The Heterogeneous Impact of
Monetary Policy on Stock Prices
Faculty: Economics & Business
Bachelor program: Economics & Business
Specialization: Economics & Finance
Student Name: Sander Burgers
Student Number: 10166505
2
Abstract
The effects on monetary policy on the economy are still not fully understood. This paper contributes to a better understanding of the monetary policy effects on the equity market by analyzing the returns of a variety of sector indexes on days of FOMC meetings. The Federal Funds target rate is taken as a measure of monetary policy decisions and Federal Funds futures are used to derive the market expectations of the policy decisions. The results show that an unanticipated 25 basis points increase of the Federal Funds target rate leads to a decrease of the daily return of the S&P500 Index of 1,58 %. Target rate changes that are anticipated by the market do not have any effect. This paper also examines whether monetary policy effects has similar effects across sectors of the S&P500. It is found that the effects of unanticipated monetary policy decisions on stock returns vary across sectors from no significant effect to a response more than 2,5 times stronger than average. It can be concluded that stocks respond in a heterogeneous fashion.
3
Table of Contents
Section 1: Introduction
4
Section 2: Theoretical Framework
5
Section 3: Method
9
Section 4: Results
16
Section 5: Conclusion
20
Reference List
21
4
Section 1:
Introduction
There has ever been a long debate on the importance of monetary policy on the macro-economy. Economists who believed the earlier Keynesian theory of the 1950s and 1960s stated that monetary policy does not affect the real economy, while the followers of Milton Friedman believed that monetary policy certainly has effects. Nowadays, the general believe is that it does matter, but it is not clear through what way exactly and which economic agents are affected most. Many theories have been developed which try to explain the effects resulting a monetary policy decision, which intentionally should affect macroeconomic variables like inflation and unemployment. These theories, or channels of the monetary transmission mechanism, often state that their first round effects are most directly on the financial markets. Studying the monetary policy effects on the equity market, which is one of the most important financial markets, is therefore essential in understanding the monetary transmission mechanisms.
The aim of this paper is to contribute to the discussion of the effects of monetary policy on the equity market. Many papers (e.g., Ehrmann & Fratscher, 2004; Bernanke & Kuttner, 2005; Gürkaynak, Sack, & Swanson 2005) have studied the monetary policy effects on the stock market before. The paper of Ehrmann and Fratscher (2004) uses an event-study methodology with daily returns from 1994 until 2003 to examine whether monetary policy effects vary across firms. Bernanke and Kuttner (2005) also use an event-study methodology with monthly data from 1989 until 2002, using Fama-French Industry Portfolios to analyze whether firms respond in the same magnitude to monetary policy. Gürkaynak, Sack and Swanson (2005) use intraday data to measure the response of the S&P500 Index returns around 30 minutes after the FOMC announced their policy decision announcement. This paper uses an event-study methodology to examine whether monetary policy effects vary across firms by using 10 sector indexes of the S&P500. The research question which will be answered is: does monetary policy have heterogeneous effects on stock returns across different sectors?
These effects are estimated by performing a event-study analysis, based on the method of previous papers (e.g., Bernanke & Kuttner, 2005; Ehrmann & Fratscher, 2004). The monetary policy decisions are measured by changes in the Federal Funds target rate and divided into an anticipated and unanticipated component (to control for the decisions the financial market has expected) (Kuttner K. , 2001). The S&P500 index is used to measure the returns of the US stock market. The ten sectors indices are according to the Global Industry Classification Standard (GICS®) and are
components of the S&P500 index and used to measure the effects per sector.
The structure of this thesis is as follows. In Section 2 the chosen method will be described and why it is suited to help answering the research question. Section 3 explains the relevant theories
5 of the research topic. Section 4 interprets the actual results and discusses the expected and
unexpected findings. Section 5 contains the conclusion and answers the research question.
Section 2: Theoretical Framework
The Fed
The Federal Reserve System, or the Fed, is the central banking system of the United States of America who tries to achieve stable prices, maximum employment and moderate long-term interest rates (Board of Governors Federal Reserve System, 2005). The tools used trying to achieve these objectives are called monetary policy. The Fed directly controls three tools to influence the
instruments which they use to set their policy target at. The tools are: the discount rate, the reserve requirements and open market operations. The open market operations are executed by the Federal Open Market Committee (FOMC), which is the most important policy making body of the Fed
(Federal Rerserve System, 2014). The members of the committee vote for the policy to be carried out between the two meetings. They regularly hold meetings eight times a year. The Fed has chosen to target the Federal Funds rate as their main policy instrument, which is the interest rate at which depository institutions lend reserves to other depository institutions overnight. Since 1982, the Federal Funds rate is a simple but appropriate indicator of the Federal Reserve policy stance (Bernanke & Milhov, 1998). Bernanke & Blinder (1992) argue that it is a good indicator since 1979. Moreover, it is a better indicator than the quantity of non-borrowed reserves, they say.
Monetary transmission channels to stocks
To estimate the effects of monetary policy on stocks returns, the valuation in stocks should be explored in more detail. The dividend-discount model is a widely used (e.g. Farrel, 1985; McQueen & Roley, 1993; Penman, 1998; Berk & DeMarzo, 2011) valuation model of stocks that states that the price of a stock should equal the sum of the present value of the future cash flows received when holding the stock1. Similar alternatives to this model often include the capital gain received when the
stock is sold at a higher price in the future. The present value of the future cash flow is obtained by dividing the cash flow by the appropriate discount rate. The cash the stockholder receives by the firm is called dividend. This division in order to get the present value is called discounting. When the To determine the amount of the cash flows and the discount rate, all the information available to investors is used. The Efficient Market Hypothesis (Fama E. F., 1970) argues that public information and expectations of the future is fully and quickly reflected in securities. The information used is both firm-specific (Fama, Fisher, Jensen, & Roll, 1969) (Patell & Wolfson, 1984)and macro-economic (Waud, 1970) (Boyd, Hu, & Jagannathan, 2005). Since a stock is a security; information about future
1 In formula:
𝑃𝑃0= � � (1+𝑟𝑟𝐷𝐷𝐷𝐷𝐷𝐷𝐸𝐸𝑛𝑛)𝑛𝑛�
∞
6 monetary policy changes is already included in the pricesand should therefore have no effect on stock returns. Therefore, only unexpected policy changes should effect stock prices.
According to the dividend-discount model, the value of a stock is determined by two
variables: the future cash flows the stock holder receives, which are the corporate dividends and the eventual capital gain (when the stock is sold), and the appropriate discounting rate. Stock prices therefore only respond to monetary policy changes when it affects the forecast of the future corporate dividends or the discount rate (also called: cost of capital or the required rate of return). To keep the possible effects organized; these two ways provide the structure of this section. The first subsection explains the possible channels through which monetary policy may affect the forecast of the future corporate dividends. The second subsection explains the impact of monetary policy on the discount rate.
Effects on expected dividends and capital gain
The future dividends and capital gains a shareholder expects to receive by the firm are affected by changes in the federal funds rate through several monetary transmission channels which will be discussed here.
