Print Reference This Reddit This

Literature Review On Internal Performance Of A Firm

Introduction

The survival of the firms’ depends upon their internal performance that gets reflected in the market. To define the internal performance, firms need to generate enough cashflow to finance its operational activities as well as satisfy and meet the demands of shareholders. With daily business activities, there occurs various cost of running the firm viz. cost of goods sold, operating cost, sales and administrative cost, interest and tax expenses, financing cost of shareholders and other related expenses. Aggregately, these costs are referred as cost of capital, in other words; more appropriately, these various costs are financed by investors that is, shareholders and lenders who expect the rate of return higher with respect to their investment made in the firm.

Since the cost of anything is a price one must pay, cost of capital defines a cost as the return that a company assures to provide in order to receive capital from the market (Pratt and Grabowski, 2008). A firm would or is not in position to set its cost of capital, in order to do so, it must reflect the market, congregating its position in the market and determining its performance adequacy.

Further, broadly, the firms can raise their finances by common shares (equity finance) and bonds (debt finance) in proportion to create the shareholder value and market value. The mixture of equity and debt finance is called as the capital structure (Modigliani and Miller, 1958). The corporate decisions undertaken by firm decide its capital structure. Therefore, the cost of each financing asset is given a proportional weight, altogether referred as weighted average cost of capital (hereafter, WACC). By taking a weighted average, firms can see how much cost the company has to pay for every dollar it finances (Christensen et. al., 2010).

On a note that the cost of capital or opportunity cost is equivalent to the risk-free rate plus a risk premium which reflects managed risk. That is, the rate used for discounting cash flows permits to incorporate two fundamental dimensions in the analysis of financial decisions: the time value of money and risk. In this manner, the risk-free rate serves for the incorporation of the time value of money and an adjustment (risk premium) is incorporated in order to recognize the flows uncertainty and volatility (Villarreal and Córdoba, 2010):

Opportunity Cost = Risk-free rate + Risk premium

Corporate firms need to make wise decisions should be able to embark upon difficulties or fluctuations like:

Where will the investment cash come from?

What are the opportunity cost (cost of capital) associated with the investment?

What is estimated net present value of the cash flows?

What is the projected risk associated in financing?

A fundamental tool in determination of future cash flow is the discount factor also called as opportunity cost which acts as measure of financing. This technical term, discount factor, in business terms is called as weighted average cost of capital (Hewlett, 2007).

The basic equation by which WACC is estimated is as stated below (Jung, 2007)

Weighted Average Cost Of Capital (WACC)

Where

Re = cost of equity 

Rd = cost of debt 

E = market value of the firm's equity 

D = market value of the firm's debt 

V = E + D 

E/V = percentage of financing that is equity 

D/V = percentage of financing that is debt 

Tc = corporate tax rate 

Source: The flows of investment and the required return on investment, Jung 2007

The above figure depicts two independent views that actually are functionally and financially interconnected viz.

Operational management

Corporate Managers (CFO, Chief Finance Officer and CEO, Chief Executive Officer)

Operators receive the capital investment become important entities for generate the returns on assets and thus monitoring the cash and profit margins. Whereas the performance of top management, which acts as umbrella for monitoring firm activities, mainly on focuses decisions made in financing and resource allocation that helps financial intermediaries and institute bridging the gap between the investors and operators (Jung, 2007).

As noted, decisions made act as the important base to analyse and forecast the cash flow as appropriately as it can be. Therefore, there should be suitable methodologies or techniques in analysing the cash flow and appropriately examining and calculating the opportunity cost associated with the financing of the projected cash flow.

Further, this article discusses the significance of WACC, and specially observing and analysing the WACC factor in emerging economies because of high developmental opportunity and the ways or the techniques used to determine WACC as being discussed in the so far literature.

As discussed by Ghersi and Sabal (2006) and Bodie et. al (2005), the financing for large scale projects has augmented over last four decades in emerging markets spreading over in wide varieties of potential projects. Thus, investment characteristics observed in case of undertaking projects in emerging markets as below:

As the financing is “made to measure”, its structuring tends to be costly, and therefore is only justifiable for large-scale projects

The bulk of the investment is aimed at tangible assets

The totality of the project’s assets are pledged to financial creditors

High leverage is usually employed

No taxes on returns and no transaction costs.

