong> Holtz-Eak<strong>in</strong> et al. (1988) is adopted by allow<strong>in</strong>g<br />

<strong>the</strong> lag coefficients <strong>of</strong> δkt and βkt to be constant for each t. In order to get rid <strong>of</strong> <strong>the</strong> <strong>in</strong>dividual effects<br />

represented by bi, Holtz-Eak<strong>in</strong> et al. employ a differenc<strong>in</strong>g transformation <strong>of</strong> (5). Moreover, based on<br />

Anderson and Hsiao (1982), Chowdhury (2001) and Hood et al. (2006) an <strong>in</strong>strumental variable<br />

framework is employed to deal with possible correlation between <strong>the</strong> error term <strong>of</strong> <strong>the</strong> transformed model<br />

and <strong>the</strong> right-hand-side explanatory variables. Granger causality from x to y is tested by a jo<strong>in</strong>t hypo<strong>the</strong>sis<br />

that all βkt are zero.<br />

Empirical Results<br />

When a Granger causality test is performed, four possible outcomes are possible:<br />

i. Per capita health expenditure Granger causes per capita GDP<br />

ii. Per capita GDP Granger causes per capita health expenditure<br />

iii. Per capita GDP Granger causes per capita health expenditure and vice versa<br />

iv. No Granger causality exits<br />

In this paper, an attempt is made to explore <strong>the</strong> causal relationship between health expenditures and<br />

output <strong>in</strong> fourteen <strong>of</strong> ECOWAS countries us<strong>in</strong>g panel data from 1998 through 2006. Table 1 show <strong>the</strong><br />

results <strong>of</strong> unit root tests which are applied to <strong>the</strong> data. The results show that both GDP and per capita<br />

health expenditures (HEC) are nonstationary <strong>in</strong> levels. The application <strong>of</strong> unit root tests to <strong>the</strong> first<br />

difference <strong>of</strong> <strong>the</strong> series reveals that <strong>the</strong> series are stationary at <strong>the</strong> conventional 5 percent level. The results<br />

for <strong>the</strong> Lev<strong>in</strong>, L<strong>in</strong> & Chu; Im, Pesaran &Sh<strong>in</strong>; ADF-Fisher and PP-Fisher tests are all consistent.<br />

Table 1: Panel Unit Root Test Results (With Individual Effects)<br />

Test Statistic GDP HEC ∆GDP ∆HEC<br />

Lev<strong>in</strong>, L<strong>in</strong> & Chu statistic<br />

Im, Pesaran & Sh<strong>in</strong> statistic<br />

ADF-Fisher chi-square statistic<br />

PP-Fisher chi-square statistic<br />

3.328<br />

5.070<br />

11.864<br />

15.642<br />

0.342<br />

2.452<br />

20.226<br />

17.583<br />

-11.574*<br />

-8.398*<br />

111.960*<br />

147.594*<br />

(5)<br />

(6)<br />

-11.052*<br />

-7.619*<br />

103.352*<br />

126.264*<br />

389

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!