The credit channel of monetary policy transmission describes the way a monetary policy decision affects the real economy through reducing the availability of credit to firms. This theory is discussed widely in Bernanke & Getler (1995).They state that monetary policy directly affects interest rates and that its effect is enhanced by changes in the external finance premium, which is the difference in the cost of external raised funds and internal generated funds (Bernanke & Gertler, 1995). A changein this external finance premium is caused by two ways; through the balance sheet channel and the bank lending channel, which will be discussed in the next paragraphs. A higher external finance premium implies higher interest expenses or no funds for firms, which in turn lowers the current and/or future profitability.This lower profitability could change the investors’ prospect for their forecasted future dividends and therefore affect stock prices.
The balance sheet channel theory says that monetary policy affects the external finance premium through the strength of the financial position of the borrower. The financial position or net worth is ‘operationally defined as the sum of the liquid assets and marketable collateral’ (Bernanke & Gertler, 1995). A monetary easing, for example, directly improves the financial position by lowering the interest expenses which has to be paid over outstanding short-term or floating-rate debt. Another (indirect) effect on the financial position occurs is due to of an increase in the revenues, because of higher customer spending due to lower cost-of-capital reasons (Bernanke & Gertler, 1995). Bernanke & Getler (1995) provide emprical evidence for this link by showing a strong correlation between the Federal Funds rate and the coverage ratio. A firms’ coverage ratio is calculated by dividing the earnings before interest and taxes by the interest expenses. The coverage
7 ratio is a common and practical measure of the strength of the financial position of the firm. A stronger financial position in turn causes less potential conflict of interest with the lender, because the borrower is more able to finance the investment internally and the higher value of the collateral makes it easier to get funds from the lender (Bernanke & Gertler, 1995). Moreover, when firms have stronger financial positions as a result of a monetary expansion, asymmetric information problems like moral hazard and adverse selection will diminish (Bernanke & Gertler, 1995). Moral hazard arises when the lender runs the risk of not receiving the loan back from the borrower when the borrower engages in risky activities (Giuliodori, Mishkin, & Matthews, 2013). With a stronger financial position, the firms has more incentive to behave appropriately and to take less risk (because the firm has more to lose). Therefore a lender is more willing to lend at a lower risk premium. Adverse selection occurs when a product or service is only purchased by a group of customer who is least profitable for the firm. In this case, adverse selection will lead to a situation where only firms with weak financial positions (who need the loans the most) borrow at certain risk premiums, because firms with strong financial position consider the risk premium too high (Giuliodori, Mishkin, & Matthews, 2013). As a result, the lender responds to increasing the risk premium, which enlarges the problem.
Asymmetric information problems, will also decrease which results like adverse selection and moral hazard, and the risk run by the lender will therefore decrease, which results into a lower external finance premium.
The bank lending channel theory says that monetary policy affects the external finance premium by ‘shifting the supply of intermediated credit’ (Bernanke & Gertler, 1995). An open market sale (for example) leads to a decline in the volume of reserves at banks and therefore to a situation wherein banks provide less loans or relatively more expensive loans, because they have less funds available to lend.This only holds true if it assumed that loans and securities are imperfect substitutes in bank portfolios to ensure that a decrease in reserves results to a decrease in loans. This
assumption is empirically reasonable (Blinder & Stiglitz, 1983). This is also suggested by an earlier paper (Bernanke & Blinder, 1988). As a result, many bank-dependent borrowers who find it difficult to obtain funds in the open market, will reduce their expenditures and will have to build a new credit relationship which costs time and money. Therefore, it is likely that the external finance premium will increase.
Firms who have fewer difficulties in obtaining external funds should be affected less by the external finance premium, whereby a variation might arise across firms (Dale & Haldane, 1995). This variation is thus not unlikely to occur across sectors.
Another general consensus is that consumer spending on durable goods is more sensitive to interest rate changes than non-durable goods. This has been shown in several papers. Bernanke & Getler (1995) have proven the different impact of monetary policy on the spending of investment
8 durable goods versus non-durable goods. Erceg & Levin (2006) showed that a monetary policy shock causes a decline in consumer durables and residential investment spending that is over three times as large as for the other GDP components. Dedola & Lippi (2005) state that output responses to monetary policy shocks are systematically related to the sector output durability, financing
requirements2 and borrowing capacity3. They suggest that sectors who produce durable goods have
greater financial requirements or smaller borrowing capacity, react stronger to monetary policy. House purchase is highly interest rate sensitive which implies that the demand declines rapidly after a monetary policy tightening (Ganley & Chris, 1997). As a result, the materials sector will suffer since the manufacturing and sales of glass, tiles, concrete, brick and wood products will decline, they say. Ehrmann & Fratscher (2004) suggest that, because of these reasons, cyclical and capital-intensive industries are affected the most. These theories therefore suggest that sectors which operational activities are more dependent on the sale of durable goods, should respond in a greater magnitude to monetary policy announcements.
Carlino & Defina (1998) explain the difference in regional effects of monetary policy in the US through the credit channel and traditional interest rate channel. They find empirical evidence of the existence of the interest rate channel by finding that states with a relative large proportion of value added by manufacturing firms of the total gross product respond in a greater magnitude to monetary policy than states with a relatively small proportion of manufacturing. Manufacturing is thought to be an interest-sensitive factor. States that consist of a relatively large proportion of small firms do not clearly show that they respond more to monetary policy, although it is significant (Carlino & Defina, 1998). Presuming that small firms are more credit constraint, some evidence is found for the credit channel effect.
Effects on the discount rate
The discount rate (also called: required rate of return or cost of equity capital) is a theoretical and a not direct observable rate which discounts the dividends and capital gain to the present. It is composed of the risk-free rate and the appropriate risk-premium for the specific stock.
The risk free interest rate is the rate of return which is earned on an investment with zero risk. This is solely a theoretical rate because in practice no investment could be completely risk-free and therefore a nearly risk-free bond of a stable government is chosen where the risk of default is so low that is negligible. In the USA, the 3-months Treasury bill is often taken for short term valuations. Evans & Marshall (1999) have found that monetary policy shocks have a significant positive but transitory effect on the short-term interest rates, a smaller effect on medium-term interest rates and
2 Working capital ratio( (current assets – current liabilities)/(total assets)) and short term debt ratio(short term debt/ total debt)
3 Measured by firm size (number of employees per firm), leverage ratio (total debt / shareholders equity) and interest rate burden (interest expense / operating profits).
9 no effect on long-term rates. Their findings imply that especially the slope of the yield curve changes. Kuttner (2001) proves partly similar results. According to his paper, there is a strong relationship between unanticipated target rate changes and market interest rates. Hence, stock investors might adjust the risk free rate used in determining the discount factor as a result of monetary policy changes. Stock prices therefore change as well. It is very unlikely that stocks of different sectors respond differently through this way because this risk-rate can be earned by any investor and therefore incurs as the same opportunity cost of capital, regardless of the security which should be priced.
