Determinants of commodity price risk exposure in the restaurant industry: An analysis by commodity price cycles

International Journal of Hospitality Management 45 (2015) 121–129

Contents lists available at ScienceDirect

International Journal of Hospitality Management journal homepage: www.elsevier.com/locate/ijhosman

Determinants of commodity price risk exposure in the restaurant industry: An analysis by commodity price cycles Chun-Hung (Hugo) Tang ∗ School of Hospitality and Tourism Management, Purdue University, 900 West State Street, West Lafayette, IN 47906, United States

a r t i c l e

i n f o

Keywords: Commodity price risk Equity exposure Restaurant Operating leverage Financial leverage

a b s t r a c t The objectives of the present study were to (1) investigate the level and the extent of commodity price risk exposure in the restaurant industry and (2) identify the determinants of risk exposure. The risk exposure was estimated by 60-month rolling regressions based on equity returns. The determinants of equity risk exposure were proposed based on a discounted cash flow model. The results found that 35.39% of sample restaurant firms are exposed to commodity price risk. The level of equity risk exposure was estimated to be 1.148 during commodity price booms and 1.031 during slumps. Empirical testing was consistent with the model prediction that operating leverage and financial leverage are effective tools in managing risk exposure, but the effects are asymmetric during commodity price booms and slumps. Financial leverage was found to be more effective than operating leverage. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction Commodity prices are considered to be a major source of business risk (Bartram, 2005). This is especially true for the restaurant industry because food costs on average accounts for 33% of the revenue (Food Prices and Small Businesses, 2008) and agricultural commodity prices have been rising and becoming volatile in recent years. Food prices have increased by 2.8% per year on average for the past 10 years (Dreibus et al., 2014) and agricultural commodity prices are becoming volatile due to climate change, disease, and rising global demand (Thorn, 2014). Considering that the revenue of the U.S. restaurant industry is estimated to be $683 billion in 2014 (National Restaurant Association, 2014), a one-percent increase in food commodity prices would lead to more than $2.25 billion additional cost for the industry. Commodity price is likely continue to be a major issue for restaurateurs given that approximately one billion people are moving from poverty into the world of consumerism (Woolley, 2010). While many restaurant companies have acknowledged commodity price risk as one of the major risk factors, the industry has little leeway to increase average check under unfavorable employment and economic environments (Jargon and Spector, 2011). When increasing prices is not an option, managing variable costs such as food cost proves to be a better profit lever compared

∗ Tel.: +1 765 409 6197; fax: +1 765 494 0327. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ijhm.2014.12.005 0278-4319/© 2014 Elsevier Ltd. All rights reserved.

to cutting fixed costs or pumping up volume (Marn and Rosiello, 1992). The restaurant industry has a wide range of strategies to manage food cost, from inventory control to menu design and financial hedging. However, it is challenging to identify a strategy that works for all types of restaurants due to the inherently diverse nature of the restaurant industry. For example, smaller or independent restaurants can quickly revise their recipes and menu to avoid using costly ingredients, but it would be a logistical nightmare for large chains like McDonald’s. Financial hedging, for another example, is not applicable to all restaurants. For example, Starbucks can use coffee futures to hedge away coffee bean price uncertainty, but Buffalo Wild Wings cannot adopt the same strategy because there is no financial derivative for bone-in chicken wings (Jannarone, 2011). Although the risk is imminent and the impact is substantial, commodity price risk has not attracted much attention from hospitality management researchers (Hesford and Potter, 2010). In the hospitality management field, there have been studies on interest rate risk (Singh, 2009), exchange rate risk (Lee and Jang, 2011), and real estate risk (Lee and Jang, 2012), but as of yet there has been no commodity price risk study. In the financial risk management literature, most studies are based on financial risks such as interest rate risk and exchange rate risk (Bartram, 2005). These studies’ implications for commodities risk management are limited because commodity price risk is closely tied to operating activities rather than financial decisions. Commodity prices are also more volatile than exchange rates, interest rates, and stock market indices (Bartram, 2005). Commodity price risk appears to be an

122

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

important issue in the restaurant industry that deserves a thorough investigation. Among the few existing commodity price risk studies, most focused on asset-induced risk exposure. For example, gold mining companies are exposed to gold prices because they own gold mines (Tufano, 1998). Few of the studies (Carter et al., 2006; Loudon, 2004) looked at operating activities-induced risk exposure (e.g., airline companies’ exposure to jet fuel price). However, the findings of these studies may not be readily applicable to the restaurant industry for two reasons. First, empirical evidence has shown that commodity price risk exposure is contingent on the type of commodity and the nature of the industry (Bartram, 2005). Second, existing theoretical models (Brennan and Schwartz, 1985; Tufano, 1998) treat commodity price risk as an output risk that affects revenue and asset value. But commodity price risk is an input risk that affects costs. For example, in gold mining firms, the output is gold. The volatility of gold prices affects the revenue and the value of the gold mines. In the restaurant context, food commodities, such as beef and flour, are inputs of the production. Input risks are different from output risks in that cash flow volatility could be affected by operating activities even after financial hedging. For example, a burger chain can use beef futures to financially hedge beef price risk. But the cash flows from selling burgers are still affected by operating activities such as contract price, production waste and pricing. As production process is industry specific, industry-specific studies can provide relevant and accurate information to industry practitioners (Loudon, 2004). When a company identifies a commodity price risk management strategy, the effectiveness of the strategy is likely to vary with the economic environment. Studies (e.g., Fabozzi and Francis, 1978; Jagannathan and Wang, 1996) showed that systematic risk is time varying. Since commodity futures price is a function of commodity cash price and systematic risk (Bailey and Chan, 1993), commodity price risk exposure could be time varying as well. In order to account for this time-varying nature, the present study investigated the exposure to commodity price risk by commodity price cycles. This study aims to contribute to the literature by (1) assessing the extent and level of commodity price risk exposure (the risk exposure hereafter) in the restaurant industry and (2) developing and testing an economic model that describes the effects of cost structure on the risk exposure. The present study focuses on equity risk exposure because corporate decisions should be made for the purpose of creating shareholder wealth (Ross et al., 2013), and strategies that affect equity value are most likely to be implemented. Given that commodity price risk is closely related to operating activities, the present study is expected to contribute to the literature by providing guidelines in using operating leverage to manage the exposure to commodity price risk. 2. Theoretical background and model development In economics, commodity is defined as a class of goods that have no qualitative differentiation across a market. Commodities that have related financial derivatives traded in exchanges and overthe-counter include agriculture (e.g., corn), non-precious metals (e.g., aluminum), precious metals (e.g., gold), and energy (e.g., crude oil; Bartram, 2005). In this study, the term “commodity” refers to agricultural commodities that are commonly used in restaurants. 2.1. Commodity price risk exposure in the restaurant industry Risk exposure is different from risk. Risk exposure is “what one has at risk” (Adler and Dumas, 1984, p. 42). In the commodity