All investors are rational mean-variance optimizers.

Investments are limited to a universe of publicly traded financial assets, such as stocks and bonds, and to risk-free borrowing or lending arrangements.

Investments are usually long-term (e.g., 20 years)

The only purpose of the financing is to complete the project, and as such it has a limited lifetime

What measure of discount factor to undertake?

Based on above characteristics, there are various factors to be considered upon which companies decide the discount factor or the WACC for carrying out business in the emerging markets. Following are the three types of WACC measures undertaken (Boere, 2006):

Corporate WACC: First type argues in favour to use single WACC measure such that a multinational company can be viewed as a portfolio of multiple (global) investments and therefore each investment can be treated with the same opportunity cost of company which reflects the total aggregated portfolio risks. This approach recognises the advantage of a multinational, which is helps to diversify country specific risks when volatilities in different countries are partly off-set by each other due to their low correlation activities. The appropriate WACC for operating in an emerging market could be achieved by adjusting the corporate WACC for the marginal contributing effect of the operation in the emerging market (based on the specific financial and operational leverage). When calculating of the nominal WACC in the foreign currency, a compensation for the different levels of inflation between the home and foreign currency will have to be added or deducted.

Stand alone and local WACC: The second method is to consider each investment project as an individual investment and valuing each of them according to a local WACC that reflects the risks of the local country and project. In that respect there is one major argument that demonstrates the need to calculate an individual WACC for an emerging market - emerging markets are, to some extent, non-integrated markets (not integrated with the global market). It is therefore said to be a segmented market. The characteristic of a segmented market is that real returns (compensated for different levels of inflation) are also determined by domestic risk factors. These are characterized by inefficiencies caused by regulatory, legal and tax barriers in emerging markets. These inefficiencies have an impact on the cost of equity. In such a case, the company determines a local project WACC, with a local cost of equity the measure of a country’s equity risk levels for the operations in the emerging markets, rather than the corporate WACC.

Middle-of-the-road/Foreign WACC: This method acknowledges the need to account for the additional sovereign risk factors in the country of the investment in the WACC. This is achieved by simply adding a sovereign risk premium to the corporate WACC to offset the cost associated with it. Sovereign risk represents the country risk and the credit risk of the country. Simply put, the sovereign risk premium is the difference between the yield of the risk-free triple-A rated government bond and a bond issued by the local government (with the sovereign risk embedded in it) minus the inflation differential of the two currencies involved. If local bonds are issued in US$, then inflation differential is not deducted.

WACC Valuation Methodologies in Emerging Market

The notion of WACC is straightforward – net operating cash flows (or forecasted) generated by investing capital into project should at least provide return equal to market determined financing cost (Galiniene and Butvilas, 2010). WACC consist of two major cost components – Cost of Equity and Cost of Debt (Hewlett, 2007). There are different valuation methodologies developed my theorists that helps to understand the significance of WACC in different economies. Some of these important methods are discussed in further part of this section.

Modified Capital Asset Pricing Model

The CAPM is derived from the capital market which is the form of discount factor valuation model. It attempts to provide a measure of market relationships based on the theory of expected returns with respect to single investment risk measure. CAPM model is based on quantifying the systematic part of risk component as it assumes that investors will eliminate unsystematic risk by holding large and diversified portfolio (Hitchner, 2006).

The cost of equity component under calculation of WACC, which is usually computed by using CAPM, is used for discounting free cash flows of equity (FCFE) (Pereiro, 2002).

With CAPM, expected risk premium on stock = beta x expected risk premium on market.

R = Rf + β (Rm –Rf)

Where:

R is expected risk premium (expected return on asset)

Rf is the risk-free rate of interest local to country,

Rm is the return on market

β (Beta) = Cov(Ri, Rm)/σ2m (that is - covariance between the market and stock i, divided by the corresponding market variance. σ2m = Market Variance. Beta can be estimated by regression analysis (Bender and Ward, 2009).

According to CAPM, the cost of equity is equal to the risk-free rate with risk premium. This premium is reflected by the sensitivity of returns of the company’s shares to the market return (the beta factor) and the so-called systematic risk premium (the difference between the average market return and the risk-free rate) (Pereiro, 2002).