The risk-premium (i.e., equity premium) is the compensation an investors wants to get for the exposure to the higher risks of a stock. Because of this risk premium; an investors is willing to take the extra risk. The risk-premium consists of several kinds of risks like: interest rate risk,
purchasing power risk (or inflation risk) and default risk. Lobo (2000) and Brandt & Wang (2003) state that monetary policy changes are often viewed by stock holders as signals of the future direction of interest rates and inflation and therefore changing the willingness of holding stock. A monetary tightening, for example, causes the interest rates of bonds to rise which makes equity stock less attractive. Therefore investors will demand a higher risk-premium. Thornton (1998b) concluded that the Fed has some credibility as an inflation fighter and that the Federal Funds Target significantly affects the inflation outlook. As a result, a monetary tightening may increase the expected future real interest rate, which implies higher risk-premium and a lower stock price.Bernanke & Kuttner (2005) add to this that a tightening directly increases the riskiness of stock because of raising interest costs. Alhough it is unclear whether these effects transmit trough the way of a higher risk-premium or revisions of the expected future dividends.
Considering the theory section, it can be expected that stocks respond in a heterogeneous fashion to monetary policy announcements. In summary, the theories suggest that several characteristics of sectors could result in a more sensitive response to monetary policy decisions. The characteristics include: the production of durable goods, a large proportion of small firms compared to large firms and firms with a high leverage.
Section 3: Method
In this section, the method used to answer the research question is described and why it is suited. It describes the strengths and weaknesses of the method. It also explains the data and how it is collected.
10 The Federal funds rate as monetary policy measure
To measure monetary policy, I use the Federal Funds target rate as indicator. This policy instrument has been oftentaken as monetary policy indicator (e.g., Cook & Hahn (1989), Bernanke & Blinder (1992), Laurent (1998), Gürkaynak, Sack, & Swanson (2005)) and is preferred over others indicators (like money stock innovations) especially if the central bank uses an interest rate as instrument (McCallum, 1983). Bernanke and Blinder (1992) argue that the Fed has implemented policy changes through changes in the Federal funds target rate over more than the past 30 years and conclude therefore that it may be used as an indicator. One shortcoming of this approach is that other instruments, like the quantity of reserves and monetary base, is not included but could also have its effects. Forward guidance is not included either, although it is not clear whether it has effects (Ehrmann & Fratscher, 2004).
In the period from January 1995 until December 2008 there were 118 meetings and target rate decisions made. I am going to use these decisions to reflect the policy of the Fed. However, to measure the effect of monetary policy, the changes of the Federal funds target rate cannot simply be used because markets might anticipate the policy change (or no change, which is a policy decision too) and therefore this change is already included in the stock price. This is an example of the Efficient Market Hypothesis. Kuttner (2001) confirmed this with empirical evidence by showing that anticipated changes have a little or no effect. Hence, the surprise or unanticipated part of the interest rate change should be taken. To calculate these extracted parts, I am going to use the method following Kuttner (2001), who used the settlement prices of the Federal funds futures. Kuttner (2001) said: “The Fed funds futures contract’s settlement price is based on the average of the
relevant month’s effective overnight Fed funds rate, rather than the rate on any specific day. The futures rate is defined as 100 minus the contract’s price. An additional twist is that the average is computed over every day in the month, with rates for weekends and holidays carried over from the previous business day.”
This method is used in this paper, because Kuttner & Krueger (1996) showed that the Fed funds future markets is very good at predicting changes in the Fed funds target rate. Moreover, more recently, Gürkaynak, Sack, & Swanson (2007) proved that Federal funds futures are the best
predictors of target funds rate changes 1-5 months ahead. The price of the future is based on the current month federal funds rate and therefore the unexpected policy rate change in basis points can be derived from the future with this formula (Kuttner K. , 2001):
11 Where D denotes the number of days in the month wherein the FOMC meeting takes place, d the day of the month the FOMC takes place, 𝑓𝑓𝑑𝑑,𝑚𝑚0 the Federal funds rate implied by the settlement price
of the current-month federal funds future traded on the day the FOMC decision is announced. And 𝑓𝑓𝑑𝑑−1,𝑚𝑚0 the Federal funds rate implied by the settlement price of the current-month federal funds
future traded on the day before the FOMC meeting. The implied federal funds rate in percentage equals 100 minus the settlement price at that day4. This formula cannot be used in two cases: when
the FOMC is at the first day of the month and at one of the last seven days of the month (Bernanke & Kuttner, 2005). In case of the first day of the month, the next-month future price of the last day of the previous month should be used instead of the current-month future price. If the meeting takes place within the last seven days of the month, the scaling factor should be omitted because it gets too large. A special occasion occurs at 15 October 1988, when the Fed announced a decrease in the target rate of 25 basis points at 3:15 p.m. (Eastern Time). Although the announcement was released at the 15th, the futures markets was already closed and therefore did not included the change until
the following day. That is why it is better to take the difference between the closing price of the 15th
and the opening price 16th.
In the formula I added the factor 100 to get the unanticipated rate in basis points, because the policy changes are normally smaller than 1% and to avoid a large amount of decimals when using percentages. The days of the FOMC meetings and statements are all published on the website of the Federal Reserve (www.federalreserve.gov). The settlement prices of the federal funds futures are available at the Chicago Mercantile Exchange Group database website of the CME Group5. All the
monthly individual futures are available from January 1988 until now. Event-study analysis
To examine whether monetary policy has effect on the stock prices of firms, I use the ordinary least squares method to estimate the following regression:
𝑟𝑟𝑀𝑀,𝑡𝑡= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝜀𝜀𝑡𝑡 (2)
Where the 𝑟𝑟𝑀𝑀,𝑡𝑡 is the daily return of the S&P500 stock index on days of the FOMC meetings, ∆𝑖𝑖𝑡𝑡𝑢𝑢 is
the unanticipated component of the Federal Funds Target Rate change on the same day and 𝜀𝜀𝑡𝑡 is the
error term. When calculating the standard errors, I correct for heteroskedasticity. This method has been used more frequently in literature (e.g., Ehrmann & Fratscher (2004) Gürkaynak; Sack & Swanson (2005) and Bernanke & Kuttner (2005)). The daily returns of the S&P500 are calculated by
4 For example, if the settlement price of the future on the 3th of May 2005 is 97, the implied rate 𝑓𝑓
𝑑𝑑,𝑚𝑚0 = 100 –
97 = 3 %
12 subtracting the closing price of the index on the day before the FOMC announcement from the closing price at the day of the FOMC announcement, divided by the closing price on the day before the announcement and multiplied by 100 to get the return in percentages6. The prices are all
available on DataStream for the chosen time span. The factor 100 is included because this is also done in previous literature (e.g., Bernanke & Kuttner(2005); Ehrmann & Fratscher (2004)).The S&P5007 is one of the most commonly used benchmarks of the US stock market. A shortcoming of
taking the S&P500 Index is that these stocks are of the 500 biggest firms of the USA and are all listed on the NYSE or the NASDAQ. These stocks have therefore different characteristics compared to stocks of smaller firms and/or stocks which are not listed.