price context, risk is the volatility of commodity prices and risk exposure is a firm’s value sensitivity to price changes. Technically, commodity price risk could be measured by the standard deviation of commodity price. Exposure to commodity price risk is commonly represented by the coefficient of regressing a firm’s stock returns on commodity price changes. Restaurants are in the business of using agricultural commodities to produce food to serve customers, so commodity prices would affect the production cost. If restaurants could raise the prices immediately to offset the increase in food cost without losing customers, the risk is passed to the customers. Restaurants would not be exposed to the price risk. In reality, no restaurant can completely pass cost increases to customers without sacrificing the market share. This leads to a situation that restaurants’ food cost tracks the changes of commodity prices quickly but the selling price reacts slowly, or not at all. This asymmetric response speed of food cost and selling price creates exposure to commodity price risk (Blake and Mahady, 1991; Jargon, 2012). This nature makes restaurants very different from commodity producers (e.g., gold mining firms), whose production cost is not necessarily tied to the commodity price. Commodity producers are exposed to commodity price risk through their assets (e.g., gold mines). To manage risk exposure, a restaurant can resort to financial hedging or operational hedging. Financial hedging aims to mitigate the effect, not the source, of the risk exposure. Starbucks’ use of coffee futures to lock in the coffee price (Jargon, 2011) is an example of financial hedging. In contrast, operational hedging addresses the risk exposure directly. For example, revenue management based on local currencies can reduce the risk exposure to the foreign exchange rate in a multinational hotel company (Chang, 2009). Both financial hedging and operational hedging could be considered as the efforts to align the response speeds of revenues and costs. For instance, marketing initiatives that reduce customers’ price sensitivity could speed up revenue responses. Financial hedging or fixed price contracts could slow down cost responses. For small firms, financial hedging may not be feasible due to the lack of expertise, financial resources, and economy of scale (Haushalter, 2000). As a result, most restaurants resort to operational adjustments to absorb or reduce the impact of commodity price risk. Operational hedging, in addition to financial hedging, could be an effective risk management tool for restaurants for three reasons. First, commodity price risk is an input risk. There are many opportunities to manage the risk exposure in the production and selling processes. Second, the firm’s expertise in the operation could help it to manage the risk exposure arising from its operations (Bartram, 2005). Third, financial hedging cannot address demand uncertainty and is very costly for long-run risk exposure (Chowdhry and Howe, 1999). Given the above reasons, operational hedging becomes an attractive alternative for firms with limited resources and financial expertise. Miller (1992) summarized operational hedging strategies into the following: (1) vertical integration by acquiring vendors, (2) increasing bargaining power against suppliers, and (3) cooperation with vendors through long-term contractual agreements. Increasing the flexibility in sourcing (e.g., multiple suppliers) also allows the firm to be more resilient to fluctuations of input prices (Aaker and Mascarenhas, 1984) and indirectly contributes to the bargaining power. However, the above categorization did not consider direct adjustments of operational activities as a way to manage the exposure to risk sources. For firms whose risk sources are closely tied to the operation, such as commodity price risk to restaurants, operational hedging could be an effective approach because operators can leverage their expertise in operational activities. Among all possible operational adjustments, this study aims to investigate the effect of operating leverage and financial leverage on the exposure to commodity price risk.

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

2.2. Model development Departing from the inductive approach that is commonly used in the hospitality literature, the present study adopted a deductive approach that is popular in the finance and economics fields. In these fields, studies often start with the mathematical model of a fundamental theory and deduct the relationships between key variables (i.e. hypotheses) through the manipulation of mathematical equations. There are two advantages for this approach. First, it explicit incorporates the key variables in the economic model. The interpretation of the directional effects of the variables is dictated by the mathematical equation, leaving no room for ambiguity. In this study, this approach allows direct incorporation of costs and revenue in the model. Second, the deductive approach is effective for studying new topics that do not have a huge body of literature. It allows the researchers to advance the frontier of the existing literature. The limited number of commodity price risk studies, especially in the hospitality management field, warrants this approach. There are two potential starting points to construct the economic model: the fixed production model and the flexible production model (Tufano, 1998). The fixed production model can be easily incorporated with the discounted cash flow (DCF) model to arrive at the present value of the firm. The flexible production model is close to the real business situation, where firms can suspend production, change output quality, or change the rate of production. Tufano (1998) recognized the limitation of a fixed production model and compared the statistics of the fixed production model to those of Brennan and Schwartz’s (1985) real options model that allows flexible production. Tufano (1998) concluded that, first, firms with a flexible production schedule should have smaller betas (i.e. risk exposure) than firms with a fixed production schedule because the firm can optimally exploit the option to adjust the production. Second, the beta is a decreasing function of the volatility of the commodity price. Other than these two points, the flexible production model and the fixed production model yield the same predictions regarding the direction of the cost components’ effects on risk exposure. Since the present study aims to identify the directional effects of the determinants, not to predict the level of risk exposure, a model based on Tufano’s (1998) approach will be sufficient for this purpose. Tufano’s (1998) model was developed for gold mining firms exposed to commodity price as an output risk. The present study modified the model for the input risk in the restaurant context. Consider a restaurant that produces to meet its demand (Q). The production incurs fixed costs (FC) such as overhead and interest expenses, as well as variable costs (VC) such as food cost and direct labor. The restaurant sells its product at a market price (P) per unit. Asset value of a firm can be presented as the present value of future incomes, a discounted cash flow model: V=

N  [Q (P − VC) − FC](1 − ) i=1

(1 + )i

,

(1)

where r = firm’s cost of capital,  = corporate tax rate, and N = number of operating periods. A firm’s asset exposure to commodity price risk (asset beta) can be defined as: ˇA ≡

∂V/V C ∂V = V ∂C ∂C/C

(2)

Since restaurants commonly produce to meet the demand regardless of the commodity price (C), Q is independent of C. FC is also independent of C. P is assumed to be independent of C because companies facing competition have limited ability to pass cost increases to customers (Blake and Mahady, 1991; Jargon, 2012). A survey conducted by Gaiotti and Lippi (2008) also showed that

123

more than half of restaurants hold their prices stable for more than one year. VC mostly consists of food cost and direct labor, and is therefore a linear function of commodity price. Based on above assumptions, Eq. (2) can be represented as: C ∂V ˇA ≡ = V ∂C

N

−C · Q (1 − ) V

1 i=1 (1+r)i

.

