Investors’ choices regarding company profits, bond yields and stock returns

Investors’ choices regarding company

profits, bond yields and stock returns.

Programme: Economie & Bedrijfskunde Track: Finance & Organisation

Supervisor: Robin Döttling

Abstract

This thesis is about how bond prices and company profits affect stock returns. Over the last years, stock prises kept increasing while the profitability of companies declined. The European Central Bank made some expansive monetary decisions which helps explaining why stock returns kept ri-sing when profits fell. With help of the three factor model from Fama & French (1993) a new model was set up to research the relationship in two time periods. These periods are compared and tested for differences. There is evidence that the influence of bond yields on stock returns significantly changed over time.

Introduction……… 3 Literature review……… 5 Methodology……….. 7 Results……….. 13 Conclusion……… 15 Appendix……….. 16 References……… 17

I. Introduction

Over the last ten years a lot has happened on the Dutch stock market, there were two crises, with the Financial crisis in 2007 and the Euro crisis following. Companies had trouble maintaining their profits, investors lost confidence in the stock market and stock prices fell (particularly in 2008) over 50%. However, from 2010 till 2014, stock prices of the AEX-index and the AMX-index went up by 20,1%. At the same time, the profits of these companies went down from 45.5 billion in 2010 to 35.5 billion reported over 2013. Which is a decline of 21,9%. This suggests that stock prices are not reflecting company profits anymore.

There is a fundamental relationship between share prices and company profits; when a com-pany generates more profits, its value increases. The value of a comcom-pany is calculated by multiply-ing its share price with the total shares outstandmultiply-ing. When assummultiply-ing that the total shares outstandmultiply-ing stays the same, the share price goes up when the company makes more profits. So higher profits implicates higher stock prices. But over the last 5 years, it looks like this is not the case anymore. The simple theory is that changes in real stock prices should be negatively correlated with changes in long-term interest rates, since the rate of discount has opposite effects on both (Schiller & Beltratti, 1992). In this paper I’m going to research if this relationship is disturbed and what factors could have caused this change in the relationship between profits and stock returns.

An instrument that investors use to forecast changes in stock prices and returns is the Pri-ce/Earnings ratio (P/E ratio). P stands for the market value per share, and E for the earnings per sha-re. A higher P/E ratio suggests that investors are expecting more earnings growth in the future com-pared to a lower P/E ratio. The P/E ratio is a simple measure that is used to assess whether a share is over- or under-valued based on the idea that the value of a share should be proportional to the level of earnings it can generate for its shareholders (Berk & DeMarzo, 2011 p. 33). In general, a high P/E suggests that investors are expecting higher earnings growth in the future compared to compa-nies with a lower P/E ratio. However it is not useful for investors to use the P/E ratio as a basis for their investment to compare as each industry has much different growth prospects which results in different P/E ratios.

The government bond markets in a monetary union act a lot like the banking system. When solvency problems arise in one country (e.g. Greece), bondholders, fearing the worst, sell bonds in other bond markets. This triggers a liquidity crisis in these other markets: investors, who sell, say, Spanish government bonds, use the proceeds to invest in other safe assets, for example German government bonds. As a result, liquidity is withdrawn from the Spanish money market, leading to a

liquidity squeeze making it impossible for the Spanish government to rollover the existing debt (De Grauwe, 2013). Because of the concern of investors about the sustainability of the debt of some of the Euro countries interest rates on these government bonds rose to the extent that these countries had to be helped out by the European Central Bank (ECB). In September 2012, the ECB decided to commit itself to provide unlimited (but conditional) liquidity support in the government bond mar-kets of the Eurozone in the context of its ‘Outright Monetary Transactions’ (OMT) program. This program certainly constituted a regime change in the Eurozone and contributed to the significant decline in interest rate spreads (De Grauwe, 2013). The ECB became a lender of last resort in the government bond market. Increasing liquidity on government bonds as there is no risk of insolven-cy anymore, this action decreases risk on bonds, increasing its price and decreasing its interest rate. Because of the low yields, it is impossible for investors to make a decent return on the government bond market. A low short-term interest rate makes risk-less assets less attractive and may lead to a search for yield by financial intermediaries with short-time horizons (Jiménez et al, 2014). They may search for this return on the stock market, driving prices up because of the increased demand.