To test whether the anticipated rate changes have no effect on stock returns, which is the case in the paper of Kuttner (2001), the following regression is estimated by using ordinary least squares:
𝑟𝑟𝑀𝑀,𝑡𝑡= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽2∆𝑖𝑖𝑡𝑡𝑎𝑎+ 𝜀𝜀𝑡𝑡 (3)
Where the added variable ∆𝑖𝑖𝑡𝑡𝑎𝑎 is the anticipated component of the Federal Funds Target rate change
on the days of the FOMC meetings, the other variables are the same as in equation (2). This regression is estimated as a robustness check. The results are shown in the appendix.
Subsequently, to examine whether these effects are heterogeneous on stocks of the equity market, the S&P500 is divided into 10 sectors according to the Global Industry Classification Standard (GICS). The 10 sectors are: Health Care, Energy, Materials, Industrials, Consumer Discretionary, Consumer Staples, Financials, Information Technology, Telecommunication Services and Utilities. Table 3 in the appendix shows the sub-sectors and industries the sectors consist of. These effects are measured in the same way as above in equation (1), by performing an OLS-regression, but instead, replacing the returns of the aggregate market by the sector specific returns:
𝑟𝑟𝑘𝑘,𝑡𝑡 = 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝜀𝜀𝑡𝑡 (4)
Where 𝑟𝑟𝑘𝑘,𝑡𝑡is the daily return of sector k. This regression will be run 10 times, each time estimating
the unanticipated monetary policy effect on a specific sector. The daily returns of the 10 sector indexes are calculated in the same way as the returns in equation (2).
To examine whether the effects of the unanticipated target rate changes per sector are significantly different from each other, I perform an F-test on the equality of coefficients. The F-test is suitable to evaluate hypotheses that involved multiple parameters. In order to perform this test, the returns of the sectors should be pooled into one OLS-regression analysis, which has the following mathematical form:
6
In formula: 𝑟𝑟𝑀𝑀,𝑡𝑡 = 𝑃𝑃𝑘𝑘,𝑡𝑡𝑃𝑃− 𝑃𝑃𝑘𝑘,𝑡𝑡−1𝑘𝑘,𝑡𝑡−1 * 100.
7 This index is a market valued weighted index consisting of 500 stocks chosen for market size, liquidity and industry grouping selected by a committee of Standard & Poor.
13 𝑟𝑟𝑘𝑘,𝑡𝑡= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡 𝑢𝑢+ 𝛽𝛽 2𝐷𝐷𝐻𝐻𝐻𝐻𝐻𝐻∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽3𝐷𝐷𝐻𝐻𝐻𝐻𝐻𝐻+ 𝛽𝛽4𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽5𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸+ 𝛽𝛽4𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽7𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀+ 𝛽𝛽8𝐷𝐷𝐼𝐼𝐸𝐸𝐷𝐷∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽9𝐷𝐷𝐼𝐼𝐸𝐸𝐷𝐷+ 𝛽𝛽10𝐷𝐷𝐻𝐻𝐶𝐶𝐷𝐷∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽11𝐷𝐷𝐻𝐻𝐶𝐶𝐷𝐷+ 𝛽𝛽12𝐷𝐷𝐻𝐻𝐶𝐶𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽13𝐷𝐷𝐻𝐻𝐶𝐶𝑀𝑀+ 𝛽𝛽14𝐷𝐷𝐹𝐹𝐼𝐼𝐸𝐸∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽15𝐷𝐷𝐹𝐹𝐼𝐼𝐸𝐸+ 𝛽𝛽16𝐷𝐷𝐼𝐼𝐸𝐸𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽17𝐷𝐷𝐼𝐼𝐸𝐸𝑀𝑀 + 𝛽𝛽18𝐷𝐷𝑀𝑀𝐸𝐸𝑇𝑇∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽19𝐷𝐷𝑀𝑀𝐸𝐸𝑇𝑇+ 𝜀𝜀𝑡𝑡 (5)
Where 𝑟𝑟𝑘𝑘,𝑡𝑡is the daily return of the sector index k on the day of a monetary policy change, ∆𝑖𝑖𝑡𝑡𝑢𝑢 is the
unanticipated federal funds rate change and the DK-symbols denote dummy variables. If the sector is
Health Care, DHCR takes value 1 and other dummies take value 0; if the sector is Energy, DENE takes
value 1 and other dummies take value 0. The same applies for the other sectors, except in case of the Utilities sector. In that case, all the dummies take the value 0.
The error terms in models (2), (3), (4) and (5) are corrected for heteroskedasticity. When correcting for heteroskedasticity, the estimated coefficients remain unbiased and consistent even if the variance of the errors appear to be uncorrelated with the explanatory variable (Stock & Watson, 2011). This is also assumed by Ehrmann & Fratscher (2004) in their similar model.
Data
I have chosen for daily data, following the event-study approach of Ehrmann & Fratscher (2004). They state that daily returns have a good balance between the identifiability of exogenous monetary policy and estimation of stock market effects. They state that a smaller window might capture overshooting effects that quickly disappear and a larger window might lead to endogeneity problems. They therefore assume that the daily window is the most reliable way to predict the longer-run impact of monetary policy on stocks.The key identifying assumption is that news about the economy on the FOMC day does not affect the policy choice. Only information available the previous day is relevant (Gertler & Karadi, 2013).
In other literature it is pointed out that, when making use of daily data, endogeneity problems might still occur, because monetary policy may then be a response to changes in stock prices when a heavy shock occurs earlier that day. This can be a response to big economic news released that day and as a result, including these observations in the analysis cause problems in the unbiasedness of the beta estimators (Gürkaynak, Sack, & Swanson, 2005). Therefore these days should be omitted from the analysis. Gürkaynak, Sack & Swanson (2005) say intraday is preferred to estimate the unanticipated interest rate, but daily day measures it very well too. Ribogon and Sack (2004) found that short-term interest rates respond to broad equity price indexes and therefore the study approach causes problems due to simultaneous causality. They also say that the event-study approach might require some more explanatory variables like information about the
14 macroeconomic outlook or changes in risk preferences to avoid omitted variable bias. Consequently, the estimator might be biased. To test whether this is true, they set up an identification method based on the heteroskedasticity of monetary policy shocks and implement the identification method as an instrumental variable regression and a generalized-method-of-moments approach (Ribogon & Sack, 2004). They compare these three methods on the quality of the estimator, by looking at the monetary policy effects on four stock price indexes (S&P500, WIL5000, NASDAQ and DJIA). The results of this analysis appear to show and upward-bias (closer to zero) in the estimated slope
coefficient of monetary policy changes under the event-study approach. However, the statistical test, the Hausman (1978) specification test cannot formally reject the hypothesis that the
heteroskedasticity-based and event-study estimates on the four stock indexes are equal. Thus, the assumption that monetary policy is the primary determinant of the stock prices on these FOMC announcements date, cannot formally be rejected (Ribogon & Sack, 2004). That justifies the use of this event-study approach with daily returns.
A simplification in this model is that there is no different treatment for positive and negative unanticipated Federal Funds Rate changes. It might be the case that the stock market responds with a greater magnitude to a rate increase than a decrease. Therefore, the results from the regression analyses will be an average effect and might be upward or downward biased. Although, Bernanke & Kuttner (2005) tested this empirically and showed no significantly different effect in the magnitude of rate changes of the different signs on stock prices.