(3)

Assuming a fixed production schedule (Tufano, 1998), Eq. (3) can be represented as: ˇA ≡

−C · Q C ·Q = . FC + Q · VC − Q · P Q (P − VC) − FC

(4)

It can be inferred from Eq. (4) that the relationship between asset beta (ˇA ) and C is a reciprocal function centered at the breakeven commodity price. When C increases, the size of positive risk exposure will decrease, while the size of negative risk exposure will increase. A restaurant’s variable cost largely consists of food costs, so CQ and QVC are highly correlated and cannot be interpreted separately. A good approach is to consider CQ and QVC as direct functions of commodity price for a given level of Q. In that case, asset beta should respond to the food cost the same way as to the commodity price. Asset beta is expected to correlate negatively with food costs. Holding everything else constant, asset beta is a reciprocal function of fixed cost centered and the breakeven fixed cost. Fixed cost is expected to have a negative relationship with asset beta. Following the same fashion, asset beta is a negative reciprocal function of revenue (QP). The relationship between asset beta and revenue is expected to be positive. 2.3. Equity’s exposure to commodity price risk Asset beta determinants are necessary but not sufficient conditions to determine equity’s exposure to commodity price risk (equity beta). Financial leverage needs to be considered. The risk of equity comes from two sources: business risk and financial risk (Ross et al., 2013). Business risk is the risk inherent in a firm’s operations and is manifested through the firm’s asset exposure to the risk source. This is the risk presented by ˇA in Eq. (4). Financial risk is determined by the firm’s capital structure. A company with a high debt-to-equity ratio is considered to have high financial leverage. For an all equity firm, the financial risk component is zero. Financial risk increases as the level of the debt. Based on M&M Proposition II (Modigliani and Merton, 1963) and the Capital Asset Pricing Model (CAPM) (Sharpe, 1964), equity beta can be represented as ˇE = ˇA (1 + D/E), where D is debt and E is equity. Equity beta is a function of asset beta but the relationship is moderated by financial leverage. Specifically, given a constant level of asset beta, equity beta will increase as financial leverage increases. Therefore, financial leverage enters the model as a determinant of equity beta. It should be noted that financial leverage does not necessarily moderate the relationships between asset beta and firm characteristics identified in Eq. (4) because ˇA is merely defined by, not equivalent to, the function of the determinants. Financial leverage moderates the collective effect of all terms in the function. From the relationships discussed, two commodity price risk management tools emerged: operating leverage and financial leverage. Here are some examples of increasing operating leverage to reduce risk exposure. A restaurant can invest in fixed assets, such as storage facilities and production equipment, to reduce the loss and waste of food commodities. The savings would lead to decreased sensitivity to commodity prices. A restaurant can also spend money in marketing programs to build up customer loyalty, thereby reducing customers’ price elasticity of demand. This will give the restaurant flexibility to adjust selling prices to reduce its

124

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

cash flow sensitivity to commodity price. The depreciation of facilities and marketing expenses are fixed costs that increase operating leverage. Alternatively, a restaurant can reduce its equity beta by decreasing financial leverage. With a lower level of debt, equity would become less sensitive to business risk, resulting in a lower equity beta (Sharpe, 1964). Furthermore, a high level of financial leverage has been shown to lead to underperformance in the product market (Campello, 2006). Reducing financial leverage would not only decrease equity’s exposure to risk, but also allow underperformance in the product market to be avoided.

index tracks the prices of nearby futures contracts and comprises the following index components: wheat, corn, soybeans, cotton, sugar, coffee, cocoa, feeder cattle, live cattle, and lean hogs. As futures contracts do not entail actual delivery of commodities and can be implemented right away with little up-front costs, futures markets generally react faster than spot markets (Silvapulle and Moosa, 1999). In other words, futures behave in line with stocks and are an effective tool to measure equity sensitivity to commodity price. Hernandez and Torero (2010) found a strong correlation between futures prices and spot prices for food commodities.

2.4. Varying exposure to commodity price risk

3.2. Measuring equity’s exposure to commodity price risk

The present study investigated the exposure to commodity price risk by commodity price booms and slumps. Intuitively, the exposure could be different because capable restaurant managers would have adjusted the operation to take advantage of commodity price drop and avoid the impacts of commodity price increases. Theoretically, the risk exposure was expected to be different through the commodity price cycles because risk exposure to systematic is time varying. Since Fabozzi and Francis’s (1978) study, subsequent researches have provided evidence that the exposure to systematic risk is time varying (Jagannathan and Wang, 1996). Exposure to commodity price risk is likely to be time varying as well because commodity futures price is a function of commodity cash price and systematic risk (Bailey and Chan, 1993). As the systematic risk component is time varying, the exposure to commodity price risk will also be time varying. Empirically, Cashin et al. (2002) showed that cycles are a dominant feature of commodity prices. During the cycles, commodity futures returns (Gorton and Rouwenhorst, 2005) and price volatility (Pindyck, 2001) vary over time. Bartram’s (2005) study of non-financial firms’ exposure to commodity price risks also indicated that commodity price risk exposure is not stable over time.

Ordinary least square (OLS) regression has long been used to estimate risk exposures (for examples, see Singh, 2009; Lee and Jang, 2012). This approach implicitly assumes that the risk exposure is stable over time. Use of the OLS involves a risk of veiling the time-varying nature of commodity price risk exposure. Two ways to address this issue are time series models based on Bollerslev’s (1986) generalized autoregressive conditional heteroscedasticity (GARCH) framework and rolling window regression. Groenewald and Fraser (2000) showed that rolling window regression dominates simple time series models in forecasting betas. Considering computational costs, this study adopted the rolling window regression approach as in Groenewald and Fraser (2000). The month m risk exposure is the regression coefficient based on a window from m − 59 to m. Following the commodity price risk exposure literature (Bartram, 2005; Petersen and Thiagarajan, 2000; Tufano, 1998), this study uses the two-factor model to estimate restaurant firms’ commodity price risk exposure:

