1
University of Amsterdam
Msc in Economics
Monetary policy and theory
Master’s thesis under the supervision of Prof. Dr. Christian
Stoltenberg
Yilang Ou
(11388560)
Economic impacts and channel
research on Chinese economic
policy uncertainty
December 2017
2
ABSTRACT:
Based on national macroeconomic data for 1995 to 2014 this paper applies the SVAR model to examine dynamic effects of economic policy uncertainty on
macroeconomics, and it discusses relevant transmission channels of such effects. Our results show that economic policy uncertainty has negative effects on increments of national output, consumption, investment and export. Specifically, the paper shows that policy uncertainty restrains consumption growth by magnifying the negative effect of consumption volatility and by increasing the risk aversion of consumers. Economic policy uncertainty diminishes investments by decreasing expected returns, aggravating the negative effect of macroeconomic volatility. Economic policy
uncertainty reduces inflation rates by limiting adaptive expectations of inflation and by dampening the output gap’s inflation effect. Policy makers can refer to this paper in evaluating and controlling policy uncertainty levels to abate their negative effects.
3
Contents
1INTRODUCTION ... 4
2METHODOLOGY ... 8
2.1THEMEASUREMENTOFECONOMICPOLICYUNCERTAINTY ... 8
2.2SVARMODEL ... 14
3DATA ... 18
4RESULTS ... 21
5ANALYSIS OF THE MACROECONOMIC INFLUENCING CHANNEL OF POLICY UNCERTAINTY ... 25
5.1CHANNELSTHROUGHWHICHPOLICYUNCERTAINTYAFFECTSCONSUMPTION ... 25
5.2CHANNELSTHROUGHWHICHPOLICYUNCERTAINTYAFFECTSINVESTMENT ... 26
5.3CHANNELSTHROUGHWHICHPOLICYUNCERTAINTYAFFECTSINFLATION ... 28
6CONCLUSIONS ... 31
4
1 Introduction
Uncertainty is an important factor that influences the economy, as it not only
influences consumers’ and investors’ economic decisions (Leland (1968), Lucas and Prescott (1971)) but also a country’s economic growth (Aizenman (1993)). Economic policy uncertainty, the variable I discuss in this thesis, also affects a country’s
economic growth as one form of uncertainty. Rodrik (1991) proves that an increase of economic policy uncertainty will decrease the reformation dividend of a developing country as well as personal investments. Aizenman (1993) regresses panel data for 46 developing countries and find a negative correlation between economic policy
uncertainty and output growth. Fatás A and Mihov I (2013) monitor the government spending of 93 different countries and find that economic policies are invalid when a government changes economic policies too frequently. N Berggren (2009) uses the International Country Risk Guide risk index to measure economic policy uncertainty levels of 132 countries and show that economic policy uncertainty negatively affects economic growth. Julio and Yook (2012) test the relationship between firm
investment and the frequency of political elections and find that frequent political elections spur an increase in economic policy uncertainty and decrease firm investments. Similarly, Yekun Xu (2013) and Haisheng Tang (2014) show that frequent political elections in China decrease firm investments. Despite the negative effects of high levels of economic policy uncertainty, Born and Pfeifer (2014) show that high levels of economic policy uncertainty do not cause serious damage when an economy is functioning well. In regards to economic depression, most economists agree that a high level of economic policy uncertainty slows the pace of economic revival. For China we also find effects of high levels of economic policy uncertainty. For example, during the 1997 Asian Financial Crisis, Subprime Mortgage Crisis and 2013 Chinese Banking Liquidity Crisis, both the government and the central bank behaved erratically. This rendered it difficult for individuals to correctly judge economic policy changes and led to fluctuations in the stock market and in the economy over the short term. In particular, as Chinese economic growth is currently slowing down, economic policy uncertainty may have more serious effects when government’s economic policy stimuli are unclear. It is thus necessary for us to examine effects and mechanisms of economic policy uncertainty to offer relevant guidance to economic policymakers.
Uncertainty refers to unknown future outcomes (Frank Knight (1921)) that cannot be determined based on probability measurements. Economic policy uncertainty, which
5 is one aspect of uncertainty, denotes the unknown state in which individuals face their expectations of future economic policies. For example, individuals expect central banks to adopt stimulus policies during a stock market crash. However, when a central bank will launch its stimulus policy and how significant of an adjustment will made through this policy cannot be determined through probability measures.
Economic policy uncertainty can be divided into two forms. Some economists suggest that economic policy uncertainty can be expressed by individuals’ sensations, which are individuals’ subjective and uncertain feelings. We thus structure uncertainty indexes based on surveys of public uncertainty. A number of uncertainty indexes apply this view, such as Baker et al’s (2013) Economic Policy Uncertainty index and the Federal Reserve Bank’s Economic Policy Uncertainty Index. Other economists suggest that economic policy uncertainty can be measured from economic data related to economic policies such as those focusing on benchmark interest rates and fiscal deficit levels (Johannsen (2014), Ulrich (2012), Mumtaz and Zanetti (2013), Fatas and Mihov (2013)). Economists estimate a potential policy rule (e.g., the Taylor rule) according to past policy data and use the estimated policy rule as a basic rule for measuring the degree of bias between real and estimated policy-related data. We then take this bias as the degree of policy uncertainty. As these two forms of policy
uncertainty involved the application of different measuring methods and data, they are classified as subjective and objective forms of policy uncertainty. Forms of objective policy uncertainty can be divided into fiscal policy uncertainty and monetary policy uncertainty according to different sources involved. In regards to fiscal policies Fernández Villaverde et al (2011) apply fluctuations in fiscal policies to the general equilibrium Keynesian Three Sector Model and find that fluctuations in fiscal policies have negative effects on both commercial activities and inflation. Johannsen (2014) shows that American fiscal policy uncertainty has more negative on the macro
economy when the monetary policy is relatively rigid. Shoudong Chen and Dongliang Yang (2009) find out that a determinate endogenous impulse negatively affects
long-term individual consumption growth rates and that a stochastic exogenous impulse positively affects long-term individual consumption growth rates. Regarding monetary policies, Ulrich M (2012) found that the impulse from policy uncertainty exacerbates fluctuations in the stock return rate through his study on the relationship between monetary policy and the capital asset price. G Bekaert (2011) uses the SVAR model to prove that monetary policy uncertainty has significant effects on individual risk-averse levels and industrial production. Mumtaz H (2013) demonstrates that an increase in monetary policy fluctuations spurs inflation while lowering industrial outputs based on the Dynamic Stochastic General Equilibrium (DSGE) model. For the
6 measurement of policy uncertainty, economists normally use public survey data to measure subjective economic policy uncertainty levels. Gentzkow and Shapiro (2006) examine policies by statistically analysing modern media content and find that
American newspaper publications can reflect policy changes in America and that such newspapers serve as a good medium for measuring policy changes. In 2013, scholars (Baker, Bloom & Davis (2013)) from Stanford University and Chicago University cooperatively developed the Economic Policy Uncertainty Index (EPU) to
comprehensively estimate economic policy uncertainty levels of different countries. This index has already been used as a proxy variable for policy uncertainty by several economists (Johannsen (2014), Aastveit (2013)). In measuring objective policy uncertainty, economists normally use economic policy uncertainty time-series data derived through model structuring and from variable estimations. For example, conditional variance has been estimated using the GARCH model and from deformation models (Fountas et al (2007), Ulrich (2012)). Fernandez-Villaverde (2012) uses time-varying parameters of the state space model to measure monetary policy uncertainty levels of a small open economy. Fatas (2005) and Bekaert (2011) argue that benchmark interest rates and government spending are the most widely used policy variables and the authors use residual terms derived by regressing policy and economic variables to represent policy uncertainty. In measuring Chinese economic policy uncertainty, Shoudong Chen and Dongliang Yang (2009) suggest that one can use the state space model to capture uncertainty in fiscal spending growth rates and argue that this approach can remedy problems with this empirical model resulting from economic structural breaks.
