Do Federal Reserves Tapering Announcements Hit the US Mortgage Reits Stock Return - Research Proposal Example

Do Federal Reserve’s tapering announcements hit US mortgage REITs stock return? In this paper, the researcher tested the effect of the Federal Reserve’s QE3 tapering announcement on US mortgage REITs stock return. Mortgage REITs stocks have become a popular asset due to their high dividend payout, while they are extremely volatile to the monetary shock in terms of decrease in money supply and increase in interest rates as implemented by the Federal Reserve. Some may question whether investing in mortgage REITs is still a good option to hold a real estate asset, as they are concerned about the fact that the tapering of the Federal Reserve’s bond purchase programme that aims to decrease the money supply would hit mortgage REITs stock returns. This paper employed daily return of iShare FTSE NAREIT mortgage REITs index ETF as a proxy of Mortgage REITs returns. Furthermore, S&P 500 was used as market return and 10-year government bond yield represented long-term interest rates. The daily price for REITs index and market index was collected from Yahoo Finance. The data for long-term government bond yield and treasury bills was collected from the website of the US Department of Treasury. The limitation of data existed in terms of choices of portfolio, index, and event dates. Two methods were employed in this project to estimate the announcement’s effect on mortgage REITs stock return. The two-step event study approach estimated the cumulative abnormal return of mortgage REITs stock during event periods, whereas the multivariate regression estimated the coefficient of event dates in relation to stock returns. Finally, the results for both methods showed that there is no significant impact of tapering announcements on mortgage REITs stock returns. This may remind investors that it is still worthy to hold stocks in mortgage REITs stocks during tapering periods. 1. Introduction Mortgage real-estate investment trusts have been considered as an important source to raise capital in the real estate sector. Unlike equity REITs acquiring shopping malls or office buildings, mortgage REITs invest in mortgage and mortgage backed securities and receive interest on them as revenue. Mortgage REITs are sensitive to the changes in the monetary policy as implemented by the Federal Reserve. In May 2013, the Federal Reserve decided to taper its $85 billion monthly bond purchase programme of agency mortgage-backed securities—QE3. Investors, who purchased shares of individual stock exchange-listed mortgage REITs or in a mortgage REIT exchange-traded fund, concerned that the QE3 tapering news could lower the stock price of this type of security and deteriorate their returns. In theory, the substantial amount of bond purchase boosts the price of agency mortgage-backed securities and increase mortgage REITs’ book values. When the Federal Reserve decides to taper the amount of bond purchase, the price of financial assets falls. Furthermore, a tightening monetary policy is likely to accompany the rise in interest rate to curb inflation. This forces the price of agency mortgage-backed securities to decrease due to the potential default risks of holding the assets for long term. Investors may rush to sell off their holdings of mortgage REITs stocks, which would deteriorate the stock returns. It may take time for investors to absorb full information, and hence the stock return may exhibit a downward trend after the announcement is made. This violates the Efficient Market Hypothesis, which states that the stock prices will accurately reflect all the information immediately. In this case, no abnormal return can be earned. However, on the other hand, behaviour finance theory indicates that the market may expect the news as the result of information leakage, and hence the stock prices may adjust prior to the announcement date. Furthermore, investors may expect that mortgage REITs have already implemented the hedging strategies against the tightening monetary policy. Thus, the stock prices may adjust to the public information rapidly even though it may exhibit a modest downward trend before the announcement. This supports the Efficient Market Hypothesis, which states that the stock price should reflect all public available information quickly. The question is whether the QE3 tapering announcements really have a significant negative impact on the stock returns of US mortgage REITs? This paper expects that semi-strong form market efficiency exists due to the market sentiment relating to information release and short-run announcement effects are tested rather than conducting a time-series analysis. Another objective is to provide a perspective to investors regarding the way negative monetary shock would affect stock prices of mortgage REITs in the short run. 