3. Methodology There were two stages of analysis, namely estimations of equity beta and evaluation of the effects of risk exposure determinants. In the first stage, equity betas were estimated monthly and converted into quarterly betas. In the second stage, the quarterly betas were regressed on the determinants identified in Sections 2.2 and 2.3 with two control variables. 3.1. Data The sample is consists of publicly traded U.S. restaurant firms (SIC code 5812 and 5813) from 1990 to 2012. This period covers the latest three business cycles defined by National Bureau of Economic Research. With this data range, the analysis results would not be limited to certain economic situations. Data were collected from three sources, as follows: CRSP for monthly returns of the market and sample firms; Datastream for Agriculture and Livestock Spot Index (S&P AL index); and Compustat for quarterly accounting data. To be included in the sample, the firms need to appear in both CRSP and Compustat datasets. Ideally, the risk exposure should be estimated with individual commodities such as wheat, corn, or cattle. However, the large number of commodity types and high correlation among commodities (Bartram, 2005) render this approach infeasible and methodologically inappropriate. Therefore, this study adopts Standard & Poor’s Agriculture and Livestock Spot Index from the S&P Goldman Sachs Commodity Index family as a proxy of commodity price. This index was originally developed by Goldman Sachs, but is now owned and published by Standard & Poor’s. The

Rjt = ˇ0j + ˇmk RMKt + ˇcm RCMt + εjt , where Rjt is the monthly stock return of company j in quarter t, RMKt is the equal-weighted market return in period t, and RCMt is the monthly percentage change in commodity price index. Coefficient ˇcm is firm j’s exposure to commodity price risk in month t. The quarterly betas used in the second stage analysis are the average of the monthly betas that are significant at 0.1 level within the quarter. The present study follows Singh (2009) in adopting the 0.1 significance level to avoid losing too many observations. 3.3. Measuring the determinants and control variables As a result of algebraic derivation, both food costs (CQ) and variable costs (VC) are in Eq. (4). Because the variable cost is mostly consisted of food costs in the restaurant industry, including the food cost and the variable cost in the same regression model would cause multi-collinearity. Therefore, an adjustment was needed to translate the economic model into a testable statistical model: using cost of goods sold as a proxy to account for both the food cost and the variable cost. Fixed costs are calculated as the sum of depreciation and interest expenses. The depreciation expense reflects the level of fixed assets, and thus operational leverage. Both cost of goods sold and fixed costs are common-sized by revenue, measured in millions. Commodity price is the quarterly average of the S&P AL Index’s monthly prices. Financial leverage is measured by the ratio of long-term debt to total assets. The present study employs two control variables: Tobin’s Q and restaurant type. Tobin’s Q, a ratio of the market value to the replacement cost of assets, is a commonly used proxy for market premium or growth opportunities. It is included as a control variable for two reasons. First, Froot et al. (1993) pointed out that firms with higher growth opportunities (i.e., high Q) have stronger incentives to reduce their risk exposure. Second, the measurement of beta is

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

based on stock market returns. It would be plausible to control for the market’s perception on the firm’s future prospects. The calculation of Tobin’s Q is based on Chung and Pruitt (1994). The present study also controls for the effect of restaurant type (limited service versus full service) on risk exposure. Full-service restaurants are defined as those with a NAICS code of 722,511, 722,514, 722,110, or 722,212. Limited-service restaurants are defined as those with a NAICS code of 722,513, 722,515, or 722,211. SIC was used to download the sample firms because its inclusion of restaurant companies is more comprehensive than NAICS. But SIC does not specify restaurant types so NAICS was used for restaurant classification. Compared to limited-service restaurants, full-service restaurants usually attract customers who are dining for the experience, as oppose to utility dining. Therefore, the demand for full-service restaurants would be relatively inelastic and give operators flexibility in passing on the price to consumers. This could be viewed as an indicator of the effect of consumer price elasticity. 3.4. Dating commodity price cycles The analyses have been conducted by commodity price cycle, investigating boom periods (trough-to-peak) and slump periods (peak-to-trough). Cashin et al. (2002) outlined a dating procedure based on an early business cycle study (Bry and Boschan, 1971). The procedure was based on monthly data. In the present study, the procedure has been modified for quarterly data as follows: Step 1. Selecting initial peaks and troughs. A peak (trough) is defined as the quarter with a price higher (lower) than the lead and lag quarter; Step 2. Enforce alternation of peaks and troughs; Step 3. Eliminate peaks and troughs within four quarters of both ends of series; Step 4. Eliminate peaks (troughs) at both ends that are lower (higher) than values closer to the end; Step 5. Merge cycles (peak-to-peak and trough-to-trough) that are less than eight quarters long; and Step 6. Merge phases (peak-to-trough and trough-to-peak) that are less than four quarters long.

125

Fig. 1. Quarterly average S&P AL Index and commodity price exposure statistics. Note: S&P AL Index and commodity price exposure (Beta) are quarterly average based on monthly data. Only betas significant at 0.1 level were included in the calculation. Beta is scaled to 100. Solid (dashed) vertical reference lines indicate the peaks (troughs) of commodity price cycles.

significant betas. The raw betas were skewed to the right, so they were log transformed for normality before they were entered as the dependent variable. The independent variables were the beta determinants and control variables discussed in section 3.3. Mills’ lambda was not significant in commodity price booms (−0.052, z = −0.31), slumps (−0.275, z = −1.36), or the whole sample period (−0.087, z = −0.73). This indicated that there was no sample selection bias and it was not necessary to run the sample selection step. In such a case, the model was reduced to an OLS regression model with the transformed beta as the dependent variable. As a robustness check, the Heckman model and the OLS model produced the same set of significant variables and the directions of the significant coefficients were the same. The size of the significant coefficients was also very close. Therefore, the present study adopted the OLS model for empirical testing of the proposed model. 4. Results and discussion 4.1. Commodity price risk exposure in restaurants

The quarters with commodity price peaks are: 1996Q2, 2001 Q1, 2004Q2, 2008Q3, and 2011Q1. The quarters with commodity price troughs are: 1992Q3, 1999Q3, 2002Q2, 2005Q3, and 2009Q3. 3.5. Statistical model Out of 2399 firm-quarter observations, only 303 (12.63%) have risk exposure at the 0.01 level. However, a Tobit regression would be inappropriate because the zero values are not due to censoring, unobservable true values below zero. A good example would be a thermometer that has a lower bound of zero reading. Any temperature below zero will be unobservable and recorded as zero. In the present sample, the zeros are observable true values (i.e., no risk exposure). In such a case, a sample selection model such as Heckman’s two-step model (Heckman, 1979) would be an appropriate choice because it accounts for the possible sample selection bias (i.e., selecting only exposed firms out of the whole sample). In the present study, we first tested the existence of sample selection bias based on the Heckman’s two-step model. The first step utilized all observations. The dependent variable was a binary variable (exposed firms versus non-exposed firms), and the independent variables were revenue, cost of goods sold percent, total fixed cost, and Tobin’s Q. The independent variables were chosen because they were significantly different between exposed firms and non-exposed firms in commodity price booms and slumps (Table 2). The second step utilized only the observations with