Nevertheless, for beta to be reliable this approach requires (Sabal, 2004 and Estrada, 2007)

the local industry stock to be very liquid

having a history of public trading

free flow of information among the buyers and sellers where equilibrium is achieved through series of transaction

Regrettably, these conditions are rarely met in emerging markets. Thus, there is need to develop certain adjustments in case of emerging economies which are primarily based on non-existence of an efficient market and more important, an unsystematic risk or the country risk. This risk represents a non-diversifiable risk, additional to market risk. It is a marginal risk and not a consequence of the associated flows volatility of a particular business, which is traditionally captured in the systematic risk (CAPM-β) (Villarreal and Córdoba, 2010).

With the aim of incorporating this additional source of risk, the corporate finance practitioners have focused on a country risk measure (as unsystematic risk) in an attempting to capture the impact of several EM conditions (such as expropriation by local governments, unstable rule of law, and lack of transparency) on the value of business activities, and on expected returns from investment (García-Sánchez et. al., 2010).

The most popular proxy for country risk has been the sovereign bond spread—that is, the spread between sovereign bonds issued by the emerging economy government and, say, the U.S. treasury. Therefore, even though most studies have suggested a variety of different amendments to the CAPM systematic risk estimate (i.e., beta), most practitioners tend to add a CR measure to the Global CAPM formulation as follows (Benserud and Austgulen, 2006 and Pereiro, 2003):

Cost-of-equity capital = Ke = Rf + βL * (Rm - Rf)

Rf = Rfg + Rcu

Where:

Rf = Local risk free rate

Rfg = Global risk free rate

Rcu = Country risk premium (unsystematic)

βL = Local asset beta with the local market

Rm = Local market return

βL * (Rm-Rf) = Risk premium asset

Several empirical studies have clearly shown that the effect on stock returns, while incorporation of country risk is frequently more sizable than the industry effect. In other words, stock performance seems to be much more integrated to the local volatility of the economy than to the fluctuations and trends of the corresponding industry at the international level.

Note: Upper curve: yield of Argentine PAR bond. Lower curve: yield of U.S. T-bonds. Difference between both (dark strip) is the Argentine country-risk premium

Source: Argentina: Evolution of the Country Risk Premium (Pereiro, 2002)

The country risk premium may reach substantial values in emerging markets. As an illustration examined in above exhibit shows the evolution of the premium for Argentina. The altitude fluctuates as the country risk premium is tightly related to both political and economic factors (e.g., the implementation of a state reform within the former category, and financial crises in Mexico, Asia, Russia, and Brazil within the latter). From the theoretical perspective, adding a country risk premium implies a de-facto use of a multifactor risk-return model, where the premium corresponds to the local country’s idiosyncratic risk.

Godfrey-Espinosa Model

Godfrey and Espinosa identify three types of risks affecting investments in emerging markets: political risk, business risk, and currency risk. As in the previous models, currency risk is accounted for by selecting a base hard currency (e.g., the U.S. dollar) whereas the other two types of risks are incorporated into the discount rate by modifying the basic CAPM (Sabal, 2004).

This model proposes two adjustments with respect to the CAPM. First, it adjusts the risk-free rate by the yield spread of a country relative to the U.S. and second, measures risk by adjusting beta defined as 60% of the volatility of the stock market of the country in which the project is based relative to the volatility of the US market (σc/σus) (Estrada, 2000).

R = (Rf +Yc) + βadj * {(0.60) * (σc / σus)}

Where σc and σus are the standard deviation of returns of country c’s and that of the US market and Yc is yield spread of a country relative to the U.S. market.

The above empirical deduction offers an understanding to the difficulties observed in calculation of historical risk premium in markets where long time series of returns are hardly available. With Portuguese data, for instance, it is (still) impossible to calculate historical risk premium over a, say, 20-year horizon, as a proxy of risk free rates is not available (Alpalhao and Alves, 2005).

Similarly, as in case of Asian economies the largest data financial collected are of the return on equities and risk-free assets are considered. Even for these economies there are differences in the data available and quality; shorter time series, less depth and breadth for the individual markets, and the possibility of greater regulation. These factors all contribute to the difficulties in estimating the estimating risk premium (Cohen, 2009).

The Godfrey-Espinosa model deals with these above problems pragmatically, by using the available long time series (either US or UK) to estimate risk premium in emergent markets. The argument is as follows: “operating risk” in every economy can be measured indirectly, through the compared volatility of domestic stock markets and the benchmark stock market (US or UK) (Alpalhao and Alves, 2005).