2. Literature Review Though there is little literature examining the effects of QE3 tapering announcements on mortgage REITs stock returns, numerous literature estimates the relationship between time-varying change in monetary policy and REITs returns. Lutz and Gabriel (2015) applied the auto-regression model and found that expansionary unanticipated monetary shocks can increase the equity market return of REITs and lower the interest rate in the housing market. Johnson and Jensen (1999) found that restrictive monetary policy has statistically significant association with four types of NAREIT indices, which include equity, mortgage, hybrid, and composite REITs. On the other hand, some studies have been focused on the fundamental principle of Efficient Market Hypothesis to examine whether stock prices reflect all available information or not. For example, Rai and Suchanek (2014) applied event study approach to study the effect of the Federal Reserve’s tapering announcements on emerging markets and showed that the reaction of emerging market on actual tapering announcement in December 2013 was muted as the markets had expected the announcement. Similarly, Keown and Pinkerton (1981) calculated the cumulative abnormal returns for companies around announcement dates of takeover event and found that there was an evidence of information leakage that generate pre-announcement effects. Rendleman, Jones and Latane (1982) applied the similar methodology to study the earning effects on companies and found a post-announcement drift in stock prices. 3. Data 3.1 Data Description This paper explores the effect of the Federal Reserve’s tapering news on stock returns of mortgage REITs through two methods: the two-step event study approach and the multivariate regression analysis. The use of both models and their results will be analyzed in the later section. To analyze the announcement effects, the prior task is to identify the specific event dates. It is almost impossible to use a single event date to capture the impact of tapering, as the information related to tapering decision flows continuously. This paper adopts the method prescribed by Rai and Suchanek (2014) to identify event dates using days when Ben Bernanke, the former chairman of the Federal Reserve, or the FOMC statements reveal the likelihood of tapering. A summary of event dates and their related description is provided in the following table. Table 1 For both event study and multivariate regression methods, the stock returns of mortgage REITs are proxied by the daily returns of iShare FTSE NAREIT Mortgage Real Estate Capped Index Exchange-Traded Fund (REM), which composed 38 U.S. REITs that hold residential and commercial mortgage. The S&P 500 serves as a market index to capture the effect generated by the overall stock market. The daily prices of both indices can be collected from Yahoo Finance. For event study approach specifically, excess returns of both indices are calculated and the three-month US Treasury Bill is used as the risk-free rate. For multivariate regression, 10-year government bond yield is employed to capture the effect caused by change in mortgage rate to Mortgage REITs returns, as the mortgage rate in housing market has a very close link with the 10-year government bond. The time periods for both methods are from January 2013 to January 2014. 3.2 Data Critique There are limitations of data used in this paper. First, this paper used a market capitalization-weighted fund as a proxy of mortgage REITs. In mortgage REITs sector, some companies, such as, Annaly Capital Management REIT and American Capital Agency REIT have relatively larger concentration ratio. Hence, their stock returns are likely to affect the whole portfolio return by a large degree. Instead, an equally weighted portfolio, which ensures the diversification of all stocks, may be considered as another measure. However, it is also skeptical that the equally weighted portfolio would well represent the whole Mortgage REITs sector as companies’ sizes do affect the overall return for this sector. In terms of index choice, while iShare FTSE NAREIT mortgage REITs index ETF excels in diversifying its portfolio and gaining broad exposure to US mortgage REITs sector, it is showed that the index has a large tracking error that it seeks to track the FTSE NAREIT All Mortgage Capped Index as a result of iShare index’s aggressive diversification. Thus, it is more likely that iShare index could generate profits or losses unexpectedly, which could distort the analysis of the announcement’s effects. Alternatively, the relatively passive ETF, Market Vector Mortgage REIT Income EFT, which seeks to track the Market Vector Global Mortgage REITs Index, could be a proxy of daily return of mortgage REITs. However, the choice is trading off between the exposure to mortgage REITs sector and standard error of the exchange-traded fund. Lastly, the choice of event dates could be qualitative. As mentioned previously, due to the continuous nature of information flow, the event dates are difficult to be identified. Any news that signals the QE3 tapering could generate effects on mortgage REITs stock returns. However, due to the complexity of collecting such kind of information, there are only four event dates identified in this paper. 