A total of 15,165 firm-month betas were estimated. Nine hundred and fifteen (6.03%) betas were significant at 0.1 level and 401 betas (2.64%) were significant at 0.05 level. To match the beta frequency with the quarterly accounting data, monthly betas significant at 0.1 level were averaged by quarter and company. After merging the two datasets and deleting beta outliers exceeding three standard deviations, there were 53 out of 139 firms with significant risk exposure in at least one quarter. It is worthwhile to note that all drinking places observations (SIC 5813) were excluded in this step. The regression results in Table 3 are pertaining to restaurants only, not drinking places. Fig. 1 shows that both the extent (number of firms with significant exposure) and level (mean of significant betas) of equity beta fluctuate over time. The percentage of firms with significant betas ranges from 0% to 27.5% (1993Q2), peaking in 1993 and 1994. Average beta ranges from 0.24 to 1.68, peaking around early 1990s. Both the level and extent of risk exposure started to decrease after 2006. This is consistent with the results of previous studies that the risk exposure is dynamic, validating the necessity of estimating risk exposures by commodity price cycles. As a reference, Bartram (2005) found that the percentage of non-financial firms with significant commodity price risk exposure ranged from 4.5% to 15.9%. The quick decline of risk exposure since 2006 coincided with the trend of rising commodity prices. This is consistent with the two variables’ negative correlation, indicated in Eq. (4). It also suggests

126

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

Table 1 Quarterly average exposure to commodity price risk. (A) Summary statistics of betas significant at 0.1 level

Number of observations Mean Standard deviation Maximum Median Minimum

Boom periods

Slump periods

171 1.104 0.560 2.623 0.942 0.345

145 1.026 0.727 4.387 0.816 0.394

(B) Number of firms with significant betas Significance level

# of firms

Mean ˇ

# of firms

Mean ˇ

10% level 5% level 1% level

49 (73) 33 (89) 9 (113)

1.026 1.098 0.949

52 (80) 34 (98) 9 (123)

1.104 1.006 0.906

Note: In the column “# of firms,” the values without parentheses are the number of firms with at least one quarter of significant beta. Values in the parentheses are the number of firms without any significant beta in the sample period.

that restaurants might have engaged in some form of operational or financial hedging after 2006. Panel A in Table 1 presents a summary of significant quarterly betas by commodity price cycles. Average betas were 1.104 for the boom periods and 1.026 for the slump periods. The difference was not significant at the 0.05 level. All estimated betas

were positive. This contradicts the expectation that rising commodity prices would adversely affect restaurant profits and result in a negative correlation. Lombardi (2013) explained that the negative beta expectation may not be plausible because it is based on the assumption that commodity prices are driven exogenously. However, commodity prices are likely to be driven by macroeconomic factors, as equity prices are. For example, commodity price increases are often driven by the rising demand due to booming economic activities, which also drive the increase of equity prices. Although Gorton and Rouwenhorst (2005) demonstrated a negative correlation at the aggregate level between a basket of commodity futures and Stand & Poor’s 500 index, the present study provides evidence to support that the nature of risk exposure varies by industries and is contingent on the type of commodities. Specifically, in the restaurant industry, the equity’s exposure to commodity price risk is positive because equity and commodity markets are driven by the same fundamental economic activities. Panel B summarizes the level and extent of risk exposure by significance level and cycle. There is no significant difference between commodity price booms and slumps in terms of the level and extent of risk exposure. 4.2. Exposed firms versus non-exposed firms Table 2 presents the comparison of key variables between firms with significant risk exposure (exposed firms) and firms without

Table 2 Descriptive summary of firms by exposure and commodity price cycle. (A) Boom periods

ˇCM Revenue Cost of goods sold Cost of goods sold % Fixed cost Fixed cost % Leverage Tobin’s Q Full-service

Firms with sig. ˇCM (a)

Firms with insig. ˇCM (b)

T test (a) − (b)

1.104 (0.560) 345.17 (906.80) 236.45 (546.07) 79.34 (10.35) 26.24 (73.94) 6.62 (3.27) 0.342 (0.273) 1.345 (0.697)

N.A. 246.39 (534.34) 182.59 (448.04) 75.95 (15.54) 15.18 (33.20) 6.68 (2.77) 0.354 (0.321) 1.543 (0.905) 54 out of 75

3.05*** 2.48** 5.90*** 6.16*** −0.43 −0.899 −5.62*** z = 0.18

Firms with sig. ˇCM (c)

Firms with insig. ˇCM (d)

T test (c) − (d)

1.026 (0.727) 327.51 (916.00) 221.70 (541.63) 78.89 (11.52) 25.34 (74.97) 6.72 (3.77) 0.318 (0.242) 1.315 (0.844) 38 out of 53

N.A. 207.94 (505.86) 147.59 (365.70) 75.90 (19.51) 12.50 (29.49) 7.64 (23.00) 0.362 (0.363) 1.522 (1.007) 61 out of 85

3.71*** 3.67*** 4.23*** 5.18*** −1.27 −3.18** −5.02*** z = −0.01

(B) Slump periods

ˇCM Revenue Cost of goods sold Cost of goods sold % Fixed cost Fixed cost % Leverage Tobin’s Q Full-service

(C) T test results of boom periods versus slump periods

ˇCM Revenue Cost of goods sold Cost of goods sold % Fixed cost Fixed cost % Leverage Tobin’s Q Full-service

Mean difference (a) − (c)

Mean difference (b) − (d)

0.078 17.68 14.74 0.45 0.90 −0.09 0.024 0.030 z = 0.20

N.A. 35.00 28.23 0.05 2.68** −0.96 0.008 0.022 z = 0.03

Note: Revenues, cost of goods sold, and fixed cost are in millions. Percentages are based on revenues. Fixed cost = depreciation + interest expenses. The “full-service” row presents the number of firms. Values in columns titled “Mean difference” represent the differences between two group means. * 0.05 level. ** 0.01 level. *** 0.001 level.

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129 Table 3 Regression results of exposures determinants. Y = ln(ˇCM )

Pooled

Boom period

Slump period

Revenue Cogs % FC % S&P AL Leverage Tobin’s Q Full-service Intercept F-statistic Adjusted R2 Obs.

−0.266 (0.036)*** −0.347 (0.211) −1.651 (0.479)** −0.001 (0.001) 0.973 (0.152)*** −0.122 (0.041)** −0.091 (0.060) 0.485 (0.261) 20.64*** 31.28% 303

−0.297 (0.056)*** −0.348 (0.230) −1.369 (0.987) 0.0003 (0.001) 1.290 (0.280)*** −0.080 (0.076) 0.048 (0.089) 0.047 (0.372) 13.52*** 34.28% 169

−0.165 (0.054)** 0.951 (0.654) −1.554 (0.502)** −0.002 (0.001)* 0.757 (0.172)*** −0.093 (0.045)* −0.226 (0.081)** −0.305 (0.557) 10.32*** 32.91% 134

Note: Revenue = quarterly revenue in billions; Cogs % = cost of goods sold/revenue; FC % = (depreciation expense + interest expense)/revenue; S&P GSCI index is deflated by100; leverage = long-term debt/total assets; Tobin’s Q = (market equity value plus net book liabilities)/book value; full-service = 1 for full-service restaurants and 0 for limited service restaurants. Values in the parentheses are standard errors. * 0.05 level. ** 0.01 level. *** 0.001 level.

significant risk exposure (non-exposed firms). Revenue and costs were higher in exposed firms than in non-exposed firms during both boom and slump periods. This may be an indication that exposed firms are larger in size than non-exposed firms are. The explanation for this is that risk exposure might not be large enough to have a material impact on small firms (Bartram, 2005). The percentage of cost of goods sold to revenue is higher in exposed firms, which may be less efficient in controlling variable cost percent than non-exposed firms. This could partially explain why non-exposed firms have higher market premium (Tobin’s Q) than exposed firms, as the market would reward the efficient firms. Since market price and expected returns are inversely correlated, a low Tobin’s Q means that investors of exposed firms are compensated with higher returns for bearing commodity price risk, an indication that commodity price risk is a part of the systematic risk. Financial leverage is higher in non-exposed firms than in exposed firms in the slump periods. This suggests that restaurateurs might be practicing risk balancing between business and financial risks commonly found among agricultural producers (Collins, 1985). The percentage of full-service restaurants is not significantly different between exposed firms and non-exposed firms. For exposed firms, cost structure and financial leverage are not significantly different between boom and slump periods (Panel C). However, for non-exposed firms, fixed cost is significantly higher in slump periods than in boom periods. This means non-exposed firms have to adjust the fixed cost if they wish to remain nonexposed through ups and downs of the commodity price cycle. This is an indication that the management of operating leverage might be essential to remain unexposed to commodity price risk when commodity price trend changes. The information in Table 2 is useful when it comes to discerning the differences between exposed firms and non-exposed ones. However, the information cannot be used directly to infer the effects of risk exposure determinants, as the extent and level of risk exposure might not be determined by the same set of factors. The effects of risk exposure determinants identified in the proposed model were examined with a sample of exposed firms. The results are presented in Table 3. 4.3. Determinants of the level of equity risk exposure Table 3 presents the results of the empirical testing of the proposed DCF model. The VIF values of all variables are below 2, lower than the customary acceptable level of 10. The empirical

127

testing supported the model prediction that operating leverage and financial leverage are effective tools in controlling exposure to commodity price risk in restaurants. The empirical testing further showed that the size of the effect is contingent on the cycle of commodity price. As a robustness test, the models were tested with franchise status as an additional control variable because franchisors receive franchise fees that are not directly exposed to commodity price risk and can pass the risk exposure to franchisees. The results of the new models are qualitative the same as the proposed models (same significant variables and directions). The reason could be that most of the observations are from franchise firms (40 out of 56 firms; 313 out of 381 observations). The models were also tested against the effect of seasonality. The results were qualitatively identical when quarterly dummy variables were included as control variables. Therefore, the discussion will be based on the results of the proposed model. Cost of goods sold percent was not significant, but fixed cost percent was negatively correlated to equity risk exposure in slump periods, suggesting that an increase in operating leverage can reduce equity risk exposure. Specifically, a 1% increase in fixed cost can lead to a 79.86% (e−1.554 −1) reduction in risk exposure. However, operating leverage only works in the slump periods. One possible explanation for this asymmetric effect lies in profitmaximizing managers’ pricing behavior. When commodity prices rise, restaurants would have strong incentives to increase price. This is a feasible action because previous studies (Kahneman et al., 1986a,b) have shown that customers are receptive to price increases justified by increases in costs. The corresponding movements in price and cost will keep risk exposure stable, resulting in a break of correlation. When commodity prices drop, restaurants tend not to reduce prices as quickly as the drop in food cost occurs (see Surowiecki, 2013, for an example), resulting in an increase in risk exposure. Financial leverage proves to be a more effective tool than operating leverage in managing the risk exposure. Financial leverage works in both boom and slump periods and its effect is stronger than that of operating leverage. A 1% increase in financial leverage will lead to an increase of 113.19% (e0.757 –1) in equity risk exposure in slump periods and 263.28% (e1.290 –1) in boom periods. Considering that the equity risk exposure is positive on average, an average restaurant would benefit from increasing (decreasing) financial leverage in boom (slump) periods. Table 3 shows that the increase of revenue can reduce risk exposure in exposed firms. However, Table 2 illustrates that revenue is higher in exposed firms than non-exposed ones. These two findings together may suggest that firms have to reach a certain size to be affected by risk exposure (Bartram, 2005), but the effect of revenue on risk exposure is negative. The sign of revenue is negative in both boom and slump periods, a distinct departure from the prediction of the proposed DCF model. There are two possible explanations for this inconsistency. First, revenue is not simply a proxy of business volume, but also a proxy of firm size. Since large firms have the economy of scale to bear the fixed cost and expertise of engaging in hedging activities (Bodnar et al., 1998), their risk exposure could be lower than that of small firms. Second, revenue incorporates the effects of price and quantity. The information of managerial pricing behavior and consumers’ price elasticity of demand is mingled in the revenue variable. This finding signals a shortcoming of applying pure economic models in industries where consumer behaviors play a big role in pricing and operating decisions. Although this inconsistency between the model prediction and empirical results is statistically significant, it is not expected to have substantial economic implications. It would take an increase of $1 billion in revenue for the risk exposure to decrease 15.21% (e−0.165 – 1) to 25.70% (e−0.297 – 1).

