Journal of Reviews on Global Economics, 2015, 4, 139-146 139 E-ISSN: 1929-7092/15 © 2015 Lifescience Global Revising Fiscal Policy and Growth in Saudi Arabia Janelle Mann 1,* and Peter Sephton 2 1 Department of Economics, University of Manitoba, Canada 2 School of Business, Queen’s University, Canada Abstract: This article empirically investigates how private investment and different categories of public expenditure (defense, education, health care, and housing) impact real non-oil GDP in Saudi Arabia. The econometric analysis couples unit root, stationarity, and cointegration analysis with vector error correction models. Impulse response functions are applied to examine the impacts of different shocks to the system. We find that public expenditures on health care and defense have decreased real non-oil GDP while public expenditure on education and housing have very little impact. Interestingly, public expenditures on health crowds-out private investment. Keywords: Fiscal Policy, Cointegration, Vector Error Correction Model, Impulse Response Function. 1. INTRODUCTION One of the ways in which to measure the efficacy of a nation’s fiscal policies is to examine the sign and magnitude of the impact of public expenditure on GDP (Marattin and Salotti 2014). In this article we employ a variety of methods to investigate the impact of public expenditure on real non-oil GDP in Saudi Arabia between 1971 and 2012. We begin by reporting how aggregated public expenditures impact real non-oil GDP, after which we report that four specific categories of public spending have very different implications for the economy. Empirical evidence on the impacts of different categories of fiscal spending indicate that defense spending tends to have a negative impact on GDP (Shahbaz, Afza, and Shabbir 2013; Chang, Lee, and Chu 2013; Al-Jarrah 2005) while that on healthcare tends to have a neutral to positive impact on GDP (Ozturk and Topcu 2014; Tang 2012; Mehrara and Musai 2011; Baldacci and others 2004). Education spending has been reported to have a positive impact (Alshahranin and Alsadiq 2014; Hussin, Muhammad, Abu, and Razak 2012; Baldacci and others, 2004) while housing or infrastructure spending appears to have a neutral to positive impact (Alshahranin and Alsadiq 2014; Fedderke and others 2006; Albala- Bertrand and Mamatzakis 2001; Baffes and Shah 1998). Our empirical analysis provides evidence that public expenditure on defense has a negative impact on real non-oil GDP. Surprisingly, spending on housing and *Address correspondence to this author at the Department of Economics, University of Manitoba, 556 Fletcher Argue Building, Winnipeg, Manitoba, R3T 2N2, Canada; Tel: (204) 474-9275; Fax: (204) 474-9207; E-mail: janelle.mann@umanitoba.ca education does not appear to be directly linked to non- oil economic growth. Unlike previous studies, we report that health spending appears to be negatively related to economic activity. This latter result may be attributed to a crowding-out of private sector investment. The empirical analysis provides three key insights for Saudia Arabian fiscal policy makers: (1) there is a long- run relationship between public expenditure variables and non-oil GDP, (2) policy makers should critically evaluate their methods of allocating government expenditures to health care, and (3) trade enhancing policies have a positive impact on non-oil real GDP. 2. DATA AND ECONOMETRIC ANALYSIS We employ the natural logarithms of annual data from 1971 through 2012 for a variety of economic series: per capita real non-oil GDP (NO-PCGDP), private investment (PRIVATEI), oil revenue (OILREV), openness to trade (TRADE) proxied by the sum of imports and exports, total capital expenditure (CAPITAL), total current expenditure (CURRENT), defence expenditure (DEFENCE), education expenditure (EDUCATION), health expenditure (HEALTH), and housing and community amenities expenditure (HOUSING). All variables were collected from the Annual Statistics report from the Saudi Arabian Monetary Agency (2014). Series were converted into real terms using 1999 as the base year. The first stage of our analysis is to determine the order of integration for each series. We employed the GLS-ADF unit root test (Elliott, Rothenberg, and Stock 1996) and the efficient fractional DF unit root test (Lobato and Velasco 2007) with critical values simulated following Sephton (2009) to determine whether the series were I(1) or not. If the series are integrated of the same order we will perform a test for