In the estimated models, it also assumed that the effect of monetary policy on stocks do not depend on the state of the market. In other words, a monetary policy decision has the same effect on stocks during a period in which the market performs well as in a period the market performs badly. However, there are theories and empirical evidence which suggest that this is not true. Kurov (2010) found evidence that stocks are affected stronger in periods of bear market conditions compared to bull market periods. He explains this by the role investor’s sentiment plays on stock returns, since monetary policy changes affect investor sentiment more in bear markets than in bull markets (Kurov, 2010). McQueen & Roley (1993) point out that if the effect is corrected for stages of the business cycle, a stronger relationship between stock price and macroeconomic news is evident. Thus, the stock market’s response to macroeconomic news is different in different states of the economy (McQueen & Roley, 1993).
I have chosen for the time span from 1995, because the required price data of several S&P500 sector indexes is unavailable before that time. If this was not the case, 1994 would be first year to start the observations, because in that year the FOMC started to explicitly announce the Federal funds target rate to the public, whereas before, changes where communicated through the size and type of open market operations (Gürkaynak, Sack, & Swanson, 2005). In 2008, the last
15 Federal Funds rate changes have been made. From that moment on, the Federal Funds target
reached the lower bond of 0 and therefore forward guidance became the only way the central bank could affect market interest rates without resorting to unconventional credit market interventions. (Gertler & Karadi, 2013)This time span leads to 118 days of FOMC meetings.
Observations
As pointed out before; these 118 observations need to be analyzed in greater detail, because FOMC days which also contain other (economic) news releases should be omitted to avoid obtaining biased estimators. To decide which news releases are meaningful enough to exclude, I follow the method of Bernanke & Kuttner (2005). They excluded days which contain Employment Report releases of the Bureau of Labor’s Statistics (BLS). This decision is justified because a paper (Boyd, Hu, & Jagannathan, 2005) has empirically proven that stocks do significantly respond to employment report releases.
Of these 118 observations, it did not occur that an Employment Report was released on the same day as the FOMC meeting and announcement, according to the Economic Calendar of Yahoo Finance8.Therefore, no observations should be omitted for these reasons.
However, three observations should be excluded from the analysis because of other, varying reasons.
The Federal funds rate cut on the 17th of September 2001 occurred at the first trading day
after the terrorist attack at 9/11 and therefore the news from that day on until 17 September are included in the stock returns.
On the 22nd of January 2008, the FOMC arranged a conference call and decided to lower its
rate with 75 basis point because of ‘weakening of the economic outlook and increasing downside
risks to growth’ (Press Release 22 Janary 2008). This conference call was held 8 days before a
planned FOMC meeting on the 30th of January. Because of the fact that these two decisions occurred
soon after each other and within the same month; the unanticipated rate change on the 22nd cannot
be derived from the futures because they do not tell if the change is expected to be made on the 22nd
of 30th. In January 2001, a similar situation occurs, but this does not create an inaccurate measure of
the unanticipated rate change. Because in January 2001, there were two FOMC meetings within the same month as well, but these were respectively on the 3rd and 31st of the month. A policy decision
made on the 3rd will affect the weighted average of the funds rate heavily and this will be reflected in
the settlement prices of the future. The decision made on the last day of the month will have a relatively (compared to the decision on the 3rd) small effect on the price. Therefore, the observation
3rd of January does not need to be excluded.
8 This calendar contains all the economic news releases in the USA, classified into A-F based on their importance. Only the Employment Report merits an importance level A. The calendar can be found on:
16 On the 18th of December of 2008 the Fed announced to set its target rate between 0 and
0.25. Because of this decision to set a target bound instead of one specific numerical target, the target rate change cannot be computed (let alone the anticipated and unanticipated components) and therefore it should not be included in the analysis.
Section 4: Results
Table 1: OLS-regressions results of the estimated coefficients of unanticipated component of Federal Funds target rate change
S&P500 return coefficients Robust standard error R2
Unanticipated target rate change
-0,063** 0,025
0,1380
constant 0,290** 0,118
Note: The response is estimated as 𝑟𝑟𝑡𝑡= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝜀𝜀𝑡𝑡 . The robust standard errors are displayed between brackets below the coefficient. * indicates that the coefficient is significantly different from 0 at a significance level of 10%, ** indicates significantly different from 0 at a significance level of 5% and *** indicates significantly different from 0 at a significance level of 1%.
Table 1 reports the estimated coefficients of the OLS regression analysis with only the unanticipated component of the Federal Funds rate change included. The daily return of S&P500 Index declines on average with 1,58 % as a result of an unexpected 25 basis point increase in the Federal Funds target rate. The result is significant at a 5% level and the negative sign is expected according to the theories of monetary policy transmission. The R2 of 0,1380 means that 13,80% of the variation of the stock
returns are explained by unanticipated Federal Funds target rate changes. This implies that a large proportion (86,20%)of stock returns variances are explained by other variables. This result is similar compared to the findings of Bernanke & Kuttner (2005) and Ribogon & Sack (2004).
The results of the analysis with both the unanticipated and anticipated component of the Federal Funds rate change included can be found in table 4the appendix. These results showed that the response of the S&P500 to anticipated rate changes is very small and not significantly different from zero. Therefore, it cannot be rejected that anticipated policy rate changes do not affect stock returns. The R2 did not increase after adding the variable and remains 0,1380. This finding is in
accordance with the Efficient Market Hypothesis and is also found at the papers of Kuttner (2001) and Bernanke & Kuttner (2005) . The effect of the unanticipated rate change on the stock prices barely differs from the analysis without the anticipated rate change. These findings are similar those of Bernanke & Kuttner (2005), Gürkaynak, Sack & Swanson (2005) and Ehrmann & Fratscher (2004).
17 Table 2 reports the results of the regressions performed with the ten GICS sectors indexes of the S&P500. The results shows that the sectors Health Care, Energy, Consumer Staples, Financials, Telecommunication services and Utilities sector have estimated coefficients which are not
significantly different from zero at significance level of 10%. Therefore, it cannot be rejected that these sectors indexes are not affected by monetary policy surprises. The R2 ‘s of these sectors,
Table 2: The response of the stock sector indexes to unanticipated Federal Funds rate changes
Sector: Unanticipated rate
change coefficient R2 Health Care 0,024 (0,024) 0,0248 Energy 0,027 (0,024) 0,0196 Materials -0,054** (0,021) 0,0784 Industrials -0,081*** (0,023) 0,2005 Consumer Discretionary -0,113*** (0,037) 0,2582 Consumer Staples 0,035 (0,026) 0,0583 Financials -0,056 (0,040) 0,0478 Information Technology -0,195*** (0,064) 0,3192 Telecommunication Services -0,043 (0,041) 0,0379 Utilities 0,033 (0,028) 0,0395
Note: The response is estimated as: 𝑟𝑟𝑡𝑡𝑘𝑘= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝜀𝜀𝑡𝑡 . The robust standard errors are displayed between brackets below the coefficient. * indicates that the coefficient is significantly different from 0 at a significance level of 10%, ** indicates significantly different from 0 at a significance level of 5% and *** indicates significantly different from 0 at a significance level of 1%.
respectively 0,03; 0,02; 0,06; 0,05; 0,04 and 0,04 are very low and indicate that monetary policy decisions do not explain much of the variation of these sectors. These results are similar to the
18 findings of Ehrmann & Fratscher (2004), Bernanke & Kuttner (2005) and Kurov (2010). A possible explanation could be found when looking closer to the caracteristics of the operational acitivities of the sectors. Intuitively, the Health Care, Energy, Consumer Staples, Telecommunication Services and Utilities sector contain mostly firms which provide products or services which are always needed for consumers, regardless of the state of economy. Therefore their sales are still purchased in periods of economic downturn and thus so-called non-cyclical. Therefore, based on intuition, it makes sense that a monetary policy decision does not significantly affect the stock returns of these sectors.