An other factor that influences stock prices are the dividends paid. When a company pays out a dividend, the share price goes down by the same amount as the dividend because of the cash that is flowing out of the company to shareholders, reducing the value of the company resulting in a lower share price. This means that when a company changes his dividend policy, and starts paying out a bigger part of its net profits in dividends, the share price goes down while the net profits do not change.

Profits should have a positive relationship with stock returns and therefore it is interesting to see if company profits have a relationship with stock prices and returns and to see what that rela-tionship is. I want to link this relarela-tionship with the interest rate on government bonds. The interest rate on a 10-year Dutch government bond has declined from 4,09% in 2004 to 1,96% in 2013. That raises the question: are the low interest rates on bonds (partially) explaining the difference in the P/E ratios and the fundamental relation of profits and stock prices and do they matter in explaining stock returns? In this thesis I build on the Fama & French (1993) three-factor model and added pro-fits and yields to explain differences in stock returns. The main conclusion is that the factor yield on government bonds significantly changed over the years and that investors’ focus has shift from fun-damental profitability to interest rates on government securities.

There has not been done any research on the influence of company profits and government bonds on stock returns and the results could be interesting taking the Euro crisis into account. That makes this subject and this thesis interesting and that is what it is adding to science.

II. Literature Review

It looks like investors’ focus has shifted from profits to interest rates and liquidity. Research from Thorbecke (1997) concludes that positive monetary shocks increase stock returns which indi-cates that expansionary monetary policy exerts real effects by increasing future cash flows or by decreasing the discount factors at which those cash flows are capitalised. Results from size portfoli-os indicate that monetary shocks have larger effects on small firms than large firms. This evidence supports the hypothesis that monetary policy matters partly because it affects firms' access

to credit.

Work from Beaver & Morse (1978) proves that accounting methods can influence the P/E ratio and therefore, the P/E ratio is not very accurate in forecasting growth. P/E ratios do not correct for risk; P/E ratios can vary either positively or negatively with market risk depending on the mar-ket conditions in a given year, marmar-ket risk is of little assistance in explaining the observed persis-tence in price-earnings ratios over periods longer than two or three years (Beaver & Morse, 1978). P/E ratios are interesting to look at. But they should not be the leading subject of this thesis.

Eychenne, Martinetti & Roncalli (2011) have found some key results regarding government bonds. Forecasts of long-run returns are low in comparison to the last quarter-century. They obtai-ned an annualised return of 4.3% for US bonds and 4.0% for eurozone bonds at the 2050 horizon, a figure contrasting with the 9.0% annualised performance posted since 1980. Regarding equities, long-run risk premiums forecasts for 2050 are close to historical standards at 4.8% for US and euro-zone equities. Interesting is the forecasted decreasing return on euroeuro-zone bonds. It shows that the monetary actions of the ECB have lead to lower interest rates on government bonds

The study from Shiller & Beltratti (1992) says that a decrease in expected long-term bond yields would seem long-term bonds as a less attractive investment and so stock prices will rise because investors shift from bonds to stocks. Both markets have a negative correlation. Concluding from the study of Eychenne, Martinetti & Roncalli (2011), bond yields are declining, saying that prices of stock should increase. An other outcome from Shiller & Beltratti (1992) is that bond- and

stock prices do not correlate with changes in inflation rates. Therefore, it is not necessary to imple-ment inflation rates in the model.

It also might be possible that the low interest rate has influence on investment behaviour of companies. Penman & Zhang (2002) have studied the relationship between company investments and earnings. A low risk free interest rate usualIy gives the incentive to do more investments (net present value is higher when interest payments are lower). Growth in investment reduces reported earnings and creates reserves. Reducing investment releases those reserves, increasing earnings. If the change in investment is temporary, then current earnings is temporarily depressed or inflated, and thus is not a good indicator of future earnings (Penman & Zhang, 2002). Firms’ investments are not taken into account in the research of this thesis. This could be an interesting addition for further research.

The Fama & French Three-factor model (1992) is an important fundamental of this thesis. Their model factors explain average returns on stocks and bonds. There are three stock-market tors: an overall market factor and factors related to firm size and book-to-market equity. These fac-tors are important for this thesis because it helps reducing the omitted variable bias. Controlling for effects, other than the explanatory is important and the Fama & French Three-factor model does that very accurately.