In relation to existing studies, this thesis makes three novel contributions. First, we discuss both subjective and objective economic policy uncertainty to avoid biases resulting from using different measuring methods for the same analysis. Moreover, while other studies on objective economic policy uncertainty only focus on monetary or fiscal issues, we consider both in this thesis. Second, through our measurement of uncertainty levels we find that uncertainty predictions should be prospective,
unforeseeable and rational. We also find that common econometric methods such as linear regression and GARCH methods do not satisfy such requirements. We thus apply the state space measurement model, as it renders our predictions more reflective of real conditions. Third, studies on transmission channels of economic policy
uncertainty effects have not been explored at length in reference to China, and we test effects of transmission channels in regards to consumption, investment and inflation.
7 Based on national macroeconomic data for 1995 to 2014, this paper applies the SVAR model to examine economic policy uncertainty’s dynamic effects on macroeconomic processes and it discusses related transmission channels. Our results show that economic policy uncertainty has negative impulse effects on increments of national output, consumption, investment and export. Specifically, the paper shows that policy uncertainty restrains consumption growth by magnifying negative effects of
consumption volatility while increasing consumer risk aversion. Economic policy uncertainty diminishes investments by decreasing the expected return, in turn aggravating negative effects of macroeconomic volatility. Economic policy
uncertainty reduces inflation rates by keeping adaptive expectations of inflation low and by dampening the output gap’s inflation effects. Policymakers can refer to this paper to evaluate and control uncertainty levels and to in turn abate their negative effects.
This paper is organized as follows. In the following section we analyse time-series data on uncertainty in Chinese economic policies and outline the SVAR model. The third section describes our variables and data in greater detail and presents the
uncertainty index of economic policies used in this paper. The fourth section analyses impulse responses of the SVAR model to explore economic impacts of economic policy uncertainty on the economy. The fifth section further discusses impacts of economic policy uncertainty channels. The sixth section summarizes the paper.
8
2 Methodology
This section is divided into three parts. I first discuss the measurement of economic policy uncertainty and why I use the measurements outlined. Second, I describe the SVAR model. Third, I identify the SVAR model.
2.1 The measurement of economic policy uncertainty
Economists have developed several methods for measuring economic policy
uncertainty, but some of these methods may not always be suitable. When measuring economic policy uncertainty it is necessary to respect features of economic policy uncertainty itself. We put forward three features of expectations regarding economic policy uncertainty:
i. Prospective.
Individuals can only make judgements on the future based on past information. Normally, when a predictor possesses information on the last t periods, we have E(𝜇) = E(𝜇|𝐷𝑡−1, 𝐷𝑡−2, ⋯ , 𝐷𝑡−𝑘) where 𝐷𝑡 refers to new information that the
predictor receives at moment t. I do not use tools such as the ordinary least squares technique or maximum likelihood estimation because these methods are retrospective. It is necessary to use future information to estimate past information, and thus we have E(𝜇) = E(𝜇|𝐷𝑇) where 𝐷𝑇 denotes overall information. Thus, tools such as
the ordinary least squares technique and maximum likelihood estimation do not meet our requirements.
ii. Unforeseeable
Only an unforeseeable change can be deemed uncertain. Knight (1921) was the first to describe the difference between risk and uncertainty. Risk can be defined by a
probability distribution while uncertainty cannot. Thus, it is not suitable to use conditional variances of GARCH to represent uncertainty (Fountas et al (2007), Ulrich (2012)). Sharpe’s (1964) capital asset pricing model assumes that investors make the only correct decision available when they understand risks related to a capital asset. Similarly, according to the consumption-based capital asset pricing model, consumers make certain consumption decisions based on a certain level of risk. However, uncertainty differs from risk and individuals cannot estimate the
9 “uncertainty premium”. Therefore, a decision made under uncertain conditions differs from uncertainty in the presence of risk.
iii. Rational
According to School of Rational Expectations theory, individuals have rational expectations based on their past experiences and constantly adjust their expectations to avoid being subject to systematic bias. Lucas (1976) note that policy changes alter an individual’s expectations. It is thus necessary to use predictive econometric models to reflect such alterations. Otherwise, misleading information will be generated. As individuals obtain future prior knowledge based on past information, we can denote a rational expectation as the prior distribution from Bayes combined with i. When the prior distribution is p(𝜇𝑡|𝐷𝑡−1, 𝐷𝑡−2, ⋯ ), 𝐷𝑡 is new information obtained at moment t,
we can have a posterior distribution written as p(𝜇𝑡|𝐷𝑡) ∝ p(𝐷𝑡|𝜇𝑡)p(𝜇𝑡|𝐷𝑡−1, 𝐷𝑡−2, ⋯ ).
The prior distribution is renewed after new information 𝐷𝑡 is received. Then, rational
expectation adjustment is achieved.
We divide economic policy uncertainty into subjective objective policy uncertainty according to its expression, and I offer a more detailed description and evaluation below:
Subjective policy uncertainty is defined as an economic subject’s views on policy uncertainty. Based on this definition, economists can quantify the public’s views on policy uncertainty to obtain the policy uncertainty index. Tools applying such an approach include Baker et al’s (2013) EPU index, Google’s index and the Philadelphia Fed’s economy-forecasting survey. This form of measurement is advantageous in that we can bypass economic policy uncertainty’s influencing
channels to directly observe its effects, which are individuals’ psychological reactions. We also believe that this type of psychological reaction is both prospective and
rational. However, this approach may not be unforeseeable, as individuals may define uncertainty differently. For instance, some individuals believe that an increase in the benchmark interest rate is to be expected while others think feel that this is not to be expected. As subjective policy uncertainty can only reveal the terminal of
uncertainty’s effects and not its origins and as we do not know what causes
uncertainty from subjective policy uncertainty, objective policy uncertainty must be considered to help policymakers examine their policy behaviours.
Objective policy uncertainty is defined as the bias between policymakers’ policy decisions and individuals’ expectations on policy decisions. It is widely understood that when the majority of a policy is carried out according to the expectations of
10 individuals, uncertainties related to the objective policy cannot be determined.