The insignificant coefficient of the Financials sector can, intuitively, not simply be explained by classifying the sector as non-cyclical, because the sector Financials is the first sector to be influenced by monetary policy changes. Monetary policy actions immediately change the
composition of banks assets and liabilities and therefore do affect the firms. However, the effects of monetary policy decisions could possibly be both positive and negative for banks and therefore have an ambiguous effect on their stock sprice.
The sectors Materials, Industrials, Consumer Discretionary and Information Technology all have estimated coefficients which are significantly different from zero. The sectors Industrials, Consumer Discretionary and Information Technology have negative estimated coefficients which have a greater absolute value than the average effect on the S&P500 Index of -0,063. This suggests that the stocks of these sectors react more strongly than average. The results also show that these three sectors have a way higher R2 compared to the others; this suggests that monetary policy
decisions play a bigger role in the variations of stock returns of firms in these sectors. A possible explanation for these results could be found when looking closer at the characteristics of the sectors. The returns of the Materials sectors decreases on average with 1,356% after an unanticipated 25 basis points increase of the Federal Funds target rate. The Materials sector might depend on the housing market, which is affected by monetary policy according to Ganley & Chris (1997). The returns of the Industrials sector decreases on average with 2,015% after an unanticipated 25 basis points increase of the Federal Funds target rate. The Industrials sector is, based on intuition, very capital intensive and contains the production of mostly capital goods, like building products and machinery, which could possibly be a reason for its relatively strong response compared to the average response. The returns of the Consumer Discretionary sector decreases on average with 2,821% after an unanticipated 25 basis points increase of the Federal Funds target rate.
The strong reaction of the Consumer Discretionary sector could be explained by the fact that it consists of mostly firms which produce durable goods, like automobiles and luxury goods, and cyclical industries, like hotels, restaurant and leisure. The Information Technology response is the strongest of all the sectors; the return declines on average with 4.875% as a result of an
19 response to unanticipated policy changes of this magnitude barely occurred because it happened only two times in the selected time frame that the financial market got surprised by a Federal funds rate change with more than 25 basis points. The sector contains the industry groups Software & Services and Technology Hardware and Equipment which appear to be very cyclical and very capital-intensive, at first sight.
Table 5, in the appendix, shows the results of the regression analysis including the dummies of each sector. Table 6, in the appendix, shows the results of the Wald-test on equal coefficients. The F-statistic (with respectively 9 and 1130 degrees of freedom in the numerator and denominator) equals 4,22 which is highly significant. This means we reject the null hypothesis which stated that the 9 coefficients are equal. This confirms that the effects of unanticipated monetary policy decisions vary by sector at a significance level of 1%.
Table 7, in the appendix, show the results when the anticipated target rate change is added as a control variable. Both the estimated coefficient of the unanticipated rate change and the R2 of
eight9 out of ten do barely change due to an insignificant coefficient of the anticipated rate change.
Only the sectors Energy and Materials show an increase in their R2 of respectively 0,02 to 0,06 and
0,08 to 0,13. The coefficients of the anticipated rate changes are significantly different from zero at a statistical significance level of 5%, suggesting that monetary policy changes which are already
expected, do also affect stock returns of these sectors. This is remarkable because the Efficient Market Hypothesis predicted differently. Hence, these findings remain unexplained and require further research.
9 The eight sectors: Health Care, Industrials, Consumer Discretionary, Consumer Staples, Financials, Information Technology, Telecommunication Services and Utilities.
20
Section 5: Conclusion
The aim of this paper has been to contribute to the discussion of the effects of monetary policy to the economy. In the period from 1995 until 2008, the responses of the stock market to decisions of the Federal Reserve System have been analyzed. The Federal Funds target rate, their main policy instrument, has been taken to quantify monetary policy and the daily returns of the S&P500 Index has been used as a measure of the equity market. Federal Funds futures are used to divide the Federal Funds target rate into an anticipated and unanticipated. The results show that an
unanticipated 25 basis point increase of the Federal Funds target rate results in a decrease of the daily return S&P500 Index of 1,58 %. Changes of the target rate that were anticipated had no effect on the S&P500 Index.
The results of the effects of unanticipated monetary policy decisions on the returns of sector indexes show a variation from no significant impact to a strong negative impact. The significant findings are the following. The returns of the Materials sectors decreases on average with 1,36 % after an unanticipated 25 basis points increase of the Federal Funds target rate. The returns of the Industrials sector decreases on average with 2,02% after an unanticipated 25 basis points increase of the Federal Funds target rate. The returns of the Consumer Discretionary sector decreases on average with 2,83 % after an unanticipated 25 basis points increase of the Federal Funds target rate. The return of Information Technology sector decreases on average with 4,88 % after an
unanticipated 25 basis points increase of the Federal Funds target rate.
The results of the Wald-test on coefficients confirmed that unanticipated monetary policy decisions of the Federal Reserve affect the returns of the sector indexes of the S&P500 in a heterogeneous fashion.
21
Reference List
Angeloni, I., Kashyap, A., Mojon, B., & Terlizzese, D. (2002). Monetary Transmission in the Euro Area: Where do we stand? Working paper no. 114.
Basistha, A., & Kurov, A. (2008). Macroeconomic cycles and the stock market's reaction to monetary policy. Journal of Banking & Finance, 2606-2616.
Berk, J., & DeMarzo, P. (2011). Corporate Finance. Harlo: Person Education Limited.
Bernanke, B. S., & Blinder, A. S. (1988). Credit, Money, and Aggregate Demand. NBER Working Paper
Series.
Bernanke, B. S., & Gertler, M. (1995). Inside the Black Box: The Credit channel of Montary Policy Tansmission. National Bureau of Economic Researh.
Bernanke, B. S., & Kuttner, K. N. (2005). What Explains the Stock Market's Reaction to Federal Reserve Policy? The Journal of Finance, 1221-1257.
Bernanke, B. S., & Milhov, I. (1998). Measuring Monetary Policy. The Quarterly Journal of Economics, 869 - 902.
Blinder, A. S., & Stiglitz, J. E. (1983). Money, Credit and Economic Activity. NBER Working Paper. Board of Governors Federal Reserve System. (2005). The Federal Reserve System: Purposes &
Functions. Washington D.C.: Board of Governors Federal Reserve System.
Boldin, M. (1994). Econometric Analysis of the Recent Downturn in Housing: Was it a Credit Crunch?
