Identifying government spending shocks in the U.K: A novel instrumental approach based on natural disasters.

Identifying government spending shocks in the U.K:
A novel instrumental approach based on natural disasters.
Following the global financial crisis of 2007, many countries resorted to significant fiscal stimulus
packages, consisting of increased government spending and tax cuts, to boost economic activity.
Most of these packages were based on the Keynesian view that expansionary fiscal policy can
mitigate the economic downfall and prevent the waste of economic resources. In line with this
rationale, in 2008, the United Kingdom was one of the major economies to lead calls for fiscal
expansion to stimulate aggregate demand and help offset the global economic downturn. Adopting
such measures generated a renewed interest among academics and policymakers in whether and to
what extent fiscal policies are effective. This interest was recently further stimulated by the massive
fiscal response that followed the Covid-19 pandemic.
Despite its importance, there is still no consensus over the effectiveness of fiscal policy in stabilising
the economy (Cloyne, 2013; Ramey, 2011; Mertens and Ravn, 2012; Leeper et al., 2013). A prime
reason for this lack of consensus lies in the difficulty of identifying primitive, unanticipated policy
shocks (i.e. changes in fiscal policy not correlated with other macroeconomic fluctuations) which are
required for conducting a causal analysis. The existing literature has proposed various methods for
identifying fiscal shocks to address this. These involve time and sign restrictions in a VAR system and
instruments (Ramey, 2016). Unfortunately, each of these techniques exhibits weaknesses that raise
concerns about the findings of empirical studies. For instance, the short-run restriction approach
suffers from anticipation issues. Specifically, because government expenditure or tax cuts are
usually announced several quarters before they take place, it fails to account for changes in
expectations which leads to underestimation of the effect of fiscal policy (see Mertens and Ravn, 2010). Similarly, sign restrictions that compute impulse responses based on a large set of plausible
models make the structural analysis less informative (Braun and Bruggemann, 2022).
A method which has been gaining popularity in recent years is the identification of shocks
using instruments. As shown by Ramey (2016), exploiting variation from proper instruments
deals with the problem of fiscal foresight (anticipation) and provides robust results. The main
challenge of identification using instruments is the selection of the instrument itself. The
selected instrument must be correlated with the shock of interest and orthogonal to the rest
of the shocks. These conditions are untestable, and as a consequence, applications of shock
identification using instruments are often criticised either for lack of strict exogeneity1 or for
low relevance. Regarding government spending shocks, most of the existing literature exploits
variation in military spending to construct instruments for identification. The underlying
assumption is that military expenditures constitute government spending unrelated to the
state of the economy (Ramey and Shapiro, 1998; Ramey, 2011; Ben Zeev and Pappa, 2017).
However, Ramey (2016) argues that proxies of government spending shocks constructed
from historical records on military spending are weak instruments for the post-Korean war
samples.