However, when the foundations of a policy do not meet the expectations of individuals, objective policy uncertainty will appear. Based on this definition,
scholars can distinguish uncertainty from the objective indicators of relevant policies. Of course, scholars adopt different measurements and measurement models, and thus policy uncertainty indicators obtained vary. This measurement method is
advantageous in that it defines the "source" of policy uncertainty - uncertainties related to a policymaker's own policy. It can help policy makers understand uncertainties related to their own policies. Of course, its disadvantages are also obvious. It measures an actual policy’s deviation from the expected policy on the basis of a certain forecasting rule, which can only be roughly in line with social reaction rules and which cannot be accurate. Therefore, it is necessary to use
uncertainty in subjective policies as a supplement to uncertainty in objective policies so that the measurement of policy uncertainty can most closely reflect real social reaction rules.
Economic policies are divided into fiscal and monetary policies by sector. Therefore, we attempt to measure uncertainty in fiscal and monetary policies and to provide reference indexes for a government and central bank. To limit biases resulting from the uncertainty measurement of objective policies, measurement methods adopted should be designed to meet the above three requirements. The estimation of the state space model based on time-varying parameters can meet such requirements, as the state space model is estimated via the Kalman filtering method, and only past
information is used to estimate time-varying parameters. It is a prospective estimation based on one-step-ahead forecasting. The estimated residual value represents
unpredictable policy changes as unforeseen uncertainties; the model adjusts a
time-varying parameter based on historical information after each phase of estimation, and it can imitate rational expectations that an economic subject adjusts at any time. Therefore, we use the state space model of time-varying parameters to estimate fiscal and monetary policy uncertainty.
The notion of using residuals of the measurement equation in a state space model to capture policy uncertainty is similar to the notion of adaptive learning, according to which the beliefs of agents are adjusted for every period according to the scale of the difference between real policy outcomes and expected outcomes, i.e., the residual is the key variable used to correct agents’ beliefs on economics, and the higher the residual is, the more agents will deviate from their original decision rules. The
11 residual directly contributes to effects of policy uncertainty, and thus it serves as a very good measure of policy uncertainty identification.
We use fiscal spending as the proxy variable of fiscal policies and the 7-day China interbank offer rate (CHIBOR) as the proxy variable for monetary policies. The use of the CHIBOR as a proxy for monetary policies is common among studies on China’s economy, as it represents the market’s expectations of short-term rates. Regarding the rule forecasting of fiscal policies, we find fiscal spending to be a function of lagged inflation and real outputs based on Shoudong Chen and Dongliang Yang (2009):
4 1 2 4 , t t t t t t y y f （1） 1, 1 1 1 1 2 2 2, 1 2 0 = + 0 t t t t t t t t （2） 2 0 0 , 0 0 t t N Q （3）Formula (1) is the measurement equation of the state space model where is fiscal expenditures, where y is the output and where is the inflation rate. The above formula implies that the government cares about outputs and the inflation rate, and the inflation of previous outputs determines current fiscal expenditures. Formula (2) is the state equation of the state space model, and time-varying parameter follows the form of AR (1). Formula (3) shows that random shocks are independent of one another and follow a normal distribution.
For the prediction of monetary policies we follow related studies by using the Taylor rule to represent monetary policy stances (Bekaert et al (2011), Mumtaz and Zanetti
(2013)) and to observe how monetary levels and nominal interest rates grow as a
function of the inflation rate, real interest rate, output gap and inflationary gap. The formula is as follows:
1, 2
*
1, 2
** t t t t t y y r i （4） 1, 1 1 1 1 2 2 2, 1 2 0 = + 0 t t t t t t t t （5）12 2 0 0 , 0 0 t t N Q （6）
In formula (4) i is the real rate of interest, is the inflation rate, y* is the natural
output and is the natural inflation rate. We obtain these variables by using the HP filter to separate long-term trends from seasonally adjusted quarterly GDP and
quarterly inflation rates. The time varying parameter of state equation (5) follows the form of AR (1). Random shocks and are the same as those of formula (3). One-step forward predictions obtained from (1), (2), (4) and (5) represent the values of policy variables predicted by individuals in period t + 1. We take the forecast error as our proxy variable of policy uncertainty to determine the degree of deviation between a policy and individuals' rational expectations. When is positive, the increase in fiscal expenditures or interest rates is greater than was expected. When is negative, fiscal expenditure or interest rate reductions are more pronounced than was expected. The greater the value of the error is, the greater the unexpected change in economic policies and the higher economic policy uncertainties are. Regarding subjective policy uncertainty, we use the China Policy Uncertainty Index (EPU Index) measured by Baker et al (2013) as a proxy variable. The index employs media content statistics to record the number of times that the Chinese news media refer to policy uncertainty over a certain period of time. The timely reporting of news media and the capturing of main political shifts serve as good indicators of uncertainty in subjective policies. The indicator is measured on a monthly basis for January 1995 to March
2015. Figure 1 shows EPU index and monetary and fiscal policy uncertainty trends.
0 100 200 300 400 500 600 90 92 94 96 98 00 02 04 06 08 10 12 14 16
13 -3 -2 -1 0 1 2 3 4 90 92 94 96 98 00 02 04 06 08 10 12 14 16 -2,000 -1,500 -1,000 -500 0 500 1,000 90 92 94 96 98 00 02 04 06 08 10 12 14 16
Figure 1 EPU Index (top), Monetary Policy Uncertainty Index (middle), and Fiscal Policy Uncertainty Index (bottom)
From Figure 1 the EPU indicators show pronounced policy uncertainty for 2002, 2009 and 2012 corresponding to China's accession to the WTO and the subprime mortgage crisis. The figure shows that during these periods society showed great concern for issues of policy uncertainty. Especially in the post-crisis period China's economic growth declined, triggering a large number of policy debates. The highest levels of monetary policy uncertainty occurred during the financial crisis in Southeast Asia, during China's accession to the WTO, during the SARS outbreak, during the subprime mortgage crisis and during the post-crisis period. In contrast to other economic data for these periods we find that both the output gap and the inflation gap fluctuate considerably and this is why it is more difficult for individuals to predict changes in policy. However, fiscal policy uncertainty remains relatively stable overall. High levels of uncertainty occurred during the financial crisis in Southeast Asia, during China's accession to the WTO, during the SARS outbreak, during the subprime mortgage crisis and during the post-crisis period. This shows that it is difficult to predict the Chinese government’s policy stance for this period. Especially in the post-crisis period growth in fiscal expenditures has fluctuated considerably, resulting in a high level of policy uncertainty. While the three uncertainty indexes present certain similarities, we still apply them all because the correlation coefficient is not significant, and the three indexes are of varying guiding significance to policymaking. Monetary and fiscal policy uncertainty is estimated from the index, which objectively
14 reflects policymaker uncertainty in formulating policies and which directly helps policymakers measure uncertainty in their own behaviour. The uncertainty index for subjective policies directly measures policy uncertainty felt in society and denotes the effect of policy uncertainty on the terminal. A government and central bank can regard the index of monetary and fiscal policy uncertainty as an intermediary index that can be directly manipulated and the EPU index as a final performance indicator.