The simplest models of time-varying discount rates tend to imply that the prices of bonds, land, and other assets (relative to their real payments streams) should covary closely with stock pri-ces, the prices of all of these assets being driven by a common underlying discount factor; the risk-less rate. Shiller (1982) reports, to the contrary, that there is little co-movement between the value of corporate equity and the prices of bonds and land, even after attempts to correct for movements in real coupons, dividends, and rents. Company profits have to be corrected for size (for a more ex-tensive explanation see the methodology section).

III. Methodology

In this paper, I’m researching companies listed on the dutch AEX index and the AMX in-dex. These are the 50 largest dutch companies listed on the Euronext Amsterdam. These two indices have 50 companies listed in total, but because I’m using the timeframe 2004-2014, ten of them have

incomplete data because these companies have not been listed long enough to have all the data avai-lable that is needed for this research. That gives the dataset 40 companies.

It is better to compare the total return of shares with company profits instead of comparing share prices. When using total returns, the dividends paid are a component of the dependent varia-ble, which is better in trying to explain why company earnings and bond yields are less/more of in-fluence on stock prices. Besides, investors are not interested in stock prices, but more interested in stock returns.

To understand what the influence of earnings is on shareholder returns, shareholder returns itself has to be defined. Shareholders have two kinds of income from shares: capital gain and divi-dends. Capital gain is the difference between the sale price and purchase price for the stock,

P1 - P0 (Berk & Demarzo 2011 p. 253). When a dividend is paid, cash flows from inside the

com-pany to the shareholders, reducing the value comcom-pany and thus, reducing its share price with the exact amount of the dividend that is paid per share. The formula to calculate the total shareholder return is:

Total Return per share = (P1 - P0) + D

P1: Price of the share at the end of the year

P0: Price of the share at the beginning of the year

D: Dividend paid per share

When dividing the total return per share by P0 the annual return will be displayed as a

per-centage. This is needed to correct for different share prices. A one euro return on a stock that costs €100 is a return of only 1%, but on a stock that is priced at €1, it is a return of 100%. In the appen-dix a figure is added to show the relationship between return in currency and return as a percentage. The total return of a share (%) is the dependent variable yt of the model. Stock prices and dividends

per share over the period 01/01/2004 - 01/01/2014 are be found on Datastream. As for the explana-tory variables I have YIELD, which is the yield on a dutch government bond with 10 years to matu-rity and PROFIT, which defines the profit that was made by a company in a particular year. Net profits of companies can be found in the Bureau van Dijk database, and yields on government bonds with 10-years to maturity are found in the database of the Dutch State Treasury Agency.

Bond Price =

F: Face Value t: Time to maturity YTM: Yield to maturity

The yield on a bond is dependent on the current interest rate, price and time until expiration. When the current interest rate drops, your bond has a bigger interest rate compared to the bonds that are issued now, so the price of the bond rises and the yield declines.

For a larger company it is easier company to make higher profits compared to smaller com-panies, but that does not mean that this bigger company outperforms smaller ones because it has more assets, so somehow I have to take this into account. There is a scatterplot added in the Ap-pendix (figure 1) which shows company profits each year. It is clear that the variable PROFIT has to be adjusted for size, because now the estimates are biased towards the profits of Royal Dutch Shell. The market value of a company is:

Market value can’t be used to correct for firm size because the term market value correlates with the dependent variable stock return. Total asset value is also unusable because some industries are very capital intensive, which means that they have more assets on their balance sheet compared to their profits. The term I’m using to correct for size is shareholders equity. The formula for total share-holders equity is:

Total Shareholders Equity = Assets - Liabilities

When a firm goes bankrupt today, shareholders equity is the amount of money that shareholders will get after the firm pays his debt. This variable will give the least biased results. I’m searching for shareholder return and total shareholders equity is interesting for investors that are looking for a return on the stock market.