Federal Reserve Bank of New York.
Boyd, J. H., Hu, J., & Jagannathan, R. (2005). The Stock Market's Reaction to Unemployment News: Why Bad News is Usually Good for Stocks. The Journal of Finance, 649 - 672.
Boyd, J. H., Hu, J., & Jagannathan, R. (2005). The Stock Market's Reaction to Unemployment News: Why Bad News is Usually Good for Stocks? The Journal of Finance, 649-672.
Brandt, M. W., & Wang, K. Q. (2003). Time-varying risk aversion and unexpected inflation. Journal of
Monetary Economics, 1457 - 1498.
Brown, G. W., & Cliff, M. T. (2004). Investor sentiment and the near-term stock market. Journal of
Empirical Finance, 139-149.
Carlino, G., & Defina, R. (1998). The Differential Regional Effects of Monetary Policy. The Review of
Economics and Statistics, 572 - 587.
Cook, T., & Hahn, T. (1989). The Effect of Changes in the Federal Funds Rate Target on Market Interest Rates in the 1970. Journal of Monetary Economics, 331 - 351.
Dale, S., & Haldane, A. G. (1995). Interest rate and the channels of monetary transmission: Some sectoral estimates. European Economic Review, 1611-1626.
22 Dedola, L., & Lippi, F. (2005). The monetary transmission mechanism: Evidence from the industries of
five OECD countires. European Economic Review, 1543 - 1569.
Ehrmann, E., & Fratscher, M. (2004). Taking Stock: Monetary Policy Transmission to Equity Markets. Erceg, C., & Levin, A. (2006). Optimal monetary policy with durable goods. Journal of Monetary
Economics, 1341 - 1359.
Evans, C. L., & Marshall, D. A. (1999). Monetary policy and the term structure of nominal interest rates: evidence and theory. Carnegie-Rochester Conference Series on Public Policy, 53-111. Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of
Finance, 383 - 417.
Fama, E. F., Fisher, L., Jensen, M. C., & Roll, R. (1969). The Adjustmebt of Stock Prices to New Information. International Economic Review, 1 - 21.
Farrel, J. J. (1985). The Dividend Discount Model: A Primer. Financial Anlysts Journal, 16 - 19 + 22 - 25. Federal Rerserve System. (2014, June 16). Federal Open Market Committee. Retrieved from Board of
Governors of the Federal Reserve System:
http://www.federalreserve.gov/monetarypolicy/fomc.htm
Ganley, J., & Chris, S. (1997). The Industrial Impact of Monetary Policy Shocks: Some Stylised Facts.
Bank of England, 1 - 35.
Gertler, M., & Gilchrist, S. (1993). The Role of Credit Market Imperfections in the Monetary
Transmission Mechanism: Arguments and Evidence. The Scandinavian Journal of Economics, 43-64.
Gertler, M., & Karadi, P. (2013). Monetary Policy Surprises, Credit Costs and Economic Activity.
Unpublished Manuscript, New York University and European Central Bank.
Gürkaynak, R. S., Sack, B. P., & Swanson, E. T. (2007). Market-Based Measures of Monetary Policy Expectations. Journal of Business & Economic Statistics, 201 - 2012.
Gürkaynak, R. S., Sack, B., & Swanson, E. T. (2005). Do Actions Speak Louder Than Words? The Response of Asset Price to Monetary Policy Actions and Statements. International Journal of
Central Banking.
Keasler, T. R., & Goff, D. C. (2007). Using Fed Funds Futures to Predict a Federal Resere Rate Hike.
Journal of Economics and Finance Education, 9-14.
Kurov, A. (2010). Investor sentiment and the stock market's reaction to monetary policy. Journal of
Banking & Finance, 139 - 149.
Kuttner, K. (2001). Monetary policy surprises and interest rates: Evidence from the Fed funds futures market. Journal of Monetary Economics, 523-544.
Kuttner, K. N., & Krueger, J. T. (1996). The Fed Funds Futures Rate as a predictor of the Federal Reserve Policy. The Journal of Futures Markets, 865-879.
23 Lobo, B. J. (2000). Asymmetric Effect of Interest Changes on Stock Prices. The Financial Review,
125-144.
Mankew, N. G. (2009). Macroeconomics. New York: Worth Publishers.
McCallum, B. T. (1983). A Reconsideration Of Sims' Evidence Concerning Monetarism. Economics
Letters, 167 - 171.
McQueen, G., & Roley, V. V. (1993). Stock Prices, News, and Business Conditions. The Review of
Financial Studies, 683 - 706.
Modigliani, F. (1971). Consumer Spending and Monetary Policy: The linkages. Federal Reserve Bank
of Boston Conference Series.
Patell, J. M., & Wolfson, M. A. (1984). The intraday speed of adjustment of stock prices to earnings and divident announcements. Journal of Financial Economics, 223 - 252.
Penman, S. H. (1998). A synthesis of equity valuation techniques and the terminal value calculation for the dividend discount model. Review of Accounting Studies, 303-323.
Press Release 22 Janary 2008. (2008, January 22). Retrieved from Federal Reserve System:
http://www.federalreserve.gov/newsevents/press/monetary/20080122b.htm Ribogon, & Sack. (2004). The Impact of Monetary Policy on Stock Prices. Journal of Monetary
Economics, 1553 - 1575.
Sims, C. (1992). Interpreting the macroeconomic time series facts: the effects of monetary policy.
European Economic Review, 975-1011.
Stock, J. H., & Watson, M. M. (2011). Introduction to Econometrics. Pearson Education.
Waud, R. N. (1970). Public Interpretation of Federal Reserve Discount Rate Changes: Evidence of the "Announcement Effect". Journal of Econometric Society, 231 - 250.
24
Appendix
Table 3: GICS’s sector, industry group and industry classification.
Sector: Industry Group: Industry:
10 Energy 1010 Energy 101010 Energy Equipm ent & Services
151010 Oil, Gas & Consum able Fuels
15 Materials 1510 Materials 101020 Chem icals
151020 Construction Materials 151030 Containers & Packaging 151040 Metals & Mining 151050 Paper & Forest Products
20 Industrials 2010 Capital Goods 201010 Aerospace & Defense
201020 Building Products
201030 Construction & Engineering 201040 Electrical Equipm ent 201050 Industrial Conglom erates 201060 Machinery
201070 Trading Com panies & Distributors 202010 Com m ercial Services & Supplies 2020 Com m ercial & Professional
Services
202020 Professional Services 203010 Air Freight & Logistics
2030 Transportation 203020 Airlines
203030 Marine 203040 Road & Rail
203050 Transportation Infrastructure
25 Consumer
Discretionary
2510 Autom obiles & Com ponents 251010 Auto Com ponents
251020 Autom obiles 2520 Consum er Durables & Apparel 252010 Household Durables
252020 Leisure Products
2530 Consum er Services 252030 Textiles, Apparel & Luxury Goods 253010 Hotels, Restaurants & Leisure
2540 Media 253020 Diversified Consum er Services
254010 Media
2550 Retailing 255010 Distributors
255020 Internet & Catalog Retail 255030 Multiline Retail 255040 Specialty Retail
30 Consumer
Staples
3010 Food & Staples Retailing 301010 Food & Staples Retailing
3020 Food, Beverage & Tobacco 302010 Beverages 302020 Food Products 302030 Tobacco
3030 Household & Personal Products 303010 Household Products 303020 Personal Products
35 Health Care 3510 Health Care Equipm ent &
Services
351010 Health Care Equipm ent & Supplies 351020 Health Care Providers & Services 351030 Health Care Technology 3520 Pharm aceuticals, Biotechnology
& Life Sciences
352010 Biotechnology 352020 Pharm aceuticals
25 Table 4: OLS-estimates of the estimated coefficients of the unanticipated and anticipated
components included
S&P500 return coefficients Robust standard error R2
Unanticipated target rate change
-0,063** 0,026
0,1380 Anticipated target rate change -0,000 0,010
constant 0,290** 0,114
The robust standard errors are displayed between brackets below the coefficient. ** indicates that the coefficient is significantly different from 0 at a significance level of 5%.
