Research On Effects Of FOMC Releases On Stock Prices In The Quantitative Easing Era

The Federal Reserve’s explicit commitment to supporting the stock market and the underlying economy through the wealth effect has been exemplified over the past few years with its Large Scale Asset Purchases (LSAPs), most notably Quantitative Easing (QE) programs. Stock markets have rallied significantly as several rounds of QE were rolled out in regular intervals beginning in2009. By "Quantitative Easing" we mean to describe a type of unconventional monetary policy where the central bank buys a certain amount of financial assets from other parties, mostly commercial banks and other private institutions. This contrasts with conventional monetary policy, which generally only involves the buying and selling of government bonds.The sheer size of the Federal Reserve’s QE programs, involving purchases of billions of dollar in value, appears to have had a distinct effect on the American stock market, with U.S.-based exchanges near their all-time highs most recently. The booming stock market contrasts somewhat sharply with the real American economy, which is still not entirely up to speed. This leads us to question the extent to which the stock market has been influenced by the Federal Reserve’s initiatives. More importantly, how can we expect this market to behave once the Federal Reserve starts tapering, i.e. winding down its asset purchases? The chance of a severe stock market downturn and the associated systemic risk make this a valid question to seek answer to.One way to approximate such a hypothetical situation is to take releases including statements and minutes from the Federal Open Market Committee (FOMC) and measure the markets’reaction to them. Such a method would expose investor psychology to an extent sufficient enough to estimate the future actual market response to tapering. This thesis therefore took FOMC releases and categorized all of them according to the their outlook. Three categories were designed to represent the Federal Reserve’s QE outlook:"Dovish"(willing to expand QE programs),"Hawkish"(wanting to reduce or wind down QE programs), and "Unchanged"(no significant changes in outlook from the previous release). Only a few preceding studies have investigated the impact of FOMC releases on asset prices, most notably Farka and Fleissig (2010) for FOMC statements and Rosa (2013) for FOMC minutes. This thesis’approach partly draws on the one proposed by Bernanke and Kuttner (2005).The stock market used for analysis was the S&P500index since it is widely perceived to offer an accurate view of the general American economy. The data featured daily percent changes in the stock market. Two hypotheses played a primary role:(1) there is a sizable and significant relationship between the outlook of FOMC release and stock market fluctuations, and (2) the responses of investors to Dovish and Hawkish statements are opposites, more specifically they respond positively to the prospect of more liquidity (Dovish releases) and negatively to less (Hawkish releases).The FOMC releases in question dated from between the18th of March2009and the19th of March2014. The number of observations (total number of categorized FOMC statements and minutes) was81. However there were only80observations for S&P500fluctuations because one FOMC release happened on a day when markets were closed. Each of these observations was assigned an outlook (Dovish, Hawkish or Unchanged) and then employed as an indicator variable in linear regression analysis (STATA). The outlook was determined by a combination of independent analysis and reporting by major newspapers, mainly the Financial Times and Wall Street Journal. The resulting binary variables were regressed onto fluctuations in the S&P500to assess the relationship between FOMC releases and stock market fluctuations. An ARCH model to better capture stock price fluctuations and take into account conditional heteroskedasticity was considered but could not be implemented. Due to the large amount of gaps in the time series data, stemming from the exclusion of all dates that did not feature FOMC releases, this type of model was rendered inefficient.Fluctuations for three different time periods were investigated:(1) S&P500percent changes on the release day itself;(2) The release day as well as the following day;(3) Before, on and after release days to provide a full picture of the state of the stock market around the release. An approach featuring multiple time periods was chosen because there is no consensus in academic literature as to the amount of time it takes for stocks to fully react to changes in monetary policy.Pre-analysis descriptive statistics indicated that stocks generally responded positively to Dovish FOMC releases (with a mean of.154) and negatively to Hawkish releases (with a mean of-.18). This appeared to support the second hypothesis:the prospect of more liquidity causes stocks to rally, while a future reduction of liquidity depressed them. However the model suffered from a number of important flaws. Firstly, it did not take into account expectations. Whereas Bernanke and Kuttner (2005) could use future contracts to estimate expectations regarding changes in the Federal funds target, here no such tool was available. If the surprise element is unaccounted for then the reaction of stock prices to policy changes cannot be accurately measured. Second, the data used was of relatively low frequency (daily). Intraday data would have given us a much more precise idea of the power of FOMC releases over stock prices. While this second problem could not be solved efficiently the first was mitigated as much as possible by employing Volatility Index (VIX) data to single out releases that could distort the data.9out of80observations showed double-digit percentage changes in the VIX. Of these9, three observations convincingly demonstrated a disconnect between expectations and actual policy decisions that led to volatility. Unfortunately these observations featured releases that announced important changes in QE policy and so could not be discarded.Concerning results, the one-day model showed that Hawkish FOMC statements, whose binary variable was significant at the10%level, have an effect of approximately-0.45%on the S&P500on the day of the release. However the Dovish binary variable was insignificant and also featured a negative coefficient instead of the positive one we expected. The Unchanged variable was insignificant too, with the coefficient indicating a positive effect.The two-day model, taking into account the day of the release and the day after that, featured an insignificant Hawkish variable with a negative coefficient of slightly larger magnitude. However the Dovish variable was significant and featured a negative coefficient of greater magnitude than its hawkish counterpart. According to these results, dovish FOMC statements have a significant and negative effect of approximately-1%on S&P500stocks, spanning two days. The Unchanged variable was highly significant too and appeared to indicate that over the past five years FOMC statements that did not change particularly much in outlook had a positive effect of1.4%on the S&P500over two days. The three-day model that took into account fluctuations on the day of the release and those around it showed the Hawkish variable to have no significant impact at all on stock price fluctuations, while the Dovish variable was still significant at the10%level with a slightly smaller, negative effect. Similar changes took pace for the Unchanged variable:a slight loss in significance (but still within10%) and a slightly smaller coefficient.The results showed that the two-day model was the most successful, featuring the highest significance and greatest coefficients for two out of three variables in the analysis (Dovish and Unchanged). There was no time period model that featured both significant Dovish and Hawkish variables. The first hypothesis was only partially confirmed:the analysis discovered no model in which all three variables were significant, even though each of them was significant in in at least one model.The second hypothesis had to be rejected, since the results showed stock price responses to Dovish and Hawkish statements were not opposites. While the one-day model appeared to indicate investors do indeed respond negatively to the prospect of less liquidity, the second and third model indicated that investors also responded negatively to the prospect of more liquidity. According to the results the only thing that investors responded truly positively to was stable policy, as represented by the Unchanged variable. One possible explanation for the discrepancies in results between the three time period models is that while fear of tapering is quick to take a hold of markets, the prospect of stable policy or increased liquidity takes longer to manifest itself.The thesis concluded that there appears to be an important relationship between FOMC releases and fluctuations in the S&P500, but more research is needed to fully understand its mechanisms. While Hawkish statements led to a0.4%average decrease in the S&P500index on release days, Dovish and Unchanged statements appeared to need more time to take effect, reaching significance only when the day after the release day was included, influencing the index by-1%and1.4%on average, respectively. The negative coefficient for Dovish statements was particularly noteworthy. The thesis also concluded that three-day models including the day before the release were not effective. The need for more advanced research was bolstered by the difficulty of accounting for expectations and the absence of intraday data. Finally, the analysis results appeared to indicate that the chance of a severe stock market downturn are limited as long as the Federal Reserve chooses to employ gradual tapering.