The above evidence discussed in their Godfrey and Espinosa (1996) paper where they construed that emerging market betas are, on average, lower than developed market ones, but that volatility of emerging markets is significantly greater that directly affects the risk premium.

Note that in this model the specific nature of the project is ignored. Put differently, it makes no difference whether a company is evaluating a project in the oil, airline, or telecommunications industries; what matters is the country in which the project is based (Estrada, 2007).

Estrada Model

The importance of an appropriate identification of the issues that determine the cross section of stock returns in emerging markets can hardly be acknowledged. Companies evaluating projects and investors evaluating companies in emerging markets need to discount expected cash flows at a risk-adjusted rate. Hence, they need to identify the variable(s) that determine such discount rates.

The framework proposed in this model enables an understanding of the discount rate applied in cross sectional returns in emerging markets. Furthermore, it is grounded in modern portfolio theory. The significance of this model is that it can be applied both at the market level by the investors and at the company level in evaluating the projected cash flow. It tries to outcast the limitation of subjective measures of the risk, it can be fine-tuned to any desired benchmark return, and it captures the downside risk that investors want to avoid.

The main results reported in Estrada (2000) can be summarized as follows. In emerging markets:

Stock returns are correlated to systematic risk measured by beta

Stock returns are correlated to total risk measured by the standard deviation

Stock returns are correlated to downside risk measured by the semi-deviation with respect to the mean, by the downside beta, and by VaR (Value at Risk)

Costs of equity based on the semi-deviation seem to be “more plausible” than those based on systematic risk or total risk

As known, the required rate of return comprises of a risk-free rate and a risk premium, such that the former is a compensation for the expected loss of purchasing power and the second is an extra compensation for bearing risk.

Additionally, consider a U.S.-based internationally-diversified investor; therefore, the risk-free rate should compensate this investor for the dollar’s expected loss of purchasing power, and the risk premium should compensate the investor for the risk of investing in the world market portfolio. Hence,

Ri = Rf + (Rpw) * (Rmi)

where

Ri is the required return

Rfus is the (U.S.) risk-free rate

Rpw is the world market risk premium

Rmi is a risk measure

i is a cross-sectional index

Therefore, the model proposed by Estrada (2001) focuses on risk measures based on downside risk and, in particular, based on the semi-deviation of returns with respect to the mean defined as

CE = Rfus + (Rmg −Rfg) * Rmi

Where Rmi is a downside risk measure that is, the ratio between the semi-standard deviations of returns with respect to the mean in market i and the semi-standard deviation of returns with respect to the mean in the world market.

Thus, Estrada (2001) manifests in case across emerging markets that

Downside risk measured by the semi-deviation with respect to the mean does explain the cross section of stock returns

Downside risk measured by the semi-deviation with respect to the risk-free rate and with respect to 0 does not explain the cross section of stock returns

Pereiro (2006) argues that Estrada (2001) model satisfies the risk perceptions of investors, as it renders cost of equity figures that are halfway between the results obtained with the standard CAPM (normally too small in the view of practitioners), and with respect to the larger figures obtained with total risk methods (which may be perceived as way too high by practitioners), the model would better reflect the partial integration under which many emerging markets operate.

Lessard model

Offshore projects in case of emerging economies are categorized as more risky due to the high volatility trading in the market. The presence of greater array of risks that are (perceived as being) primarily of a downside nature, such as currency inconvertibility, expropriation, civil unrest, and general institutional instability creates that hinders the shareholder value. Further, because such risks are relatively unfamiliar to the investing companies, the companies are likely to make costly errors in early years and to require more time to bring cash flows and rates of return to acceptable steady-state levels (Lessard, 1996 and Pereiro, 2006).

To deal with these higher risks and greater unfamiliarity, many companies include an extra premium in their discount rate and, particularly, in case of emerging-market projects. However, the basis for these discount rate adjustments is often subjective. Such adjustments do not properly reflect objective information available about either the nature of these risks, or about the ability of management to manage them. Nor do they take into account the reality that the risks stemming from unfamiliarity fall over time as the firm progresses along the learning curve (Lessard, 1996).

