STOCK MARKET RETURNS, RISK AVERSION AND CONSUMPTION GROWTH: EVIDENCE FROM THE NIGERIAN ECONOMY

Transcription

1 STOCK MARKET RETURNS, RISK AVERSION AND CONSUMPTION GROWTH: EVIDENCE FROM THE NIGERIAN ECONOMY Favoured Mogbolu Department of Economics and Statistics, University of Benin, Benin City Abstract This paper analyses interdependencies between stock market returns, risk aversion and consumption growth to identify the nature of risk aversion of Nigerian investors/consumers. The nature of risk aversion is important for understanding the behavior of stock returns, and the interdependencies may explain monetary policy effect on the stock market. The previous literature supports risk averse investors/consumers but is inconclusive about the measured level. However, the literature consists largely of developed economy studies. We employed a Generalised method of moment (GMM) estimation techniques using quarterly and annual Nigerian data for the period We attempt to improve the performance of the model by using a measure of monetary policy condition to instrument the relationship among stock return, risk aversion and consumption growth. We found a value of 0.63 as measured level of risk aversion and conclude that the response of stock returns to consumption growth supports existence of weakly risk averse investors/consumers in Nigeria. Keywords: Stock, returns, risk, aversion, consumption Introduction This study analyses the nature of risk aversion of investors/consumers and seeks to derive evidence on the degree of risk aversion and its effect on stock return variability for the Nigerian economy. Interdependencies exist between risk aversion, variability of expected consumption and stock market returns, arising primarily from the assumptions of consumption-based asset pricing models, and receive support from assertions in monetary transmission studies. For example, Ludvigson, Steindel and Lettau (2002) posit that stock value responds to monetary policy and consumption responds to asset value. Furthermore, Bernanke and Kuttner (2005) suggest that monetary policy influence on stock returns may be explained by changes in investors degree of risk aversion which results from variability in expected levels of consumption as shown by Campbell and Cochrane (1999). Likewise, Cook and Hahn (1989) argue that the theoretical negative response of the stock market to monetary policy may increase with increased variability and magnitude of policy changes. The nature of investors risk aversion is important for understanding the behavior of stock returns, and the interdependencies between investors risk aversion, variability of expected consumption level and stock market returns may explain monetary policy effect on the stock market. The existing studies of asset returns and consumption growth are based largely on the more developed stock markets and they are moreover, inconclusive on the degree of agents risk aversion as well as the extent to which risk aversion explains stock returns. These studies include Hansen and Singleton (1982, 1983) and Hunter and Wu (2009). Hansen and Singleton s (1983) study could not explain a significant amount of the variation in returns, while Hunter and Wu (2009) assumed that including individuals aversion to a wealth reference variable was required to explain the premium on stock returns. Their results reported small estimated values, in contrast, Grossman and Shiller (1981) found very large values of the coefficient of risk aversion, and concluded that implausibly large values of the estimated coefficient were required to explain actual returns. Little however, is known about the nature of risk aversion in Nigeria. The concern of this study is that investor s aversion to risk may be relevant in explaining how stock returns are linked to increased macroeconomic and monetary policy variability in the Nigeria economy, following implementation of economic and policy reforms in Nigeria (Ibeawuchi, 2007; Mogbolu, 2014). Meristem (2008) argued that increased variation in monetary policy preceded the 2007/2008 Nigerian stock market deflation, which was accentuated by the U.S subprime mortgage based global financial crisis. Literature on stock market returns in Nigeria establishes that conditions of monetary policy influences stock returns. Of these, Osisanwo and Atanda (2012) show that monetary policy variables determine stock market returns. Similar findings are provided by studies of monetary policy transmission conducted by Okpara (2010) and Nnaji and Ohuche (2008). The existence of interdependencies between stock returns, risk aversion and consumption growth implies that if investors are risk averse, high policy variability may imply adverse influence on stock values (Cook and Hahn, 1989; Pindyck, 1984; Vickers, 2000). The questions arising are (1) to what extent are