In the end, I want to conclude that the focus of investors has shifted from company profits to yields on government securities over time. So the basic relation that I want to research is:

ytc = β0 + β1 ･ PROFITct + β2 ･ YIELDct + αi + � (1)

yt is defined in two periods: 2004-2008and 2009-2013. I’m comparing the two time periods to see

if there are any changes in the factors influencing returns on stock. But as stated above, the variable PROFIT has to be adjusted for size, which done by dividing PROFIT with Total Shareholders

Equity. Total shareholders equity is found in Datastream. This creates a new variable;

PRO-FIT_EQUITY. So the new relationship that has to be researched is:

ytc = β0 + β1 ･ PROFIT_EQUITYct + β2 ･ YIELDct + αi + � (2)

It is important to include control variables in a regression. A control variable is not the ob-ject of interest in the study; rather it is a regressor included to hold constant factors that, if neglec-ted, could lead the estimated causal effect of interest to suffer from omitted variable bias (Stock & Watson 2012 p. 272). As for controlling for omitted variable bias, the total market return, minus the risk free interest rate is included as the variable mrkt. There is extensive evidence that the level of the risk free rate predicts the expected return on common stock (Barsky, 1989). Also the SMB and the HML factors from the Fama & French three-factor model (1993) are included. The variable SMB (small minus big) is meant to mimic the risk factor in returns related to size. It is the differ-ence, each year, between the simple average of the returns on the three small-stock portfolios (Small/Low, Small/Medium, and Small/High) and the simple average of the returns on the three big-stock portfolios (Big/Low, Big/Medium and Big/High). Thus, SMB is the difference between the returns on small- and big-stock portfolios with about the same weighted-average book-to-market equity. This difference should be largely free of the influence of BE/ME, focusing instead on the different return behaviours of small and big stocks. The variable HML (high minus low) is meant to mimic the risk factor in returns related to book-to-market equity, is defined similarly. HML is the difference, each year, between the simple average of the returns on the two high-BE/ME portfolios (Small/High and Big/High) and the average of the returns on the two low- BE/ME portfolios (Small/Low and Big/Low). The two components of HML are returns on high- and low-BE/ME portfolios with about the same weighted-average size. Thus the difference between the two returns should be largely free of the size factor in returns, focusing instead on the different return behaviours of high- and low- BE/ME firms. (Fama & French 1993). The variables mrkt, SMB and HML for Europe can be found in the data library of Kenneth R. French. Unfortunately, these factors are not available for dutch portfolio’s and computing them myself would be very

time-consuming, that means that the European factors are the most accurate available. Including the con-trol variables, the model looks like this:

ytc = β0 + β1 ･ PROFIT_EQUITYct + β2 ･ YIELDct + β3 ･ SMBct + β4 ･ HMLct + β5 ･ mrktct + αi + �

(3)

This dataset is a panel dataset. Panel data refers to data for n different entities observed at T different time periods. (Stock & Watson 2012 p. 390) There are 40 companies, so the number of entities is n = 40, where each entity is observed in T=10 time periods (each of the years

2004,…2014). There are no missing values so this dataset is a balanced panel, which means that total # observations = 10 x 40 = 400. Regression software typically computes the OLS fixed effects estimators in two steps. In the first step, the entity-specific average is subtracted from each variable. In the second step, the regression is estimated using “entity-demeaned” variables. Specifically, con-sider the case of a single regressor in the version of the fixed effects model and take the average of both sides (Stock & Watson 2012 p. 398):

The fixed effects regression model: Yit = β1 ･ Xit + αi + ui

Take the average on both sides:

or:

where:

Thus β1 can be estimated by the OLS regression of the “entity demeaned” variables Yit on Xit. In fact,

this estimator is identical to the OLS estimator of β1 obtained by estimation of the fixed effects model

(Stock & Watson 2012 p. 399). Fixed effects regression is a method for controlling for omitted va-riables in panel data when the omitted vava-riables vary across entities but do not change over time. Fixed effects regression can be used when there are two or more time observations for each entity. The fixed effects regression model has n different intercepts, one for each entity. These intercepts can be represented by a set of binary (or indicator) variables that differ from one entity to the next but are constant over time (Stock & Watson 2012 p. 396)

fo-with β2004-2008 the estimator YIELD in the period 2004 - 2008 and β2009-2013 the estimator YIELD in

the period 2009 - 2013. To test the null hypothesis a dummy variable needs to be added to separate the two periods (2004 - 2008 & 2009 - 2013). This dummy variable is called TIME and is 0 from 2004 until 2008 and 1 from 2009 until 2013. Secondly the variable TIME_YIELD is added, this variable is the product of YIELD and TIME. When the variable TIME_YIELD is significantly dif-ferent from 0, the conclusion can be made that the variable YIELD has changed over the two peri-ods:

ytc = β0 + β1 ･ PROFIT_EQUITYct + β2 ･ YIELDct + β3 ･ SMBct + β4 ･ HMLct + β5 ･ mrktct