This model tweaks an additional factor in measuring risk similarly to, but also somewhat differently from the CAPM. More precisely, it proposes measuring specific risk as the product between a project beta (βp) and a country beta (βc) (Li and Hoyer-Ellefsen, 2008); that is,

Risk factor = βp * βc

Where

βp is the risk of the industry

βc is the risk of the country in which the company will invest.

In this approach the required return on equity when investing in industry p located in country c is given by (Estrada, 2007)

CE = Rf + Rc + βp * βc (Rm −Rf)

Where

Rf is the risk-free rate,

Rc a country risk premium that includes the chance of ex-proprietary actions, payment difficulties and other risks,

βp the country beta (relative sensitivity of the returns of the local [emerging] stock market to the U.S. market returns) and

βc is the beta of a project that is comparable to the offshore project

The project beta and the country beta can be estimated as the beta of the relevant industry with respect to the US market.

Goldman Sachs Model

Mariscal and Hargis (1999) offer an overview of the approach used by Goldman Sachs to estimate the cost of equity in emerging markets. Their approach, somewhat similar to that of Godfrey and Espinosa (1996), adjusts risk-free rate and the risk premium, the latter by replacing beta by the standard deviation as the appropriate measure of risk in emerging markets. In their framework, discount rates are driven by global (or relative to US), country, and firm-specific factors (Bruner et. al. 2002).

More precisely, Goldman Sachs proposes replacing the fixed adjustment of 0.60 by one minus the observed correlation between the stock market and the bond market of the country in which the project is based Mariscal and Hargis (1999). In other words, the investment bank proposes estimating the specific risk as

Risk factor = (1–ρsb) * (σc/σus)

Where ρsb denotes the correlation between the stock and bond markets of country c relative to US.

In this approach, the required return on equity when investing in country c is given by

CE = Rf + Rc + [(σL/σus) * β * (Rm −Rf)](1− ρsb)+ Rid

Where

Rf is risk-free rate,

Rc a country risk premium,

σL the standard deviation of returns in the local market,

σus the standard deviation of returns in the U.S. equity market,

β the beta of the target local company computed against the local stock market index,

ρsb the correlation of dollar returns between the local stock market and the sovereign bond used to measure country risk and

Rid is an idiosyncratic risk premium related to the special features of the target firm (e.g., specific firm credit rating as embodied in its corporate debt spread, industry cyclicality, percentage of revenues coming from the target country and other related risk).

The intuition behind this more sophisticated double counting adjustment is as follows. If the stock market and the bond market are perfectly correlated (ρsb=1), they both reflect the same sources of risk; in that case Rc will capture c will capture call the relevant risk of investing in country c and, therefore, R = Rf + Rc (Estrada, 2007).

If, on the other hand, the stock market and the bond market are uncorrelated (ρSB=0), each reflects different sources of risk; in that case, Rc quantifies the risk reflected by the bond market, σc/σus the additional risk reflected by the US the additional reflected by the US stock market. For all practical purposes it is the case that 0<ρsb<1. Therefore, this model incorporates the risk reflected by both the stock market and the bond market without double counting sources of risk (Estrada, 2007).

Salomon Smith Barney Model

This model measures the correlation of the risk of investing in a specific industry with the risk of investing in a specific country having characteristics of both the project and the company considering the investment. The first adjustment is implemented by incorporating the project beta (βp) and the other two by incorporating an adjusted political risk premium (Estrada, 2007).

Zenner and Akaydin (2002) suggested that there is adverse effect of political events and the associated pressure on cash flows. That is, if it is hard to quantify the political events then the expected cash flows losses will be too high. In that case, the expected cash flow is discounted at the cost of capital adjusted for financial and business risk and adding a political risk premium to the discount rate. This approach assumes that potential cash flow losses are correlated with the political risk premium estimates.

To estimate the political risk premium, we use four different sources of information:

Sovereign bond spreads

S&P country sovereign debt ratings

Country ratings from Institutional Investor

Euro-money magazines and Macroeconomic variables

and make some further adjustments to the raw political risk premium estimates (Zenner and Akaydin, 2002 and Bruner, 2002).

The sensitivity analysis carried out by Zenner and Akaydin (2002) conferred that under the reasonable assumptions, averagely, a common political risk premium adjustment of around 5% implicitly corresponds to an expected probability of the loss of all cash flows due to 50% of political intervention. This probability is significantly higher than what most investors anticipate for most investments in emerging markets.