4. Methodology 4.1 Event Studies Methodology As explained in the previous section, the first methodology applied in this paper is the traditional two-step event study. The market model is used in the first step to calculate abnormal returns. Before going through the mechanism behind, the reasons for using this model are explained here. A number of approaches are available to calculate the normal return of stocks that induce abnormal return. The constant mean return model indicates that the normal return of stock will be induced by its mean return over a time period. Brown and Warner (1985) concluded that this model could yield results similar to those generated by the more sophisticated models. On the other hand, the market model calculates the return of security in relation to the returns of market portfolio with assumption of joint normality of asset returns. The model specification is described in Table 2. Constant Return Model Market Model Normal Return Rᵢt=µᵢ + ζᵢt Rᵢt = αᵢ + βᵢRmt+ εᵢt Expectation of Disturbance Term E(ζᵢt) = 0 E(εᵢt) = 0 Variance of Disturbance Term var(ζᵢt) = σ2ζᵢ var(εᵢt) = σ2εᵢ Abnormal Return ARᵢt = Rᵢt - µᵢ ARᵢt= Rᵢt– (αᵢ + βᵢRmt) Note: µᵢ is the mean value of stock returns; ζᵢtand εᵢtrepresent disturbance term Table 2 The market model can be considered as a more appropriate model in this study. The constant mean return model is unlikely to capture the normal return if there is a substantial variation in return during the estimation period. One may argue that a multi-factor model could be a more favourable solution. However, the empirical studies shows that there is limited gain from using multi-factor models for event studies, as the marginal explanatory power of additional variables is too small to reduce the variance of abnormal returns. The first step of the event study methodology is to define the event window and the estimation window. This paper defines each event date as an event window and each event window contains 21 trading days (-10 to +10 of event date). Estimation window is defined from -120 to -11 trading days of the first event date (May 22, 2013) in order to provide estimators for the parameters of the normal return model. Post-event date ends at around the 30th trading day following the last day of the event window. Source: Mackinlay (1997) Figure 1 The second step is to run regression for data in the estimation window (-120 to -11 day) to get estimates of parameters (αᵢ, βᵢ) of the normal return model. The normal return can simply be calculated as the percentage change in the daily price. Then, abnormal returns in the event window can be induced. As explained above, the market model is employed here: ARᵢt= Rᵢt– (αᵢ + βᵢRmt) Then, the cumulative abnormal return for the index is aggregated from the abnormal return. CAR(t1, t2) = t To test whether the announcement has price impact on mortgage REITs, this paper examines the cumulative abnormal return by setting the null hypothesis: H0 :CAR(t1, t2) = 0. The H0 can be tested using t statistics, which is calculated as following: t = Then, CAR for each event window is plotted to test whether stock prices incorporate immediately into new information. 4.2 Multivariate Regression Model This paper employs a multivariate regression model to solve the problem of non-announcement effect in the event period. Some literature employs autoregressive conditional heteroskedasticity (ARCH) model to study the effect of monetary policy on REITs returns over long horizon. While the models can efficiently capture the way the volatility of stock returns varies in respond to the positive and negative news using time series analysis and examine the pre-announcement effect, they are more suitable for a longer horizon with a significant numbers of event dates. Thus, this study uses an alternative method, the multivariate regression, which is similar to approach of Rai and Suchamek (2014). The time period is from 4th January 2013 to 21st January 2014. Considering the likelihood of information leakage and the time for adoption of announcements, this method identifies that each event contains five days (-2,-1, 0, 1, 2 day of the event). The equation and model specification are given below: rit = α + βrmt + γbt + δrit-1 + inMnt + ε The rit is the percentage daily return on the mortgage REITs stocks and is similar to the two-step event study approach. The return is proxied by iShare FTSE NAREIT Mortgage REITs Capped Index ETF. The rmt is the daily return on market portfolio, proxied by S&P 500 index. The rit-1 is the one-day lagged return that captures the stock autocorrelation effect. The bt is long-term bond yield proxied by the 10-year government treasury yield to capture the change in mortgage rates that could affect the book value of mortgage REITs. The Mnt is a vector of dummy variables for each of four events and is equal to 1/5 for each day in the five-day event of the event and ‘0’ otherwise. The ε is the error term. The tapering announcement effect is estimated as the coefficient of dummy variables corresponding to a specific event date. The hypothesis is that the tapering announcements have no impact on the mortgage REITs returns. This hypothesis can only be rejected if the coefficient of the dummy variable is significantly greater than zero. 5. Analysis and Result 5.1 Two-step event study approach The result is shown in the table 3 to conclude whether the announcements have price impact on mortgage REITs or not. The first two events show around 9% decrease in the cumulative abnormal return, while the cumulative abnormal returns in the rest two events do not deviate from zero a lot. Not surprisingly, the cumulative abnormal returns for the four events are not significantly greater than zero at 1%, 5%, and 10% significant level. Hence, the null hypothesis can be not rejected, which means that the tapering news do not significantly hit the mortgage REITs stock returns. 22-May 2013 19-June 2013 18-September 2013 18-December 2013 Day AR CAR AR CAR AR CAR AR CAR -10 0.00676 0.00676 -0.00175 -0.00175 0.00154 0.00154 0.00287 0.00287 -9 -0.00351 0.00324 0.01499 0.01325 -0.00995 -0.00841 -0.00529 -0.00242 -8 -0.01481 -0.01157 -0.02253 -0.00929 0.01439 0.00598 -0.00926 -0.01169 -7 -0.01696 -0.02853 -0.00980 -0.01909 0.00002 0.00600 0.01460 0.00292 -6 -0.01156 -0.04009 -0.00615 -0.02523 -0.01778 -0.01178 0.01381 0.01673 -5 -0.00307 -0.04316 -0.00844 -0.03368 0.00782 -0.00396 0.00529 0.02201 -4 0.01465 -0.02851 0.03665 0.00297 0.00435 0.00039 -0.00131 0.02071 -3 -0.00237 -0.03088 0.00243 0.00540 0.00796 0.00835 0.00108 0.02178 -2 -0.01317 -0.04405 -0.02009 -0.01469 -0.00255 0.00580 -0.00983 0.01195 -1 -0.01320 -0.05726 0.00305 -0.01164 -0.00225 0.00356 0.01810 0.03005 0 -0.00848 -0.06573 -0.01706 -0.02870 0.02996 0.03352 -0.00138 0.02866 1 0.00167 -0.06407 -0.01539 -0.04409 -0.01178 0.02174 -0.00956 0.01910 2 -0.00426 -0.06833 0.00199 -0.04209 -0.01106 0.01068 0.01355 0.03265 3 -0.03713 -0.10546 -0.01773 -0.05982 0.01178 0.02246 -0.03829 -0.00564 4 0.00833 -0.09712 0.01201 -0.04781 -0.03152 -0.00906 -0.00294 -0.00859 5 -0.00336 -0.10048 -0.03522 -0.08302 0.00635 -0.00271 -0.00346 -0.01205 6 -0.01110 -0.11158 0.01229 -0.07073 -0.00498 -0.00769 -0.00309 -0.01514 7 -0.01311 -0.12468 0.00260 -0.06814 0.00157 -0.00612 -0.00322 -0.01836 8 0.01754 -0.10715 -0.00633 -0.07447 -0.00192 -0.00804 0.00589 -0.01247 9 -0.00175 -0.10889 -0.00737 -0.08184 0.00491 -0.00313 0.01380 0.00133 10 0.01499 -0.09390 -0.01086 -0.09270 -0.00266 -0.00579 0.00642 0.00775 Table 3 To see whether stock prices incorporate immediately into any new information, the cumulative abnormal return for each event window is plotted below: Figure 2 Figure 3 Figure 4 Figure 5 It can be seen from the figure 2 that the CAR decreased prior to the event date consistently and the trend continued after the announcement. This is a post-announcement drift driven by investors’ under-reaction to the new information. However, a conclusion that EMH is violated cannot be simply reached as the hypothesis is not rejected statistically. The graph for the second event illustrates a similar pattern as the first event but the magnitude of downward trend is slightly smaller. The decrease in CAR prior to the event dates can be explained by information leakage or investors’ anticipation for the announcements and this does not violate the semi-strong efficiency of EMH theory. The third event in September shows an increase in CAR one day prior to the announcement. However, the CAR decreases sharply due to a substantial decline in the abnormal return on the announcement day (Day 0) and does not deviate a lot in the following days. The decrease in the abnormal return can be explained by investors’ concerns of rising interest rate after the FOMC announcement. On the actual tapering date, 18th December 2013, there is not much variation in mortgage REITs stock return despite a decline observed on the third day after the announcement. This may indicate that the investors have anticipated the tapering announcement at the last meeting of FOMC in 2013. 