+ β6 ･ TIMEct + β7 ･ TIME_YIELDct + β8 ･ TIME_PROFITct + αi + � (4)

Summarising, I am doing three regressions. Regression (3) is done over period 2004-2008 and 2009-2013 and it explains what the influence of profits and bond yields have on stock returns in the two periods. I am comparing the two periods and try to explain differences with economic theo-ry. Regression (4) is made to statistically prove that one of the variables significantly changed over the two time periods.

IV. Results

The dependent variable is total return (%), which is defined as the total return on a stock per year. PROFIT_EQUITY is annual company profits divided by shareholders equity. YIELD is the yield on a government bond with 10 years to maturity. SMB is included to mimic the risk factor in returns related to size. HML is included to mimic the risk factor in returns related to book-to-market equity. mrkt is the market return minus the risk free interest rate. TIME is a dummy variable that is 1 for each year after 2008 and 0 for 2004, 2005, 2006, 2007 and 2008. TIME_PROFIT is the varia-ble TIME multiplied by PROFIT_EQUITY and TIME_YIELD is TIME multiplied by YIELD. All standard errors are robust to heteroskedasticity.

Sample 2004-2008 Sample 2009-2013 Sample 2004-2013 2004-2013 with dummy

PROFIT_EQUITY .300 -.0389 .2589 .4558 (1.28) (-0.12) (1.17) (1.58) YIELD .0506 .0541 .0427 -.1877 (0.29) (0.87) (1.74)* (-2.06)** SMB -.1600 .0094 .1794 .0097 (-2.77)** (2.85)** (7.33)*** (3.01)** HML .2883 -.0128 -.0010 .0072 (2.88)** (-1.47) (-0.46) (1.31) mrkt -.0084 -.2405 .0072 .0062 (-1.61) (-4.03)** (9.08)*** (7.56)*** TIME -1.0367 (-2.55)**

Sample 2004-2008 Sample 2009-2013 Sample 2004-2013 2004-2013 with dummy TIME_PROFIT -.1017 -(0.41) TIME_YIELD .3238 (3.07)** Observations: 200 199 399 399 R2 0.5340 0.5166 0.4589 0.4910

Panel data regression with fixed effects and robust standard errors

Constant included. Absolute value of t-statistic in parenthesis *Significant at 10%; ** Significant at 5%; *** Significant at 1%

The results show that PROFIT_EQUITY always has a positive influence on the return, ex-cept for in the sample 2009-2013, but the estimate is highly insignificant (t = -0.12). This is in line with theory, which says that higher profits should mean higher returns. An interesting observation is that the coefficient of PROFIT_EQUITY is smaller in the second sample, even though it is highly insignificant, it still suggests that company profits are of less influence on stock returns in the pe-riod 2009-2013. Another outcome is that bond yields and stock returns are having a positive rela-tionship. The estimations from the subsamples are not significant, but do suggest that the relations-hip is at least zero. The second sample is far more significant, which means that the chance of this estimator being zero is less. Bond yields have a bigger influence on stock returns in the last 5 years, the estimator TIME_YIELD is significant at the 5% level. The estimate of TIME_PROFIT is nega-tive, which suggests that the estimator PROFIT_EQUITY is smaller in the second period TI-ME_PROFIT is not significant at the 10% level, so that concludes that the influence of company profits has not changed significantly in the last five years.

Standard errors are robust, results from the modified Wald test for groups wise

hete-roskedasticity supports the alternative hypothesis that hetehete-roskedasticity is present. Results from the Hausman test (Prob>chi2 = 0.0368) rejects the hypothesis that the difference in coefficients are not systematic. So fixed effects estimation should be used.