2 stock market investors risk averse in Nigeria? and (2) how much does the degree of investors aversion to consumption variability explain stock return variability in Nigeria? This study focuses on analyzing risk aversion using asset pricing theory in the tradition of the Consumption capital asset pricing model (CCAPM) which relates variability in asset returns to consumption growth. The study employs generalized Method of Moments (GMM) methodology and Nigerian data on consumption and stock returns for the period 1986 to We extend the scope of existing studies with evidence on agents risk aversion and its effect on stock market returns by estimating the level of risk aversion in a consumption-based model of stock returns in the emerging Nigerian stock market. Furthermore, the study contributes to the literature on monetary policy effects on stock market returns by conditioning individuals predictions of expected consumption growth on conditions of monetary policy. The rest of the study is presented in five additional sections. Section Two includes a brief review of monetary policy and stock prices in Nigeria during the period under study. Section Three is the review of previous literature while in Section Four we present the theoretical framework and empirical methodology. Sections Five and Six, include the empirical results and conclusions respectively. Monetary Policy and Stock Prices in the Nigerian Economy ( ). This section presents changes in monetary policy and stock prices in Nigeria. We assume that policy variability is indicative of expected variability of economic activity, (monetary authorities are assumed to have superior information about future economic activity, as noted by Romer and Romer, (2000), therefore, observed changes in monetary policy should reflect expectations of changes in macroeconomic activity). We measure monetary policy changes using changes in the monetary policy rate (MPR), and changes in stock prices as changes in the aggregate stock Price Index. We present changes in monetary policy rate and stock prices denoted as ADMPR and ADSR respectively in Table 1 and Figure 1 and examine the data for information on the extent of association between monetary policy and stock return variability.

3 Table 1: Changes in monetary policy and stock prices Year ADSR ADMPR Source: Computed by author based on CBN data.. The information in Table 1 tends to support high association between monetary policy and stock variability, as extent of variability of the monetary policy rate generally corresponds with that of stock prices over the period A key feature of the data is that both ADMPR and ADSR show little variability before early For instance, ADMPR was small before 2000 and in particular zero in six of the 14 years, between 1986 and This trend 1991 and 1993 when reflecting government policy of interest rate liberalization and 1994 when the government adopted open market operations (OMO). Likewise, till early 2000 decade the value of ADSR was small relative to the latter years of the decade. was broken in a few years such as in 1987, 1989, Note: (1) Source: Author based on data in Table 1.

4 (2) The value of ADSR is graphed along the left axis while ADMPR is on the right axis. The low variability of both monetary policy and stock prices before the 2000 decade may not be interpreted to indicate low variability of economic activity, as monetary policy is widely accounted to be inactive, and in the case of stock market, prices were administratively fixed for the most part of this period. An implication of the inactive posture of policy and the administered stock prices is that policy changes may not have predictive content for changes in economic activity, and stock price movement may not contain information on consumers expectations of variation in future consumption. In contrast to the picture before 2000, substantial variability characterized both stock prices and monetary policy rate from Specifically, values of ADMPR varied frequently between 2000 and It was zero only in 2013, implying only one instance of no change in MPR from previous year, in contrast to the picture in the period before An implication of the data on ADMPR and ASR in the later years of the 2000 decade is that variability in the monetary policy is highly associated with stock price variability. This is even more likely, as during this period monetary policy was persistently expansionary, particularly, in the period immediately preceding the 2007/2008 stock price deflation as shown by the negative value of ADMPR, in 2005, in 2006, and in Review of relevant literature In financial economics, studies based on asset pricing models identify risk aversion as a preference parameter governing agents investment behavior. Consumption-based studies including Grossman and Shiller (1981), Hansen and Singleton (1982, 1983), and Epstein and Zin (1989, 1991) align with the tradition of the CCAPM, which is based on the premise that consumption is state dependent and a function of the same state variables as asset returns (where assets includes default-free zero coupon bonds and shares of stocks.) This assumption results in a joint distribution of asset returns and consumption growth, whose parameters are the agents preference parameters that is, the discount factor for discounting future consumption and risk aversion. The use of this class of models in studies of risk aversion yields a relationship between variation in expected consumption growth, relative risk aversion (or equivalently, consumers aversion to variability in consumption) and stock return variability. A prediction of this class of models is that consumption variability may induce a temporal effect on stock price variability whose magnitude depends on the degree of risk aversion. The resulting theory of asset returns has the empirical properties that it can be generally applied, holding for all assets and portfolio as well as for aggregate consumption so long as some people s consumption is well represented by aggregate consumption (Grossman and Shiller, 1981). This claim of generality of the model receives support from findings in U.S studies of the stock market wealth effect on consumption (in a period when U.S. stock holding household were about 25%) that changes in stock prices portend large consumption growth, (see Mankiw and Zeldes, 1991; Porteba, Samwick and Shiller, 1995) but the relation is independent of household stockholding (see Porteba et al, 1995). The general nature of the joint distribution of consumption and stock price movement proposed in this model indicates the potential to obtain correct characterization of aggregate agents risk aversion from its application. In contrast to the use of consumption growth in consumption-based models, an alternative approach called the empirical pricing kernel models analyse investors risk aversion by substituting a summary of the asset s payoffs for consumption growth as the stochastic discount factor in the asset pricing model. Studies based on this approach attempt to identify risk premia across wealth status from observed stock and bond prices. Empirical findings on nature of investors aversion to consumption risk, and associated stock return variability, includes the studies of Grossman and Shiller (1981), Hansen and Singleton (1982, 1983), Garcia, Luger and Renault (2003), Hunter and Wu (2009). These studies provide empirical evidence based on developed economy conditions. An important task in the empirical studies is specifying assumptions about the joint distribution of returns, risk aversion and consumption. The studies differ in how they address this issue. Some impose no restrictions on the form of the joint distribution, and adopt distribution free methods. Among these group, Grossman and Shiller (1981) use non parametric method, and Hansen and Singleton (1982) use nonlinear GMM method. Other studies impose restrictions on the form of the joint distribution. For example, while Hansen and Singleton (1983) assume a lognormal joint distribution of the serial correlation among stock returns and consumption growth, and applied maximum Likelihood method to a VAR of stock

5 returns and consumption variability, Hunter and Wu (2009) assume lognormalilty of the joint distribution, based on a multiplicative error term, and employed linear GMM and nonlinear GMM. Findings in the studies are characterized by differences in the measure of risk aversion. For example, Grossman and Shiller (1981) determined large values of risk aversion ranging from one to six, with increasing model performance corresponding to use of larger values of the coefficient of relative risk aversion. Hansen and Singleton (1982) obtained highly statistically significant estimates of relative risk aversion, with magnitudes below one across models. Hansen and Singleton (1983) found statistically significant coefficient of relative risk aversion equal 1.25 but the overidentifying restrictions imposed on the model were rejected on the basis of the X 2 tests. Garcia et al (2003) adopted switching fundamentals and Eipstein and Zin (1989) preferences, and obtained estimates of risk aversion with average values of and similar to Hansen and Singleton (1982). Hunter and Wu (2009) examined U.K data and augmented the CCAPM models with a U.S wealth reference. They found estimates of 0.471, 2.862, and respectively, for the Linear- IV, the Linear-GMM and the Nonlinear-GMM specifications. An important feature of these results is that the measure of risk aversion is not invariant to the lag length employed in the analysis. Hansen and Singleton (1983) attributes the relatively large values of Grossman and Shiller (1981) to the absence of lags in their analysis to account for serial correlations. But the studies differ on the optimal lag length, for example, Hansen and Singleton (1982) results support a lag length of two while Hunter and Wu (2009) showed support for 12 lags. Empirical studies based on empirical pricing kernel models include Ait-Sahalia and Lo (2000), Jackwerth (2000), Rosenberg and Engle (2002) and Bliss and Panigirtzoglou (2004). These studies provide estimates of risk aversion with substantial variation, for example, Rosenberg and Engle (2002) finds estimated risk aversion ranging between 2.36 to 12.55, while Bliss and Panigirzoglou (2004) estimated average values of 2.33 and Theoretical framework This study employs Hansen and Singleton s (1982, 1983) theoretical model as framework for empirical analysis. This is because their models specifies the conditional co-moments of stock returns and consumption growth based on both use of minimal restrictions as well as the lognormal assumptions. Hansen and Singleton (1982, 1983) assume a production exchange economy with a single good, N assets with w t holdings at date t, and q t vector of prices of the N assets in w t, net of distributions, with vector of values of distributions during period t given by q t * and a representative consumer with a time additive utility function given as, [ ] (1) Where, C t is consumption in time period t, β is the discount factor in consumption and is a strictly concave function. Moreover, The consumer chooses a consumption plan so as to maximize the expected value of (1). The consumer implements consumption plans by changing the value of their portfolio of financial assets (which include fixed income, default free securities as well as equity shares). In each period, the consumption and investment plans must satisfy the budget constraint given by, C t + q t.w r+1 (q t + q t* )(w t +y t) (2) Where, y t is labour income. it was assumed that the agent has a utility function of the Constant relative risk aversion (CARA) class, that is (C t) = and hence the first order condition for the representative consumer to maximize utility involves the equilibrium prices of the N asset which is given as, E t[ ( ) ]=1; I = 1... N (3)

6 Where is the coefficient of relative risk aversion, R it = the return on assets. Equation (3) states that conditional on period t information, the joint variation of expected discounted consumption growth, ( ), the level of risk aversion, and returns on the i th asset, R it, in period t+1 is non-zero. Defining u it = (C t/c t-1) SR it, the first order condition can be written as E t-1[u it] = ; i = 1... N (4), Hansen and Singleton (1983) assume that the joint distribution of asset returns and consumption is lognormal and define LCG t = log (C t/c t-1); LR it = log R it; Y t = (LCG t, LR it,... LSR nt); U it = log u it (i = 1,..., n), and ѱ t-1 as the information set {Y t-s: s 1}. Assuming that {Y t} is a stationary Gaussian process, the distribution of U it conditional on ѱ t-1 is normal with a constant variance and a mean µ it, which is a function of ѱ t-1. Then, the relationship among the serial correlation of consumption, risk aversion and serial correlation of asset returns implied by the first order condition can be written as, ( ) (5) Where V it is derived as V it U it - µ it-1 and denotes the serially uncorrelated error on the joint distribution of asset returns, risk aversion and consumption. It follows that E( ) = 0 and E = ( ), i = 1,..., n (6) Where, E and E[LCG t\ ѱ t-1)] are the expected log of asset return and consumption growth respectively, based on the information set ѱ t-1. Equation (6) shows that variability in stock returns depends on variability in consumption growth by a factor that measures the level of consumers aversion to risk. Empirical methodology To specify the empirical model we consider the nonlinear and linear forms given by equations (3) and (6) respectively in the economic model above. That is, E t*β ( ) +=1 (7) E = ( ), i = 1,..., n (8) The nonlinear specification implicitly expresses the joint distribution of stock returns and consumption growth, and the linear specification describes the conditions of stock returns dependence on consumption growth. We adopt the General Method of Moments, GMM econometric method to estimate both equations. The use of GMM is common in consumption-based studies of risk aversion as it provides consistent results. It also avoids the need to specify either the distributions of the endogenous variables of the model, or of instrumenting variables. Furthermore, the results based on this method are still consistent in conditions where the disturbances are conditionally heteroscedastic and serially correlated (Hansen and Singleton, 1982). We estimate equations (7)and (8), using the traditional approach replacing the expectations of Cg t+1, Cg t, SR t+1 and SR t with the instrumented actual values. We initially, adopt the common practice in GMM studies of risk aversion which follows Hansen and Singleton (1982) and employ lags of the endogenous variables of the model as the instruments. However, we rely on Hansen and Singleton (1982) and assume that different subsets of variables that are in agents information set at time period t and (t-1) are equally candidate instruments, specify an alternative set of instruments in which we include current and lagged values of changes in broad money demand (DM2) as proxy variable for monetary policy condition. This enables us to address the concern of the study that conditions of monetary and macroeconomic variables over the period studied may affect stock returns via agents aversion to consumption risk. Our empirical analysis thus involves obtaining estimates of equations (7) and (8) by applying the GMM method and using two different set of instruments, Z1 and Z2, where Z1 include N-lags of the endogenous variables, consumption growth and