128

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129

Market premium (Tobin’s Q) is negatively correlated to equity risk exposure in slump periods. An increase of one percent of market value in relative to book value would lead to a decrease of 8.88% (e−0.093 –1) in equity risk exposure. The market premium cannot be used as a strategic lever, since it is not in the direct control of managers. However, this finding indicates that restaurants with higher market premium have smaller risk exposure. This may reflect that highly valued firms are not as motivated to pursue risky propositions, such as adjusting operation to chase falling commodity price. This finding fits nicely with Froot et al.’s (1993) theory that firms with higher growth opportunities (i.e., high Q) have stronger incentives to reduce their risk exposure. Full-service restaurants experience a lower level of risk exposure than limited-service restaurants in slump periods. However, it is not likely that a change of service style will affect the risk exposure. A more plausible interpretation may be that the target customers of a full-service restaurant have a relatively low level of price elasticity of demand, which allows full-service restaurants flexibility to respond to commodity price changes. Commodity price was negatively correlated with the risk exposure in slump periods. This was consistent with the model prediction, but the effect size was minimal in relative to other determinants. 5. Conclusion The study found that both the level and extent of exposure to commodity price risk in the restaurant industry fluctuate over time. This is consistent with the prediction of the proposed theoretical model (Eq. (4)). The rapid decline of risk exposure after 2006 also suggests that restaurants might have engaged in some form of operational or financial hedging after 2006. The trend of decreasing risk exposure coincided with the trend of rising commodity prices. A comparison between exposed firms and nonexposed firms revealed that (1) the risk exposure might not be large enough to have material impacts on small firms (Bartram, 2005), (2) investors of exposed firms are rewarded for bearing commodity price risk, and (3) restaurateurs might be practicing risk balancing between business and financial risks. Empirical testing of the proposed model confirmed that an increase in operating (financial) leverage can reduce (increase) equity risk exposure to commodity price risk, but the effects are asymmetric during commodity price boom and slump. Financial leverage was found to be more effective than operating leverage. The present study contributes to the literature by proposing and testing a theoretical model of commodity price risk exposure determinants in the restaurant industry. The results show that the effects of operating and financial leverage on risk exposure are asymmetric during commodity price booms and slumps. The study also showed that the exposure to commodity price risk is positive in the restaurant industry, contradicting the intuitive assumption that the risk exposure should be negative because high commodity price leads to low profits. This suggests that restaurant stocks and commodity prices are driven by the same macroeconomic forces (Lombardi, 2013). Froot et al. (1993) theorized that firms with higher growth opportunities have stronger incentives to reduce their risk exposure. Many studies have tested this theory in the context of financial risks; the present study showed that Froot et al.’s (1993) theory also works in the context of a business risk by demonstrating that exposed firms have lower Tobin’s Q than non-exposed ones. The findings have useful managerial implications. Considering that the risk exposure is positive, restaurants are encouraged to increase their risk exposure during commodity price booms and reduce their risk exposure during commodity price slumps. They can do so by increasing (decreasing) financial leverage during commodity price booms (slumps). But financial leverage might not

be adjustable at times due to its adverse effects on credit rating and financial distress costs. In such situations, operational hedging would be a feasible choice. During commodity price slumps, equity returns would benefit from an increase in operating leverage. The study found that the extent and level of risk exposure quickly decreased after 2006. It may be an indication that restaurants have engaged in operational hedging to manage the exposure. The analysis results are based on stock returns. The implications may not be generalizable to private companies. The direction and size of the determinants’ effect might be different between equity risk exposure and cash flow risk exposure. Managers are advised to use caution when applying the findings to manage cash flow risk exposures. Future studies also can benefit from measuring risk exposure by matching restaurants with the major food commodity used, instead of using an index. This approach has potential to provide firm specific information. Although the discounted cash flow model provides insights into the effects of determinants of commodity price risk exposure, the model may not reflect managerial flexibility or optionality (Tufano, 1998), such as pricing decisions. Future studies that incorporate managerial pricing decisions and consumers’ price elasticity of demand would prove valuable to industries where consumer behaviors play a significant role in pricing and operating decisions. This direction of research is consistent with the future direction of the restaurant industry, in the aim to transform products from simple commodities into unique, monopolistic offerings (Muller, 1999). The findings of the present study could serve as a stepping-stone for interdisciplinary studies that cross-pollinates between finance and marketing disciplines in the hospitality context. Acknowledgement The author is grateful to the support of Purdue Research Fund (#205957). References Aaker, D.A., Mascarenhas, B., 1984. The need for strategic flexibility. J. Bus. Strateg. 5 (2), 74–82. Adler, M., Dumas, B., 1984. Exposure to currency risk: definition and measurement. Financ. Manag. 13 (2), 41–50. Bailey, W., Chan, K.C., 1993. Macroeconomic influences and the variability of the commodity futures basis. J. Financ. 48 (2), 555–573, http://dx. doi.org/10.1111/j.1540-6261.1993.tb04727.x. Bartram, S.M., 2005. The impact of commodity price risk on firm value – an empirical analysis of corporate commodity price exposures. Multinatl. Financ. J. 9 (3/4), 161–187. Blake, M., Mahady, N., 1991. How mid-sized companies manage risk. J. Appl. Corp. Financ. 4 (1), 59–65, http://dx.doi.org/10.1111/j.1745-6622.1991.tb00572.x. Bodnar, G.M., Hayt, G.S., Marston, R.C., 1998. 1998 Wharton survey of financial risk management by US non-financial firms. Financ. Manag. 27 (4), 70–91. Bollerslev, T., 1986. Generalized autoregressive conditional heteroskedasticity. J. Econom. 31 (3), 307–327. Brennan, M.J., Schwartz, E.S., 1985. Evaluating natural resource investments. J. Bus., 135–157. Bry, G., Boschan, C., 1971. Cyclical Analysis of Time Series: Selected Procedures and Conputer Programs. Campello, M., 2006. Debt financing: does it boost or hurt firm performance in product markets? J. Financ. Econ. 82 (1), 135–172. Carter, D.A., Rogers, D.A., Simkins, B.J., 2006. Does hedging affect firm value? Evidence from the US airline industry. Financ. Manag. 35 (1), 53–86, http://dx.doi.org/10.1111/j.1755-053X.2006.tb00131.x. Cashin, P., McDermott, C.J., Scott, A., 2002. Booms and slumps in world commodity prices. J. Dev. Econ. 69 (1), 277–296, http://dx.doi.org/10.1016/ S0304-3878(02)00062-7. Chang, C., 2009. To hedge or not to hedge: revenue management and exchange rate risk. Cornell Hosp. Q. 50 (3), 301–313, http://dx.doi.org/10.1177/ 1938965509333168. Chowdhry, B., Howe, J.T.B., 1999. Corporate risk management for multinational corporations: financial and operational hedging policies. Eur. Financ. Rev. 2 (2), 229–246, http://dx.doi.org/10.1023/a:1009778703889. Chung, K.H., Pruitt, S.W., 1994. A simple approximation of Tobin’s q. Financ. Manag. 23 (3), 70–74. Collins, R.A., 1985. Expected utility, debt-equity structure, and risk balancing. Am. J. Agric. Econ. 67 (3), 627–629, http://dx.doi.org/10.2307/1241085.