2.2 SVAR model
In this paper we construct a structural vector autoregressive model (SVAR) to reflect the impacts of policy uncertainty on key macroeconomic variables. Due to the high endogeneity of economic policy uncertainty (Aastveit et al (2013)), we use policy uncertainty as an endogenous variable and work with other economic variables to construct a SVAR model. SVAR applies priori constraints determined by principles of economics to variables based on the VAR model.
n AA( )Y = ADZ + Au Au = Be (e ) = 0 (e e ) = I t t t t t t t t L E E （7）
Y , , , , , u VGW(0, ) t t t t t t t t u Y C I X In the SVAR model of formula (7), Y is a k-dimension vector of endogenous variables, Y is the output, C is consumption, 𝐼 is investment, 𝜋 is the inflation rate,
𝑋 is exports, and 𝜎 is policy uncertainty. Therefore, dimension 𝑘 is valued at 6. Z is
the exogenous variable vector, and in this paper is considered the world energy price index. As a manipulative variable, energy prices can capture anomalous changes in economic variables caused by external supply shocks. D is the coefficient matrix of exogenous variables, utis the random perturbation vector at time t, and Vector
Gaussian White Noise (VGW) denotes the Gaussian white noise process, A(L) = Ip−
A−1A1L − A−1A2L2− ⋯ − A−1APLP, where L is the lag multiplier and where Aiis the
coefficient matrix of lagged terms of endogenous variables. B can be regarded as the Cholesky orthogonal decomposition matrix, and the B matrix adjusts residuals Aut to
form an orthogonal perturbation et. Therefore, matrices A and B have the following
15
t t t t t t′
A (A ) =Pe (Pe ) =P(e e )P =PP （8）
The A and B matrices take the following form:
12 16 11 12 16 21 26 21 22 26 6 6 6 6 61 62 61 62 66 1 a 1 A B 1 a b b b a a b b b a a b b b （9）
𝑎𝑖𝑗represents the immediate effect of the jth variable on the ith variable for the
current period. We restrict the interaction between these variables based on the theory of economics. When the jth variable has no effect on the ith variable in the current period, 𝑎𝑖𝑗= 0. Matrix B uses mutually independent perturbation term et to generate Aut from a certain linear combination.
To ensure that the SVAR model can be identified, 72 constraints must be added to its parameters. As (8) applies 21 conditions to the P matrix, only 51 more constraints need to be added.
As the estimates of SVAR models are consistent, they are not suitable for analysing the relationship between variables. Therefore, we use an impulse response to analyse any variable's impact on other variables. The impulse response function is as follows:
, L ( ) = i t s ， 1,2,3, ij jt y s s e （10）
Formula (10) shows the impact of the use of the j-th variable of 𝑦t+s on the i-th
variable of Yt for period t+s. As the error terms ejtare independent of one another,
we can define them as innovation from the jth variable or as the pulse impulse from the jth variable. We observe the impact of the new variable for policy uncertainty on other economic variables. When policy uncertainty can have a negative impact on certain economic variables, policy uncertainty is negative for the impulse response function of the economic variable.
Then I will show the identification of the SVAR model: (a) Aggregate demand curve (IS curve)
16 Equation (11) shows the aggregate demand curve in the form of an IS curve with uncertainty added. The current output consists of consumption, investment, exports and constant items covering other items subject to uncertainty σ.
(b) Aggregate supply curve (Phillips curve)
𝜋𝑡 =0+1𝑦𝑡+1𝜎𝑡 （12）
According to the theory of the sticky price, divergence between outputs and natural outputs and divergence between the price level and the expected price level are positively correlated (Ball and Mankiw (1994)). Therefore, the inflation rate can be expressed as a function of outputs. Under assumptions of the sticky price, the adjustment of commodity prices is comparatively slow, as current prices are not affected by current consumption, investments or exports.
(c) Consumption function
𝐶𝑡= 𝛽0+ 𝛽1𝑦𝑡+ 𝛽2𝜋𝑡+ 𝛽3𝜎𝑡 （13）
Formula (13) shows the Keynesian absolute income function and expresses
consumption as a linear function of total income and policy uncertainty. Price levels can affect consumer incomes but can also be seen as a determinant of consumption. At the same time, consumption theory often does not apply investment and exports as the determinants of consumption, and thus we set current investments and exports relative to current consumption impacts as zero.
(d) Investment function
𝐼𝑡 = 𝛾0+ 𝛾1𝑦𝑡+ 𝛾2𝜋𝑡+ 𝛾3𝜎𝑡 （14）
According to neoclassical theory, investments are a function of interest rates and of the marginal product of capital. Formula (14) shows that investment is affected by current outputs, inflation and uncertainty. The investment function does not generally consider the impacts of consumption or exports, and so we limit impacts of current consumption and exports on current investments to zero.
(e) Export function
At the same time, we assume that current exports are only a function of foreign demand and exchange rates. As exchange rates can be affected by prices and outputs, we express the export function as a function of inflation and policy uncertainty:
𝑋𝑡 =0+1𝜋𝑡+2𝜎𝑡 （15）
Economic policy uncertainty should be independent of other variables in the current modelling period. According to Amisano and Giannini (2012), we can limit the orthogonal decomposition matrix B to a diagonal matrix. From the above conditions
17 we impose exactly 51 constraints on the SVAR model to ensure that the SVAR model is precisely identified. After applying constraints, matrices A and B are as follows:
11 1 2 3 4 5 22 1 2 3 33 1 2 3 44 1 2 55 1 2 66 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 A= B 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 b b b b b b （16）
18
3 Data
We use data for the first period of 1995 to the second period of 2017. We collect data from the National Bureau of Statistics of China, Wind Info (a Chinese financial information database), the China Center of Economic Research (CCER) and World Bank Data. Wee seasonally adjust seasonal data following Census X-12’s method.
Table 1: Data adjustment methods Variables/
Shorthand notations
Variable meanings Calculation methods
Adjuste d or not
RGDP/ Y Real GDP
Divide gross national product by the consumer price index for the current period
Yes
CONS/ C Consumption
Consumable retail levels for the current period
Yes
INV/ I Investment
Increases in investment for the current period
Yes
EXPO/X Export
Export increases for the current period
Yes
CPI
Consumer Price Index
Taking 1986 as the base year (=100), the current value is equal to the mean value for three months for the current season
Yes
INF/𝜋 Inflation rate
Chain relative ratios of the Consumer Price Index
19 GAPGDP Output gap
Seasonally adjusted GDP minus time trends estimated from the Hodrick-Prescott filter
No
GAPINF Inflation gap
Inflation rate minus time trends estimated from the Hodrick-Prescott filter
No
FISCAL/f Fiscal expenditures
Fiscal expenditures of the current season
Yes
CHIBOR/i
Nominal interest rate
The 7-day China interbank offer rate
No
PREMIUM Risk premium
The 4-month China interbank offer rate minus the 7-day China
interbank offer rate
No
STOCK Stock return rate
Weighted mean values of stock return rates for the Shanghai stock market
No
FPU
Fiscal policy uncertainty
Prediction errors of formula (5) No
MPU
Monetary policy uncertainty
Prediction errors of formula (6) No
EPU
Baker’s economic policy uncertainty
Drawn from Baker’s website No
Note: Real outputs, investments, consumption, net exports, the consumer price index, and fiscal expenditures undergo seasonal variations. Therefore, this paper adjusts them seasonally.
20 To ensure the stability of the SVAR model, we first-order difference the first-order single integer. Therefore, outputs, consumption, investments and exports are defined as output growth, consumption growth, investment growth, export growth and inflation growth. Below we present our Augmented Dicky-Fuller (ADF) monolithic test of variables considered in the SVAR model:
Table 2: ADF test
Variable
T-test The first difference test
Integration ADF value P value ADF value P value
Y 1.3788 1.0000 -7.6815*** 0.0000 I(1) C 3.1295 1.0000 -3.9268** 0.0145 I(1) I -3.8252** 0.0200 — — I(0) X -1.6884 0.7489 -7.1779*** 0.0000 I(1) FU -8.5246*** 0.0000 — — I(0) MU -4.6306*** 0.0019 — — I(0) EPU -4.1072*** 0.0092 — — I(0)
Note：The lag period is chosen according to SIC guidelines. ***, **, and * denote significance levels of 1%, 5%, 10%, respectively. (C, T, L) respectively denote whether the ADF test applies constant items (0 means no, 1 means yes), whether there is a time trend item (0 means no, 1 means yes), and the hysteresis order.
21 4 Results -15 -10 -5 0 5 10 15 1 2 3 4 5 6 7 8 9 10 -12 -8 -4 0 4 8 1 2 3 4 5 6 7 8 9 10 -12 -8 -4 0 4 8 1 2 3 4 5 6 7 8 9 10
Figure 2 The impulse of the output caused by economic policy uncertainty.
Note: From left to right the figure shows EPU, MPU, and FPU impacts. (The same applies for the following figures). -.006 -.004 -.002 .000 .002 .004 1 2 3 4 5 6 7 8 9 10 -.008 -.006 -.004 -.002 .000 .002 .004 .006 .008 1 2 3 4 5 6 7 8 9 10 -.012 -.008 -.004 .000 .004 .008 1 2 3 4 5 6 7 8 9 10
Figure 3: The impulse of the consumption growth caused by economic policy uncertain
-400 -200 0 200 400 600 1 2 3 4 5 6 7 8 9 10 -600 -400 -200 0 200 400 1 2 3 4 5 6 7 8 9 10 -800 -600 -400 -200 0 200 400 1 2 3 4 5 6 7 8 9 10
Figure 4: The impulse of investment growth caused by economic policy uncertainty.
-.003 -.002 -.001 .000 .001 .002 1 2 3 4 5 6 7 8 9 10 -.004 -.003 -.002 -.001 .000 .001 .002 1 2 3 4 5 6 7 8 9 10 -.003 -.002 -.001 .000 .001 .002 .003 1 2 3 4 5 6 7 8 9 10
22
Figure 5: The impulse of the inflation rate caused by economic policy uncertainty.
-4,000,000 -2,000,000 0 2,000,000 4,000,000 6,000,000 1 2 3 4 5 6 7 8 9 10 -8,000,000 -6,000,000 -4,000,000 -2,000,000 0 2,000,000 4,000,000 6,000,000 1 2 3 4 5 6 7 8 9 10 -8,000,000 -6,000,000 -4,000,000 -2,000,000 0 2,000,000 4,000,000 1 2 3 4 5 6 7 8 9 10
Figure 6: The impulse of export growth caused by economic policy uncertainty. According to Figure 2 the three uncertainty indexes have a negative impact on output
growth. However, after the impact of EPU reaches a minimum value in the second phase, the rebound has a positive impact and the positive impact slowly declines. This reveals certain degree of overreaction in social subjective perceptions of policy
uncertainty. From the outset, individuals grow very cautious of uncertainty and choose to limit their investments and spending. However, after half a year (two periods), policies applied six months prior begin to become more evident. Individuals find that they had been overly cautious in the past and thus restore and engage in more production activities. Monetary policy uncertainty had a optimistic impact on output growth in the first two periods, but this effect quickly declined to under zero and fluctuated around zero, revealing a short-term and unsustainable impact. Uncertainties in fiscal policy had a positive impact on output growth in the first phase but became negative in the second. Changes in China’s fiscal policies accompanied by the execution of a large number of positive public information campaigns can increase social confidence, which may spur more overreaction. After the first quarter (one period) individuals found that effects of fiscal policy before the first quarter did not reach expected levels, and the macroeconomy thus contracted.
In Figure 3 both the EPU and the monetary policy uncertainty index have a negative impact on consumption growth for the first seven periods, but this impact fluctuates and slowly declines to zero. This shows that individuals pay more attention to subjective perceptions of the EPU and to monetary policy uncertainty and that the
23 process of adjusting the consumption path is relatively slow and stable. Moreover, there is no over-reaction but repetitive. The impulse response of fiscal policy uncertainty to consumption growth is similar to the impulse response of monetary policy uncertainty, indicating that individuals' consumption decision-making is less affected by fiscal policy uncertainty. This may be attributed to the fact that the consumption growth examined in this article reflects a change in consumption occurring over a relatively short period of time and that many fiscal policies have a relatively slow economic impact on the economy as monetary policies do. We thus find consumer-to-policy uncertainty as sensitive.
Figure 4 shows that for the first two periods the investment growth rate has obviously been negatively impacted by the three forms of policy uncertainty. The impact
convergence from monetary policy uncertainty clearly manifests faster than subjective and fiscal policy uncertainty, as the capital market can respond promptly to an
adjustment in monetary policy. For a weak effective market, external shocks can quickly be absorbed by the market. In contrast, the impacts of subjective policy uncertainty and fiscal policy uncertainty on investment are persistent and repetitive and in the opposite direction. After period 4, subjective policy uncertainty has had a positive impact on investment growth while fiscal policy uncertainty has had a negative impact on investment growth, but both shocks have not significantly moved away from the zero line and we cannot prove that subjective and fiscal policy
uncertainty have a lasting impact on investment.
Figure 5 shows that both subjective and fiscal policy uncertainty have a significantly negative impact on the inflation rate in the first phase, implying that the two forms of policy uncertainty have inhibitory effects on inflation. The inhibition of subjective policy uncertainty and fiscal policy uncertainty rapidly disappears, but monetary policy uncertainty has only converged to zero after two years (after the eighth period). That is, the impact of monetary policy uncertainty on prices is relatively long lasting. The inflationary inhibitory effect of policy uncertainty is related to the fact that
24 uncertainty leads to the development of a more cautious economic entity, to a
reduction in demand for consumption and investment, and to cooling in the economy. However, this inflation-inhibiting effect is based on assumptions of sacrificing output and employment. It cannot be assumed that policy uncertainty can become a tool for price control.
In Figure 6 both subjective and monetary policy uncertainty have a short-term
optimistic impact on export growth. At the same time, we find a slight overreaction in the impact of exports on the uncertainty of fiscal policies. Monetary policies can have a more timely impact on exports, as they can effectively affect the exchange rate. However, fiscal policies are often targeted at the government spending and they can indirectly affect export-oriented enterprises. Therefore, fiscal policy uncertainty has significant impact on export growth.
In general, policy uncertainty has a negative impact on outputs, consumption, investment and inflation in the short run. This warns policymakers that unstable economic policies can have a counterproductive effect on the economy by generating uncertainty. Therefore, when the economy is in a downturn, a government and central bank should express policy objectives in a timely manner and implement them
25
5 Analysis of the macroeconomic channel of policy uncertainty
Consumption and investment are two of the most important components of China's GDP, and thus consumption growth and investment growth are often the two most central statistical subjects for domestic policy-making officers. This paper has empirically shown that policy uncertainty has an impact on consumption and investment. However, it goes without saying that we next question through which channel policy uncertainty can affect consumption and investment. We show above that policy uncertainty has an inhibitory effect on inflation. Naturally, we would also like to determine through which channel policy uncertainty inhibits inflation. To investigate this, we develop a model of V (an economic variable affected by policy uncertainty) that refers to consumption growth, investment growth and inflation. In this model, Channel proxies for the channel we intend to test:
V = 𝛽0+ 𝛽1𝑐ℎ𝑎𝑛𝑛𝑒𝑙 + 𝛽2𝐸𝑃𝑈 ∗ 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 + 𝛽3𝑀𝑈 ∗ 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 + 𝛽4𝐹𝑈 ∗ 𝑐ℎ𝑎𝑛𝑛𝑒𝑙 （17）
V is the macro variable of interest and includes consumption growth, investment growth and inflation. Note that when a channel is significantly correlated with variable V, we expect β1 to be significantly non-zero. When subjective policy
uncertainty, monetary policy uncertainty and/or fiscal policy uncertainty can increase or decrease a channel’s impact on consumption or other economic variables, we expect β2, β3 and β4 to be nonzero.
5.1 Channels through which policy uncertainty affects consumption
The precautionary saving theory developed by Leland (1968) proposes that
individuals are cautious about risk and namely, that the third moment of the consumer utility function is not zero. When individuals expect to exhibit more volatility in their consumption, individuals will be concerned that consumption will not be guaranteed within a certain period in the future, and as a result they will increase their savings and limit their current consumption. Consumption volatility is proxied by the conditional variance in the consumption growth rate (Sundaresan (1989), Tédongap (2014)). This paper uses GARCH (1, 2) to separate volatility from the growth rate of consumption. Meanwhile, precautionary savings theory states that in addition to individuals' expectations, the relative level of personal risk aversion is an important factor that alters the consumption direction. Highly risk-averse consumers will limit their current consumption to protect future consumption levels. In measuring the
26 overall social risk aversion coefficient, the risk premium of venture capital can be used as a rough substitute proxy (Sheng Wang and Mingchao Cai (2011)). Referring to Ross (1986) we use the short-term interest rate difference as the risk premium. Unlike Ross (1986), we use the interest rate differential between the 4-month lending rate for Chinese banks and the inter-bank 1-month lending rate. Through a correlation analysis we confirm that both consumption volatility and risk aversion are positively correlated with policy uncertainty. We can thus assume that policy uncertainty affects consumer spending through two potential channels: consumer expectation and risk aversion.
Table 3 shows that consumption fluctuations reduce consumption growth and further proves that the consumption utility function of Chinese residents corresponds with the precautionary saving model proposed by Leland (1968). The cross-term coefficient of monetary and fiscal policy uncertainty and of consumption volatility is negative, indicating that monetary and fiscal policy uncertainty aggravates the adverse effect of consumption fluctuations in household consumption growth. Rather, when monetary and fiscal policy uncertainty is observed, residents become more concerned about effects of fluctuations in consumption, further reducing consumption under the same consumption fluctuations. The coefficient of the channel variable of the risk aversion channel is negative, indicating that an increase in risk aversion will also significantly reduce consumption. This result shows that China’s intertemporal consumption utility function shows a tendency towards risk aversion and that an increase in risk aversion tendencies can reduce current consumption while increasing the proportion of savings. Moreover, the cross-coefficient of monetary policy uncertainty and risk aversion is negative, implying that monetary policy uncertainty will aggravate negative effects of risk aversion on consumption. At the 10% confidence level subjective policy
uncertainty also has a negative impact on consumption through risk aversion channels.
5.2 Channels through which policy uncertainty affects investment
Investment decisions made in firms experiencing uncertainty are affected by expected returns and by the volatility of returns on investment (Dixit (1994), Pindyck (1990)). The lower the return on investment and the higher the volatility of returns is, the more limited investment activity becomes. We thus predict that policy uncertainty spurs fluctuations in investment income and that policy uncertainty may limit investments by affecting the investment return. We use stock prices to calculate investment returns,
27 as stock prices are the outcome of the listed company that uses the funds to invest in carefully selected business projects, and it can thus reflect the income level of such investment projects. We thus use the SSE stock index as the proxy variable of the social total investment return. Bernanke (1980) and Bloom (2009) demonstrate that macroeconomic volatility can also significantly affect firm returns on investment. Clearly, the stock market is often influenced by economic uncertainty. Yizhong Wang (2013) also shows that macroeconomic fluctuations have a negative impact on the investment needs of enterprises. We thus believe macroeconomic fluctuations may also serve as channels of policy uncertainty. Following Bredin and Fountas (2009) and Yizhong Wang (2013) we use the conditional variance of GARCH (1, 1) as the proxy variable for macroeconomic volatility for the first-order difference in quarterly GDP. We also take corporate financing costs into account.
As is shown in Table 3, the coefficient of returns on investment is significantly positive, meaning that an increase in investment income can stimulate investment. However, the crossover coefficient of monetary policy uncertainty and of investment income is negative, indicating that positive news on return increases should be partially offset by monetary policy uncertainty. Black-Scholes-Merton's option pricing model (Black and Scholes (1973), Merton (1973)) implies that when returns on an investment project fluctuate, the IRR of an investment is lower than the return of the investment itself, and furthermore the higher volatility levels are, the greater the difference between the project yield rate and the IRR becomes and the less attractive the project becomes. Monetary policy uncertainty causes variables (e.g., interest rates, exchange rates and inflation) that affect the rate of return to fluctuate, reducing the IRR of such projects and rendering good news less attractive to investors. We thus also come to a conclusion on small policies that monetary policy uncertainty may weaken the monetary stimulus to an economy. In the channel of macroeconomic fluctuations we find that macroeconomic fluctuations lead to a reduction in
investment. This occurs because uncertainty increases the value of investors waiting and because individuals seek to invest until uncertainty diminishes, and thus
uncertainty reduces current investments (Bernanke, 1980; Dixit, 1994). From the table we find that fiscal policy uncertainty aggravates the negative effect of economic fluctuations on investment, as fiscal policy reflects government attitudes towards the economy. When a government adopts a laissez-faire attitude, individuals expect economic fluctuations to have a greater impact. By contrast, when a government is determined to stabilize fluctuations, individuals expect economic fluctuations to be reduced. Shugin Zhu and Mingyong Lai (2005) and Yang et al. (2014) also confirmed the relationship between fiscal policies and economic fluctuations. They showed that
28 fiscal policy uncertainty increases uncertainty in economic growth. We can thus conclude that there is a strong internal relationship between fiscal policy uncertainty and overall macroeconomic fluctuations.