7 stock returns, and Z2 also includes lags of change in broad money, DM2. The magnitude of the estimates of, the coefficient of risk aversion, and its statistical significance in stock returns variation yields findings on the extent of risk aversion and its role in stock returns variations over the short run. Interpretation of the measure of risk aversion is based on the absolute value of, where a value of = 0 implies risk neutrality on the part of investors. On the other hand, > 1 implies that investors exhibit risk aversion, with higher values representing greater risk aversion. We rely on the Hansen J-statistic (Hansen, 1982) for model diagnostic as is commonly adopted in consumption based studies. The Statistic uses over-identifying restrictions to test the validity of the instrument set and hence the validity of the model specification. Its computed probability values (pvalues) indicate the probability of rejecting the null hypothesis that the instrument sets are valid. Moreover, in addition to the conventional use of Hansen J-Statistic and t-values we also employ the Durbin Watson statistic test for serial correlation and the Jarque-Bera tests for normality. While the existence of serial correlation is unimportant for the validity of Nonlinear GMM estimates (Hansen and Singleton, 1982), this is not the case for the linear model given the imposition of distributional assumption of lognormality. Data sources and measurement Quarterly and annual Nigerian data from 1985 to 2015 on household consumption and consumer price index, CPI, Broad money (M2) as well as the Treasury bill rate are obtained from Central Bank of Nigeria Statistical Bulletin (various issues). In addition, data on the NSE All share index are also obtained from the SEC Bulletin and the NSE fact books. Real aggregate consumption was derived by deflating nominal household consumption data CPI. We compute stock returns using the one period return measured as ((ASI t/asi r-1) + 1) We use both quarterly and annual data in other to overcome data limitations arising from the non availability of quarterly data on consumption for Nigeria before We used quarterly data for the period 2007 to 2015 and used annual data for the entire sample period. Presentation and analysis of estimated results We now present the results from estimations based on quarterly and annual time series. We estimated the models using various lag lengths (N-lags) specification, but we present results for N-lags equal one and two for each instrument set Z1 and Z2. This is because across the models using N-lags higher than two resulted in rejections of the specifications by all model diagnostics. Results based on quarterly data We present in Table 2, results derived from model specifications involving only lags of the endogenous variables as instruments. The computed p-values for the Hansen J-Statistic are highly statistically significant and indicate rejection of the restrictions imposed on the models except for the Linear GMM specified using 1-Lag of the instruments. But across all model specifications, the t-values show very large standard errors for the computed coefficients. In particular, the estimate of, the coefficient of risk aversion from the linear GMM model with 1-Lag specification equals , but it is not statistically significant given the computed t-value of The implication is that the specifications of the models employed here are not valid for estimating the level of investors risk aversion. These results are different from the nonlinear GMM estimates of Hansen and Singleton (1983). In Table 3, we present the results when we use 1-lag and 2-lags of the instrument set Z2 in which we include change in broad money along with the lagged endogenous variables. We first consider the appropriateness of the instrument sets and thus the validity of the model specification using the Hansen J-statistic. The computed statistics are also highly statistically significant across models and estimation methods based on their p-values implying incorrect model specifications, similar to the results we derived from using only lags of the endogenous variables as instruments. These results are only with exception in the case of the linear GMM with 1-lag specification for which the Hansen J-Statistic at has a calculated p-value of supporting acceptance of the restrictions and the model specification. However, the estimate of (0.250) has a t-value of which is statistically insignificant at any conventional level of significance. The key information in this set of results in the light of previous literature is that a small lag length seems to characterise the relationship.

8 Table 2: Results from GMM estimation of the linear and non-linear models of Risk aversion using lagged values of the endogenous variables. Model Non-linear GMM Linear GMM N-lags Β R 2 DF Hansen J- Statistic NA [1.83E-08] [0.000] [0.717] lag (76.030)*** (-0.086) 2-lag (48.299)*** (1.100) 1-lag (2.850)*** (-0.094) 2-lag (174.97)*** (1.307) [0.000] Notes: Source: Author s computations using Eviews-7 econometric software Figures in parenthesis are t-values and those in brackets are computed p-values. (*), (**), (***) = significant at the 10%, 5% and 1% levels of significance respectively. (i) and (ii) are respectively the Durbin-Watson and the Jarque-Bera Statistics. (i) DW Statistic (ii) Tests Normality [0.0012] [0.553] [0.957] [0.395] for Table 3: Results from GMM estimation of the linear and non-linear models of Risk aversion using current and lagged Values of the endogenous variables and DM2. Model N-lags Β R 2 DF Hansen J- Statistic NA [3.34E-05] NA [0.000] (0.878) (i) DW Statistic 1-lag Nonlinear (56.399)*** (-1.453)** GMM ( )*** (-1.177) Linear GMM (2.882)*** (-0.125) ( )*** (0.359)* [0.000] Notes: Source: Author s computations using Eviews-7 econometric software Figures in parenthesis are t-values and those in brackets are computed p-values. (ii) Tests Normality [0.761] [0.0045] [0.988] ] for (*), (**), (***) = significant at the 10%, 5% and 1% levels of significance respectively. (i) and (ii) are respectively the Durbin-Watson and the Jarque-Bera Statistics.

9 Results based on annual data Table 4: Results from GMM estimation of the linear model of Risk aversion Panel A. Results from Linear GMM models of Risk aversion using lagged values of the endogenous variables. N- Hansen J- (i) DW (ii) Test for Model Lags Β R 2 Statistic Statistic Normality 1-lag (15.99)*** (-2.351)** [0.155] [0.229] Linear GMM (22.331)*** (-3.592)** [1.64E-13] [0.258] Panel B. Results from Linear GMM models of Risk aversion using lagged Values of the endogenous variables and DM2. Linear GMM (19.092)*** (-0.858) (0.021) [0.888] Model (22.85)*** (-2.358)*** [0.000] 8 [0.673] Notes: Source: Author s computations using Eviews-7 econometric software Figures in parenthesis are t-values and those in brackets are computed p-values. (*), (**), (***) = significant at the 10%, 5% and 1% levels of significance respectively. (i) and (ii) are respectively the Durbin-Watson and the Jarque-Bera Statistics.

10 This study further present findings based on annual time series covering form , the entire sample period. Results from the nonlinear GMM were uniformly poor and similar in pattern to the those based on quarterly time series reported above. We therefore present only the results for the linear GMM., based on instrument sets Z1 and Z2 respectively in panels A and B of Table 4. A key finding from these results is that the Hansen J-statistic shows more support for the restrictions on the models across instrument sets using the 1- lag rather than the 2-lag specification of the linear GMM. Specifically, the Hansen J-statistic p-value of 0.16 and 0.02 for the 1-lag specification, are each larger than the 1.64E-13 and 0.00 computed for the 2-lags specification. The results also suggest that the measure of risk aversion is 0.63, based on 1-lag of instrument set Z1. The estimated measure is statistically significant at the 1% level. This finding of a level of risk aversion below one is in accordance with the findings of Hansen and Singleton (1982) and Garcia et al (2003), but it is lower than the value of 1.25 estimated by Hansen and Singleton (1983). The low value of risk aversion we have estimated may be due to the limitations in the stock return data. As earlier discussed for part of the study period, stock prices were not allowed to vary freely in response to investors demand but were outcomes of administratively fixed stock prices, which may affect the extent to which stock prices reflect investors risk attitude. The Jarque-Bera normality test shows that the residuals are normally distributed, and the DW statistic shows absence of serial correlation indicating that we can rely on the estimate value. Two other key information follow from the results in Table 4. First, it is difficult to interpret the highly statistically significant estimate of to mean that investors aversion to consumption variability explains expected stock returns variation. This position arises from the fact that the model with 1- lag of instrument set Z1 has a negative R 2 an occurrence usually observed in models with overidentifying restrictions. Second, monetary and macroeconomic conditions do not appear to inform the level of risk aversion as based on either 1-lag or 2-lags, we cannot accept validity of the specification of the model using instrument set Z2 (panel B). Policy implications Investors in Nigerian stock market on the aggregate appear to be risk averse based on stock response to consumption growth. The measured value of risk aversion is 0.63 for the Nigerian economy. This implies a weak risk aversion for Nigerian investors/consumers which further implies that the effect of changes in risk on investors demand for stocks and for risky assets generally, is small. Monetary policy and macroeconomic conditions do not appear to predict the dependence of expected stock returns on expected consumption growth through investors risk aversion. Conclusions This study analysed the extent of risk aversion exhibited on the aggregate by investors in Nigerian stock market and was directed at establishing for the developing Nigerian economy, the nature of risk aversion that characterize the behavior of investors/consumers. Based on the single factor consumption-based model of asset return which we estimated using the GMM estimation technique, we found an estimate of risk aversion of We conclude, based on the results from the estimation exercise that for the Nigerian economy, the response of stock returns to consumption growth supports that investors are risk averse but that conditions of monetary policy do not seem to affect the stock market, via the mechanism of investors risk aversion. We also conclude that the measure of the level of risk aversion appears to be 0.63 for the Nigerian economy. References Ait-Sahalia, Y. & Lo, A. W. (2000), Non-parametric risk management and implied risk aversion. Journal of Economtrics, 94, Bernanke, S. B. & Kuttner, K. N. (2005). What explains the stock market reaction to Federal Reserve policy? Journal of Finance, 60 (3), Bliss, R. & Panigirtzoglou, N. (2004). Option implied risk aversion estimates. Journal of Finance, 59 (1), Campbell J. Y. & Cochrane, J. H. (1999). By force of habit: A consumption-based explanation of aggregate stock market behaviour. Journal of Political Economy, 107 (2),

11 Cook, T. & Hahn, T. (1989). The effect of changes in the federal fund rate target on market interest rates in the 1970s. Journal of Monetary Economics, 24 (3), Epstein, L. G. & Zin, S. E. (1989). Substitution, risk aversion, and the temporal behavior of consumption and asset returns: A theoretical framework. Econometrica, 57, Epstein, L. G. & Zin, S. E. (1991). Substitution, risk aversion, and the temporal behavior of consumption and asset returns: An empirical analysis. Journal of Political Economy, 99 (2), Garcia, R., Luger, R. & Renault, E. (2003). Empirical assessment of an intertemporal option pricing model with latent variables. Journal of Econometrics, 116, Grossman, S. J., & Shiller, R. J. (1981). The determinants of the variability of stock market prices. American Economic Review, 71 (2), Hansen, P. L. (1982). Large Sample Properties of generalized method of moments estimators. The Econometric Society, 50 (4), Hansen P. L. & Singleton, J. K. (1982). Generalized instrumental variables estimation of non-linear rational expectation models. Econometrica, 50, Hansen P. L. & Singleton, J. K. (1983). Stochastic consumption, risk aversion, and the temporal behaviour of asset returns. Journal of Political Economy, 91(21), Hunter J. & WU, F. (2009). A multifactor consumption based asset pricing model of the UK stock market: The US stock market as a wealth reference. Paper presented at the EEPS conference in Warsaw, 2009 and the Infiniti Conference in Dublin, 2010 and Brunel University Working Paper 09, 01. Ibeawuchi, S. N. (2007). Overview of monetary policy in Nigeria. Central Bank of Nigeria Economic and Financial Review, 45 (4), Jackwerth, J. C. (2000). Recovering risk aversion from option prices and realized returns. The Review of Financial Studies, 13 (2), Ludvigson, S., Steindel, C. & Lettau, M. (2002). Monetary policy transmission through the consumption wealth Channel. Economic Policy Review, 8 (1), Mankiw, N. & Zeldes, S. (1991). Consumption of stockholders and non stockholders. Journal of Financial Economics, 29 (1), Meristem Securities (2008). The Nigerian stock market: Gauging the fundamentals. Accessed on Mogbolu, F. (2010). Stock price response to monetary policy in Nigeria: Analysis of sectoral stock returns. Finance and Banking Review, 4 (1and 2), Mogbolu, F. (2014). A review of monetary policy reforms: Effects on financial markets and economic performance in Nigeria. Nigerian Journal of Business Administration, 12 (1&2), Nnaji, O. & Ohuche, F. K. (2008). The stock market and monetary policy transmission mechanisms. Central Bank of Nigeria, Economic and Financial Review, 46 (3), Okpara, G. C. (2010). Monetary policy and stock market return: Evidence from Nigeria. Journal of Economics, 3(1), Osisanwo, B. G. & Atanda, A.A. (2012). Determinants of stock market returns in Nigeria: A time series analysis. African Journal of Scientific Research, 9 (1), Pindyck, R. (1984). Risk, inflation, and the stock market. American Economic Review, 74, (June), Porteba, M. J. Samwick, A. & Shiller, R. J. (1995). Stock ownership patterns, stock market fluctuations and consumption. Brookings

12 Paper on Economic Activity, 1995 (2), Romer, C. D. & Romer, D. H. (2000). Federal Reserve information and the behaviour of interest rates. American Economic Review, 90, Rosenberg, J. & Engle R. (2002). Empirical pricing kernels. Journal of Financial Economics 64, Vickers, J. (2000). Monetary policy and asset prices. The Manchester School Supplememt, , 1-22.