C.-H. Tang / International Journal of Hospitality Management 45 (2015) 121–129 Dreibus, T.C., Josephs, L., Jargon, J., 2014. Food prices surge as drought exacts a high toll on crops. Wall Str. J., 2014, March 18. Fabozzi, F.J., Francis, J.C., 1978. Beta as a random coefficient. J. Financ. Quant. Anal. 13 (1), 101–116. 2008. Food Prices and Small Businesses: Hearings Before the Committee on Small Business U.S. House of Representatives, pp. 1–8. Froot, K.A., Scharfstein, D.S., Stein, J.C., 1993. Risk management: coordinating corporate investment and financing policies. J. Financ. 48 (5), 1629–1658. Gaiotti, E., Lippi, F., 2008. Pricing Behaviour and the Introduction of the Euro: Evidence from a Panel of Restaurants. Springer. Gorton, G., Rouwenhorst, G.K., 2005. Facts and Fantasies About Commodity Futures. Working Paper. Financial Institutions Center. University of Pennsylvania. Groenewald, N., Fraser, P., 2000. Forecasting beta: how well does the ‘five-year rule of thumb’ do? J. Bus. Financ. Acc. 27 (7–8), 953–982, http://dx.doi.org/ 10.1111/1468-5957.00341. Haushalter, D.G., 2000. Financing policy, basis risk, and corporate hedging: evidence from oil and gas producers. J. Financ. 55 (1), 107–152. Heckman, J.J., 1979. Sample selection bias as a specification error. Econometrica 47 (1), 153–161. Hernandez, M., Torero, M., 2010. Examining the Dynamic Relationship Between Spot and Future Prices of Agricultural Commodities. IFPRI Discussion Paper. Markets, Trade and Institutions Division, International Food Policy Research Institute. Hesford, J.W., Potter, G., 2010. Accounting research in the Cornell quarterly: a review with suggestions for future research. Cornell Hosp. Q. 51 (4), 502–512. Jagannathan, R., Wang, Z., 1996. The conditional CAPM and the cross-section of expected returns. J. Financ. 51 (1), 3–53, http://dx.doi.org/10.2307/2329301. Jannarone, J., 2011, February 8. Leaving profit on the restaurant table. Wall Str. J. Jargon, J., 2011, March 7. Coffee talk: Starbucks chief on prices, McDonald’s rivalry. Wall Str. J. Jargon, J., 2012, August 29. Costlier food puts restaurants in bind. Wall Str. J., B8. Jargon, J., Spector, M., 2011, November 1. Hamburgers, fries and a shakeout. Wall Str. J., B8. Kahneman, D., Knetsch, J.L., Thaler, R., 1986a. Fairness and the assumptions of economics. J. Bus., S285–S300. Kahneman, D., Knetsch, J.L., Thaler, R., 1986b. Fairness as a constraint on profit seeking: entitlements in the market. Am. Econ. Rev., 728–741. Lee, S.K., Jang, S., 2011. Foreign exchange exposure of US tourism-related firms. Tour. Manag. 32 (4), 934–948, http://dx.doi.org/10.1016/j.tourman.2010.08.008.

129

Lee, S.K., Jang, S., 2012. The real estate risk of hospitality firms: examining stock-return sensitivity to property values. IJHM 31 (3), 695–702, http://dx.doi.org/10.1016/j.ijhm.2011.09.005. Lombardi, M.J., 2013. On the Correlation Between Commodity and Equity Returns: Implications for Portfolio Allocation. Bank for International Settlements. Loudon, G.F., 2004. Financial risk exposures in the airline industry: evidence from Australia and New Zealand. Aust. J. Manag. (University of New South Wales) 29 (2), 295–316. Marn, M.V., Rosiello, R.L., 1992. Managing price, gaining profit. Harv. Bus. Rev. 70 (5), 84–94. Miller, K.D., 1992. A framework for integrated risk management in international business. J. Int. Bus. Stud. 23 (2), 311–331. Modigliani, F., Merton, H.M., 1963. Corporate income taxes and the cost of capital: a correction. Am. Econ. Rev. 53 (3), 433–443. Muller, C.C., 1999. The business of restaurants: 2001 and beyond. IJHM 18 (4), 401–413, http://dx.doi.org/10.1016/S0278-4319(99)00045-6. National Restaurant Association, 2014. Restaurant Industry 2014 Forecast and Trends. Petersen, M.A., Thiagarajan, S.R., 2000. Risk measurement and hedging: with and without derivatives. Financ. Manag. 29 (4), 5–30. Pindyck, R.S., 2001. The dynamics of commodity spot and futures markets: a primer. Energy J. 22 (3), 1–29. Ross, S.A., Westerfield, R.W., Jordan, B.D., 2013. Fundamentals of Corporate Finance, 10th ed. McGraw-Hill Irwin, New York. Sharpe, W.F., 1964. Capital asset prices: a theory of market equilibrium under conditions of risk. J. Financ. 19 (3), 425–442. Silvapulle, P., Moosa, I.A., 1999. The relationship between spot and futures prices: evidence from the crude oil market. J. Futur. Mark. 19 (2), 175–193. Singh, A., 2009. The interest rate exposure of lodging firms. IJHM 28 (1), 135–143. Surowiecki, J., 2013. The Surprising Complexity of Lobster Prices, Retrieved http:// www.newyorker.com/talk/financial/2013/08/26/130826ta talk surowiecki (accessed 20.12.13). Thorn, B., 2014. 4 Gactors Affecting Commodity Prices, Retrieved http://nrn.com/ commodities/4-factors-affecting-commodity-prices (accessed 14.04.14). Tufano, P., 1998. The determinants of stock price exposure: financial engineering and the gold mining industry. J. Financ. 53 (3), 1015–1052, http://dx. doi.org/10.1111/0022-1082.00042. Woolley, S., 2010, October 4–10. Coping With Commodity Shock. Bloomberg Businessweek.