40 Financials 4010 Banks 401010 Banks
401020 Thrifts & Mortgage Finance 4020 Diversified Financials 402010 Diversified Financial Services
402020 Consum er Finance 402030 Capital Markets
4030 Insurance 403010 Insurance
4040 Real Estate 404010 Real Estate -- Discontinued effective 04/28/2006
404020 Real Estate Investm ent Trusts (REITs) 404030 Real Estate Managem ent & Developm ent
45 Information
Technology
4510 Softw are & Services 451010 Internet Softw are & Services
451020 IT Services 451030 Softw are
452010 Com m unications Equipm ent 4520 Technology Hardw are &
Equipm ent
452020 Technology Hardw are, Storage & Peripherals
452030 Electronic Equipm ent, Instrum ents & Com ponents
452040 Office Electronics - Discontinued effective 02/28/2014
452050 Sem iconductor Equipm ent & Products -- Discontinued effective 04/30/2003. 4530 Sem iconductors &
Sem iconductor Equipm ent
453010 Sem iconductors & Sem iconductor Equipm ent
50 Telecommunica
tion Services
5010 Telecom m unication Services 501010 Diversified Telecom m unication Services
501020 Wireless Telecom m unication Services
55 Utilities 5510 Utilities 551010 Electric Utilities
551020 Gas Utilities 551030 Multi-Utilities 551040 Water Utilities
551050 Independent Pow er and Renew able Electricity Producers
26 Table 5: OLS-regression results of the model including the dummy variables
Variable: Estimated
coefficient: Robust standard error:
∆𝑖𝑖𝑡𝑡𝑢𝑢 0,033 0,028 DummyHCR * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,008 0,038 DummyENE * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,006 0,037 DummyMAT * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,087** 0,035 DummyIND * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,113*** 0,036 DummyCOD* ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,145*** 0,046 DummyCST * ∆𝑖𝑖𝑡𝑡𝑢𝑢 0,002 0,038 DummyFIN * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,089* 0,048 DummyINT * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,227*** 0,070 DummyTEL * ∆𝑖𝑖𝑡𝑡𝑢𝑢 -0,076 0,050
The table shows the coefficients of the interaction effects of the regression of the following model:
𝑟𝑟𝑘𝑘,𝑡𝑡= 𝛽𝛽0+ 𝛽𝛽1∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽2𝐷𝐷𝐻𝐻𝐻𝐻𝐻𝐻∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽3𝐷𝐷𝐻𝐻𝐻𝐻𝐻𝐻+ 𝛽𝛽4𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽5𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸+ 𝛽𝛽4𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽7𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀+ 𝛽𝛽8𝐷𝐷𝐼𝐼𝐸𝐸𝐷𝐷∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽9𝐷𝐷𝐼𝐼𝐸𝐸𝐷𝐷 + 𝛽𝛽10𝐷𝐷𝐻𝐻𝐶𝐶𝐷𝐷∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽11𝐷𝐷𝐻𝐻𝐶𝐶𝐷𝐷+ 𝛽𝛽12𝐷𝐷𝐻𝐻𝐶𝐶𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽13𝐷𝐷𝐻𝐻𝐶𝐶𝑀𝑀+ 𝛽𝛽14𝐷𝐷𝐹𝐹𝐼𝐼𝐸𝐸∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽15𝐷𝐷𝐹𝐹𝐼𝐼𝐸𝐸+ 𝛽𝛽16𝐷𝐷𝐼𝐼𝐸𝐸𝑀𝑀∆𝑖𝑖𝑡𝑡𝑢𝑢 + 𝛽𝛽17𝐷𝐷𝐼𝐼𝐸𝐸𝑀𝑀 + 𝛽𝛽18𝐷𝐷𝑀𝑀𝐸𝐸𝑇𝑇∆𝑖𝑖𝑡𝑡𝑢𝑢+ 𝛽𝛽19𝐷𝐷𝑀𝑀𝐸𝐸𝑇𝑇+ 𝜀𝜀𝑡𝑡
* indicates that the coefficient is significantly different from 0 at a significance level of 10%, ** indicates significantly different from 0 at a significance level of 5% and *** indicates significantly different from 0 at a significance level of 1%.
Table 6: the results of the test on equal coefficients
Results Wald-test:
Critical region at 1% significance: (degrees of freedom 9;1130)
F > 2,42
F = 4,22
Probability > F 0,0000
Note: the F-test evaluated the following hypotheses:
𝐻𝐻0: 𝛽𝛽2= 𝛽𝛽4= 𝛽𝛽6 = 𝛽𝛽8= 𝛽𝛽10= 𝛽𝛽12 = 𝛽𝛽14= 𝛽𝛽16= 𝛽𝛽18 = 0
𝐻𝐻1: 𝛽𝛽2 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽4 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽6 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽8 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽10 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽12≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽14 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽16 ≠ 0 𝑎𝑎𝑎𝑎𝑎𝑎/𝑜𝑜𝑟𝑟 𝛽𝛽18≠ 0
27 Table 7: The response of stock indexes to unanticipated and anticipated Federal Funds target rate changes
Sector index: Unanticipated rate change coefficient Anticipated rate change coefficient R
2 Health Care 0,023 (0,024) (0,006) 0,005 0,0342 Energy 0,030 (0,023) -0,012** (0,006) 0,0581 Materials -0,050** (0,019) -0,015** (0,006) 0,1301 Industrials -0,079*** (0,023) (0,009) -0,003 0,2039 Consumer Discretionary -0,112*** (0,038) (0,011) -0,002 0,2593 Consumer Staples 0,035 (0,025) (0,005) 0,001 0,0595 Financials -0,054 (0,041) (0,017) -0,006 0,0529 Info Technology -0,196*** 0,066 (0,016) 0,007 0,3237 Telecom -0,043 (0,425) (0,014) 0,002 0,0389 Utilities 0,033 (0,028) (0,005) -0,003 0,0437
The robust standard errors are displayed between brackets below the coefficient. * indicates that the coefficient is significantly different from 0 at a significance level of 10%, ** indicates significantly different from 0 at a significance level of 5% and *** indicates significantly different from 0 at a significance level of 1%.