5.3 Channels through which policy uncertainty affects inflation
The theory of the sticky price (Ball and Mankiw (1994)) and incomplete information theory (Lucas (1977)) show that the degree of deviation between inflation and inflation expectation is positively correlated with the output gap, and thus we try to examine whether inflation expectations and output gaps can transmit impacts of policy uncertainty. In measuring inflation expectations, previous research (Feige and Pearce (1976)) has shown that the autoregressive term of inflation can better represent rational expectations of inflation. This autoregressive term is also the adaptive
expectation (Mankiw et al (2003)). The measurement of the output gap is illustrated in Table 1.
According to Table 3, wherein the coefficient of the adaptive expectation channel variable is very close to 1, adaptive expectations for inflation may be present in China. To further investigate this point, we conducted a Wald Test to estimate the coefficient of the first-order inflation lag of the expected adaptiveness channel. The coefficient of the null hypothesis is equal to 1 and we found that we could not reject the null
hypothesis, meaning that there is an adaptive expected effect on inflation in China. However, the coefficients of sub-policy uncertainty and the crossover term between monetary policy uncertainty and the first-order inflation term are significantly negative, indicating that policy uncertainty mitigates effects of adaptive expectations on inflation. Rational expectation theory holds that monetary and fiscal policies significantly contribute to the existence of adaptive expectations (Sargent (1982)) whereas policy uncertainty reduces adaptive expectations of inflation, as policy uncertainty has unfavourable impacts on the economy in the short term and causes the economy decline slightly while mitigating inflationary pressure. The fact that output gap channels stimulate inflation while the cross-term coefficient of monetary policy uncertainty and of the output gap is significantly negative at the 10% confidence level means that the existence of monetary policy uncertainty weakens inflation effects of the output gap. In the adaptive expectation and output gap channels, policy
uncertainty has inhibitory effects on inflation consistent with SVAR model results presented in this paper.
29 Dependent
variable
The dependent variable is the first-order difference of
consumption
The dependent variable is the first-order difference of
investment
The dependent variable is the inflation rate Channel Consumption fluctuation channel Risk aversion channel Investment yield channel Economic fluctuation channel Adaptive expectation channel Output gap channel Constant term 97.11*** （8.9024） 1047.20*** （7.8108） -36.01 （-0.4483） 181.92*** （5.0068） 0.01* （1.8928） 0.20*** （11.1900） Channel -189.06*** （-2.0543) -342.92** （-2.4759） 0.08** （2.1724） -27.92** （-2.1580） 1.02*** （22.147） 0.01** （2.1158） EPU*channel 0.1421 （1.5478） -2.24* （-1.8185） 0.00 （0.8467） 33.35 （1.4690） -0.01* （-1.8564） 0.00 （0.1392） MPU*channel -126.70*** （-1.8556） -133.45* （-1.8999） -0.03** （-1.9605） -1800.11 （-1.4025） -0.092*** （-4.1883） -0.01* （-1.7486） FPU*channel -59.47*** （-2.5314） 62.77 （0.8086） -0.00 （-1.1881） -1335.97* （-1.6698） 0.02 （1.1833） -0.01 （-0.7183） Sample size 202 228 237 237 237 77
Note: “Channel” is the channel’s proxy variable. Proxy variables of the seven channels from left to right in the table are consumption volatility, short-term and inter-bank interest rates, Shanghai SSE, output volatility, short-term and inter-bank lending rates, the first-order lagged inflation rate, and the output gap.
Channels through which policy uncertainty affects macroeconomics provide policy makers with a new perspective. As monetary and fiscal policies employ different channels of influence, policy makers can reduce impacts of policy uncertainty on consumption and investment by deliberately adopting certain policies. From Table 3 we can see that fiscal policy uncertainty did not significantly affect the long-term interest rate differential or the Shanghai Composite Index, but it also did not
30 exacerbate negative impacts of risk aversion and risk premiums on consumer
investments nor did it offset investment incentives to stimulate investment. Hence, the impact of fiscal policy uncertainty is weaker than that of monetary policy uncertainty. The effect of monetary policy is often quickly reflected in the capital market, and its uncertainty effects can therefore spread quickly. However, the main instruments of fiscal policy (i.e., government purchases and government bonds) tend to be regional and time-lagged, rendering them less likely to generate pronounced uncertainty across a large area. Therefore, in the face of an economic downturn, the government may first consider altering and refining its fiscal policy to stimulate the economy.
Moreover, the government should exercise caution towards the formation of monetary policies and adopt a stable monetary policy to limit negative impacts of policy
31
6 Conclusions
This paper divides policy uncertainty into subjective and objective policy uncertainty. The latter can be divided into monetary and fiscal policy uncertainty. In this paper, the uncertainty index of monetary and fiscal policies is obtained by applying the state space method with time-varying parameters. All three forms of policy uncertainty are heterogeneous in an economic sense. Subjective policy uncertainty reflects the ending role of policy uncertainty while monetary and fiscal policies reflect the starting point of policy uncertainty. Major policy-making bodies such as central banks and
governments can evaluate uncertainties related to their own policies by evaluating monetary and fiscal policy uncertainty levels. Furthermore, policy-making bodies can monitor the degree to which individuals feel that policies carry uncertainty through an examination of subjective policy uncertainty levels. The three measures outlined can help policy-making bodies focus on issues of policy uncertainty from multiple perspectives while limiting unfavourable impacts of policy uncertainty on the macro economy.
Using the SVAR model we tested dynamic impacts of policy uncertainty on economic variables and found that policy uncertainty initially has a negative impact on output growth, consumption growth, investment growth and export growth while policy uncertainty has an inhibiting effect on inflation. There are some differences between subjective, monetary and fiscal policy uncertainty. The impacts of subjective policy uncertainty on output growth and investment growth are found to rebound and become sustained while impacts of monetary policy uncertainty on inflation are insignificant. The impacts of fiscal policies on consumption growth and export growth are insignificant as well. However, these differences do not affect our conclusion that policy uncertainty negatively affects the economy.
By examining influencing channels of policy uncertainty we find that subjective and monetary policy uncertainty aggravate negative impacts of risk aversion on
consumption. Monetary policy uncertainty limits the positive effect that increased investment returns have on investment, and fiscal policy uncertainties amplify negative impacts of economic fluctuations on investment. Subjective and monetary policy restrain adaptive expectations of inflation, but monetary policy uncertainty can also weaken inflation effects of the output gap.