From the results of the model and the underlying economic theory I can conclude that yields on Dutch government bonds have had more influence on stock returns in the last five years. This is significant at the 5% level. The theory suggests that lower interest rates on bonds should drive pri-ces of stock up. And this is exactly what is happening. Because of the monetary policy of the Euro-pean Central Bank, bond yields are on their lowest point ever, and because of that it is impossible

for investors to get a decent return on the bond markets. Because of this investors are investing mo-re in stocks. Driving the prices of stock up. The outcome of this model shows that the low bond yields are responsible for the increase in prices on the stock market. It is not proven that investors are neglecting economic fundamentals such as the profitability of a firm but the model suggests that company profits are of less influence of determining stock prices. However, the results are not sig-nificant. Further research is necessary to statistically prove this assumption.

V. Conclusion

The main goal of this thesis is to find a relationship between company profits, government bond yields and stock returns and research if their relationships have changed over the years. I have analysed three samples, one over 2004-2008, one over 2009-2013 and one over 2004-2013. First I have analysed the influence of bond yields and company profits over time, secondly I have added a dummy variable to statistically prove that at least one of these variables has changed over time. This thesis proves that the relationship between bond yields, profits and returns has changed over the last 5 years. Investors are not looking as much at the fundamentals of companies as they did before. Because of the monetary policy of the European Central Bank, supplying unlimited liquidity to countries in the EU, the Dutch government bonds are very safe which means that their yields are very low. This means that investors are searching for returns on the stock market, driving market prices up. One of the limitations of this thesis are the lack of diversification, companies scattered over more countries would give a more accurate and more significant result for the European Union as a whole. Now it fixates on one country, and it is not a very important economy like the U.S or Germany's. Investors also anticipate on forecasts, when forecasts are good, stock prices go up. This thesis is too small to search further into this topic, but it is interesting to implement this in further research. It is also interesting to implement P/E ratio’s more to see if they have risen and that inves-tors are now paying more for a stock with the same returns as 5 years ago.

Government bonds are considered to be risk-free. It could be interesting for further research to see if investors are more risk-averse after the financial crisis than they were before. It is common knowledge that excessive risk taking has caused the financial crisis but it would be a nice addition. Most of the timeframe I used was pre- mid- or post crisis, it is interesting how returns are when the world is “normal” and compare that to this period.

VI. Appendix

e

Figure 1: Company profits over the time period 2004-2013

Figure 1 is showing that entity 20, which is Royal Dutch Shell generates far more profits compared to other companies in the sample. In order to get unbiased estimation results, this has to be cor-rected.

VII. References

Brennan, Schwartz & Lagnado, Strategic asset allocation, Journal of Economic Dynamics and

Barsky, Why don't the prices of stocks and bonds move together, American Economic Review, 79 (1989), pp. 1132–1145

Campbell, Measuring the persistence of expected returns, American Economic Review, 80 (1990), pp. 43–47

Fama & French, Business conditions and expected returns on stocks and bonds, Journal of

Finan-cial Economics, 25 (1989), pp. 23–49

Beaver & Morse, What Determines Price-Earnings Ratios?, Financial Analysts Journal, 34 (1978), pp. 65-76

Shiller, R. "Consumption, Asset Markets, and Macroeconomic Fluctuations." CarnegieRochester

Conference Series on Public Policy, 17 (1982) pp. 203-238

Shiller & Beltratti, Stock prices and bond yields: Can their comovements be explained in terms of present value models?, Journal of Monetary Economics, 30 (1992), pp. 25-46

Penman & Zhang, The Accounting Review, 77 (2002), pp. 237-264

De Grauwe, The European Central Bank as Lender of Last Resort in the Government Bond Markets

CESifo Economic Studies, 59 (2013), pp 520-535

Jiménez, Ongena, Peydró & Saurina, Hazardous times for monetary policy: what doe twenty-three million bank loans say about the effects of monetary policy on credit risk taking? Econometrica, 82, (2014), pp. 463–505

Thorbecke, On Stock Market Returns and Monetary Policy, The Journal of Finance, 52, (1997), pp 635-654

Eychenne, Martinetti & Roncalli, Strategic Asset Allocation, Lyxor White Paper Series, 2011 Berk & DeMarzo, Corporate Finance, Pearson, (2011), pp. 1001