5.2 The Multivariate Regression Tapering Dummies Control Variables May-22 Jun-19 Sep-18 Dec-18 rm b rt-1 R2 Observation -0.0274 0.0022 -0.0068 0.0152 0.8370*** -0.3166*** -0.1189** 0.4328 262 (0.2004) (0.9217) (0.7514) (0.4766) (0.0000) (0.0000) (0.0146) Note: *** indicates p < 0.01; ** indicates p < 0.05 Table 4 The result of multivariate regression (table 4) is similar to that of the two-step event study approach. The coefficients of dummy variables measure the tapering announcements. From the graph, it can be seen that tapering news had led to 2% decrease in mortgage REITs on 22 May 2013. The coefficients for the second and third events do not deviate from zero a lot. The actual tapering announcement on 18 December 2013 even brought a 1.5% increase in the mortgage REITs. This can probably be explained by the Federal Reserve’s claim that interest rate would remain low at the current level. However, similar to the previous approach, the four coefficients for dummy variables are not statistically significant at 1%, 5%, and 10% level. Hence, the null hypothesis should be rejected. It can be concluded that QE3 tapering announcements do not lower mortgage REITs stock returns significantly. In terms of control variable, as expected, the coefficient of market index S&P 500 is positive and significant at a conventional level. On the other hand, the coefficients of lagged return and bond market are both significantly negative. 6. Discussion and Conclusion Both empirical models make an assumption that the systematic risk remain constant during the event period and do consider the change in systematic risk in relation to stock market over the period of the information flow. Xu and Yiu (2010) used the product of the dummy variable and market return as a proxy of systematic change over the event period and found the coefficient to be statistically significant. Moreover, all four events have happened within the period of six months, which is a relatively shorter time period, similar method should be considered in the future study. The problem of omitted variables may exist in the multivariate regression. The existing literature shows that the volatility of REITs return proxied by variance of daily returns can also affect the stock returns. Furthermore, the short-term interest rate is the borrowing cost of the mortgage REITs in order to invest in the long-term mortgage. As the level is very close during the whole event period, it is not included in the model. To sum it up, despite there are limitations in terms of data and models, the results still offer some perspective to remind investors that in mortgage REITs sector, people do not respond aggressively to information shock. It is wise to keep a portion of mortgage REITs in the portfolios during the tapering period. However, the position should not be too large in case of rising interest rates in the near future. To conclude, the purpose of this project was to study whether the Federal Reserve’s QE3 announcement hit the mortgage REITs stock returns or not. The empirical models applied do not show evidence that there is a negative or significant announcement effect. The reason behind might be that the mortgage REITs are adapting to the changes in the monetary policy by implementing hedging strategies, such as, reducing portfolio size in mortgage-backed securities and adjusting its asset-mixed portfolio. Investors are well informed about those hedging strategies. Meanwhile, on the actual announcement date, the Federal Reserve claimed that the interest rate would not rise in the near future. This may boost the investors’ confidence in mortgage REITs sector because of which they do not sell their stocks aggressively. However, the limitation of the project methodology does exist and refinement will be needed in the future to improve the results. References Mankinley, C 1997, ‘Event Studies in Economics and Finance’, Journal of Economic Literature, vol. 35, no. 1, pp. 13–39. Brown, S & Warner, J 1985, ‘Using Daily Stock Returns: The Case of Event Studies’, Journal of Financial Economics, vol. 14, no. 1, pp. 3-31. Keown, A & Pinkerton, J 1981, ‘Merger Announcements and Insider Trading Activity’, Journal of Finance, vol. 36, no. 4, pp. 855-869. Randleman, R, Jones, C & Latane, H 1982, ‘Empirical Anomalies Based on Unexpected Earnings and the Importance of Risk Adjustments’, Journal of Financial Economics, vol. 10, no. 3, pp. 269–287. Johnson, R & Jensen, G 1999, ‘The The Federal Reserve Monetary Policy and Real Estate Investment Trust Returns’, Real Estate Finance, vol. 16, no. 1, pp. 52–59. Gabriel, S & Lutz, C 2015, ‘The Impact of Unconventional Monetary Policy on Real Estate Markets’, viewed 14 April 2015, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2493873 Rai, V & Suchanek, L 2014, ‘The Effect of the Federal Reserve’s Tapering Announcements on Emerging Markets’, Bank of Canada Working Paper 2014-50, viewed 14 April 2015, November 2014, http://www.bankofcanada.ca/wp-content/uploads/2014/11/wp2014-50.pdf Xu, Y & Yiu, C 2010, ‘The Effects of Tax Reforms on REITs: an International Empirical Study’, viewed 14 April 2015, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1878505. Read More