This paper shows that policy-making bodies must consider policy uncertainty levels when examining their own policies. To curtail negative effects of policy uncertainty
32 on the macro economy, governments and central banks may reduce levels of policy volatility by implementing stable and predictable monetary and fiscal policies. When there is a need for policy adjustments, they may consider adjusting fiscal policies to limit uncertainty effects while maintaining the stability of on-going monetary policies. However, in the long run, governments and central banks must focus on rendering policy formulation and implementation more transparent by stressing the public nature of policies and by expressing their own policy positions in a timely manner. Such bodies must also accelerate expected adjustments of social uncertainty and shorten the periods of policy uncertainty to limit negative impacts of policy uncertainty on the macro economy.
33
References
Aastveit, K., Natvik, G. and Sola, S. (2013). Economic uncertainty and the effectiveness of monetary policy. Oslo: Norges Bank.
Aizenman, J. and Marion, N. (1993). Policy Uncertainty, Persistence and Growth. Review of International Economics, 1(2), pp.145-163.
Amisano, G. and Giannini, C. (2012). Topics in structural VAR econometrics. Springer Science & Business Media.
Ball, L. and Mankiw, N. (1994). A sticky-price manifesto. Carnegie-Rochester Conference Series on Public Policy, 41, pp.127-151.
Baker, S., Bloom, N. and Davis, S. (2013). Measuring Economic Policy Uncertainty. SSRN Electronic Journal.
Baker, S., Bloom, N., Canes-Wrone, B., Davis, S. and Rodden, J. (2014). Why Has US Policy Uncertainty Risen Since 1960?. American Economic Review, 104(5), pp.56-60.
Bekaert, G., Hoerova, M. and Lo Duca, M. (2011). Risk, Uncertainty and Monetary Policy. SSRN Electronic Journal.
Bergh, A., Berggren, N. and Bjørnskov, C. (2009). The Growth Effects of Institutional Instability. SSRN Electronic Journal.
Bernanke, B. (1983). Irreversibility, Uncertainty, and Cyclical Investment. The Quarterly Journal of Economics, 98(1), p.85.
Black, F. and Scholes, M. (1973). The Pricing of Options and Corporate Liabilities. Journal of Political Economy, 81(3), pp.637-654.
Bloom, N. (2009). The Impact of Uncertainty Shocks. Econometrica, 77(3), pp.623-685.
Born, B. and Pfeifer, J. (2014). Policy risk and the business cycle. Journal of Monetary Economics, 68, pp.68-85.
Bredin, D. and Fountas, S. (2009). Macroeconomic uncertainty and performance in the European Union. Journal of International Money and Finance, 28(6), pp.972-986.
34 Sharpe, W. (1964). Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. The Journal of Finance, 19(3), p.425.
Schwartz, E., Dixit, A. and Pindyck, R. (1994). Investment Under Uncertainty. The Journal of Finance, 49(5), p.1924.
Fatás, A. and Mihov, I. (2013). Policy Volatility, Institutions, and Economic Growth. Review of Economics and Statistics, 95(2), pp.362-376.
Feige, E. and Pearce, D. (1976). Economically Rational Expectations: Are Innovations in the Rate of Inflation Independent of Innovations in Measures of Monetary and Fiscal Policy?. Journal of Political Economy, 84(3), pp.499-522. Fernández-Villaverde, J., Guerron-Quintana, P., Kuester, K. and Rubio-Ramirez, J. (2011). Fiscal Volatility Shocks and Economic Activity. SSRN Electronic Journal. Fernández-Villaverde, J., Guerron-Quintana, P., Rubio-Ramirez, J. and Uribe, M. (2009). Risk Matters: The Real Effects of Volatility Shocks. SSRN Electronic Journal.
Fountas, S. and Karanasos, M. (2007). Inflation, output growth, and nominal and real uncertainty: Empirical evidence for the G7. Journal of International Money and Finance, 26(2), pp.229-250.
Gentzkow, M. and Shapiro, J. (2006). What Drives Media Slant? Evidence from U.S. Daily Newspapers. SSRN Electronic Journal.
Handley, K. and Limao, N. (2012). Trade and investment under policy uncertainty. Cambridge, MA: National Bureau of Economic Research.
JULIO, B. and YOOK, Y. (2012). Political Uncertainty and Corporate Investment Cycles. The Journal of Finance, 67(1), pp.45-83.
Johannsen, B. (2014). When are the Effects of Fiscal Policy Uncertainty Large?. SSRN Electronic Journal.
Knight, F, H. (1921). Risk, uncertainty and profit. New York: Hart, Schaffner and Marx.
Klomp, J. and de Haan, J. (2009). Political institutions and economic volatility. European Journal of Political Economy, 25(3), pp.311-326.
35 Leland, H. (1968). Saving and Uncertainty: The Precautionary Demand for Saving. The Quarterly Journal of Economics, 82(3), p.465.
Lucas Jr, R.E., Prescott, E.C. (1971). Investment under uncertainty. Econometrica: Journal of the Econometric Society, 1971, 659-681.
Lucas, R. (1976). Econometric policy evaluation: A critique. Carnegie-Rochester Conference Series on Public Policy, 1, pp.19-46.
Lucas, R. (1977). Understanding business cycles. Carnegie-Rochester Conference Series on Public Policy, 5, pp.7-29.
Menegatti, M. (2010). UNCERTAINTY AND CONSUMPTION: NEW EVIDENCE IN OECD COUNTRIES. Bulletin of Economic Research, 62(3), pp.227-242.
Merton, R. (1973). Theory of Rational Option Pricing. The Bell Journal of Economics and Management Science, 4(1), p.141.
MUMTAZ, H. and ZANETTI, F. (2013). The Impact of the Volatility of Monetary Policy Shocks. Journal of Money, Credit and Banking, 45(4), pp.535-558.
Pindyck, R, S. (1990). Irreversibility, uncertainty, and investment (No. w3307). National Bureau of Economic Research.
Rodrik, D. (1991). Policy uncertainty and private investment in developing countries. Journal of Development Economics, 36(2), pp.229-242.
Stock, J. and Watson, M. (2012). Disentangling the Channels of the 2007–09 Recession. Brookings Papers on Economic Activity, 2012(1), pp.81-135. Tédongap, R. (2014). Consumption Volatility and the Cross-Section of Stock Returns*. Review of Finance, 19(1), pp.367-405.
Sundaresan, S. (1989). Intertemporally Dependent Preferences and the Volatility of Consumption and Wealth. Review of Financial Studies, 2(1), pp.73-89.
Ulrich, M. (2012). Economic policy uncertainty & asset price volatility. Available at SSRN 1566909.
Haisheng, Yang., Shaoling, Chen., Danglun, Luo. and Guoman, Yu. (2014). Policy certainty and Economic growth.
36 policy uncertainty’s effects on individuals’ consumption behaviors.
Sheng, Wang. and Mingchao, Cai. (2011). Chinese residents risk aversion index estimation and influencing factors analysis.
Yizhong, Wang. and Min, Song. (2014). Macroeconomic policy uncertainty, Funding needs and Corporate investment.
Yekun, Xu., Xianhang, Qian. and Weian, Li. (2013). Political uncertainty, Political affiliation and Private enterprise investment.