Technically, to understand the process of incorporation of political measures, there are three factors which need to be adjusted for political risk premium: A company’s access to capital markets, the susceptibility of the investment to political risk, and the financial importance of the project for the company. More precisely, in this model,

A = {(γ1 +γ2 +γ3) / 30} * Yc

Where each γ coefficient is measured on a scale from 0 to 10.

In particular,

γ1 captures a company’s access to capital markets, with 0 indicating full access and 10 indicating no access;

γ2 captures the susceptibility of the investment to political risk, with 0 indicating no susceptibility to political intervention and 10 indicating maximum susceptibility;

γ3 captures the financial importance of the project for the company, with 0 indicating that the project involves a small proportion of the company’s capital and 10 indicating a large proportion.

Note that γ1 will be low for large international companies and high for small undiversified companies. Also note that γ2 is directly related to the probability of expropriation; hence it will be high for projects in industries that are likely to be expropriated (such as natural resources) and low for projects in industries where that is unlikely (such as retail). Finally, γ3 will be low for large companies investing in relatively small projects and large for small companies investing in relatively large projects (Estrada, 2007).

It should be clear that the sum of the γ coefficient will vary between 0 and 30, which in turn implies that the adjustment to the yield spread will vary between 0 and 1. As a result in the worst-case scenario A=Yc and in the best-case scenario A=0.

To illustrate, a large international company investing a small proportion of its capital in an industry unlikely to be expropriated would have to make no adjustment for political risk (A=0); a small undiversified company investing a large proportion of its capital in an industry likely to be expropriated, in turn, would have to incorporate a full adjustment for political risk (A=Yc).

This model further proposes quantifying the specific risk with the project beta which, as mentioned above, is the beta of the relevant industry with respect to the global market.

In sum, according to this approach, the discount when investing in industry p located in country c is given by (Pereiro, 2006)

CE = Rf + Yc * [(γ1 + γ2 + γ3)/30] + βlg (Rmg −Rfg)]

Where

Rf is the risk-free rate of the home country of the multinational corporation doing the valuation

βlg is a Global CAPM beta for the industry of the investment, properly levered for the financial structure of the target

Yc is the ‘c’ country yield spread relative to US market.

It is important to keep in mind that in this model, the required return on equity for a specific project in a specific country may be different depending on the company that considers the investment. In other words, according to the above models discussed, given the project and the country in which the project will take place, the required return on equity would be the same regardless of the company that considers the project; in this approach, in contrast, that is not necessarily the case (Estrada, 2007).

Damodaran model

The model particularly focuses on estimation of risk premium over equity cost. It is derived based on modified approach over historical risk premium methodology. That is, The actual returns earned on stocks over a long time period is estimated, and compared to the actual returns earned on a default-free (usually government security). The difference, on an annual basis, between the two returns is computed and represents the historical risk premium (Damodaran, 2006).

While users of risk and return models may have developed a consensus that historical premium is, in fact, the best estimate of the risk premium looking forward, there are surprisingly large differences in the actual premiums we observe being used in practice that are evident comparing various financial databases. Therefore, it is difficult to predict the risk premium when the markets with short and volatile histories (Johnson and Soenen, 2009 and Damodaran, 1999).

These short comings are amended with the indulgence of country risk premium and the β of the market. The basic proposition that the risk premium in any equity market can be written as:

Equity Risk Premium = Base Premium for Mature Equity Market + Country Premium

The most general, and our preferred approach, is to allow for each company to have an exposure to country risk that is different from its exposure to all other market risk. We will measure this exposure with l, and estimate the cost of equity for any firm as follows

CE = Rf + Rc * γ + [β * (Rm −Rf)]

Where γ is a firm specific exposure to country risk ranging from zero to one, and BLL is the local company beta computed against a local market index. The exposure factor γ could be, for instance, the percentage of revenues to the parent firm coming from the local or emerging market (Estrada, 2007).

Thus the equity cost can be priced and valued depending upon the expected growth, cash flows which are reflected in the risk premium with respect to firm specific that reflect the current market perceptions and not relying on historical knowledge.

Need help with your literature review?

Our qualified researchers are here to help. Click on the button below to find out more:

Literature Review Service

Related Content

In addition to the example literature review above we also have a range of free study materials to help you with your own dissertation: