Labor Cost and Sugarcane Labor Cost and Sugarcane Mechanization in Florida: Mechanization in Florida: NPV and Real Options Approach NPV and Real Options Approach Nobuyuki Iwai Nobuyuki Iwai Robert D. Emerson Robert D. Emerson International Agricultural Trade and Policy Center International Agricultural Trade and Policy Center Department of Food and Resource Economics Department of Food and Resource Economics University of Florida University of Florida
Background Background � Foreign workers in US agriculture Foreign workers in US agriculture � � 77% of farm workers in 2002-04 77% of farm workers in 2002-04 � � At least 50% of farm workers are At least 50% of farm workers are � undocumented in 2002-04 (16% 1989-91, 16% 1989-91, undocumented in 2002-04 ( 38% in 1993-95) 38% in 1993-95) � Wage gap Wage gap � Average predicted wages for authorized, Average predicted wages for authorized, permanent resident, and citizen worker are 10%, 10%, permanent resident, and citizen worker are 18% and 14% higher than for unauthorized higher than for unauthorized 18% and 14% workers (Iwai et al. 2006). workers (Iwai et al. 2006).
Immigration Policy Reform Reform Immigration Policy � Increased border and domestic Increased border and domestic � enforcement, amnesty and guest worker enforcement, amnesty and guest worker programs programs Goal: legal labor force legal labor force Goal: � Concern that the reform might lead to Concern that the reform might lead to � labor cost increase in US ag ag. . labor cost increase in US � Need to study the impact of labor cost � Need to study the impact of labor cost increase: Mechanization Mechanization , Termination , Termination increase: (Emerson 2007) (Emerson 2007)
Previous Studies on Labor Cost Studies on Labor Cost Previous and Sugarcane Mechanization and Sugarcane Mechanization � Zepp Zepp and Clayton (1975) found cost advantage and Clayton (1975) found cost advantage � of mechanical harvesting operation over hand- of mechanical harvesting operation over hand- cut as early as 72-3 season. cut as early as 72-3 season. � Reduced revenue due to large field losses with Reduced revenue due to large field losses with � mechanical harvesting, resulting in $40.70 lower mechanical harvesting, resulting in $40.70 lower net returns per acre in comparison to hand-cut. net returns per acre in comparison to hand-cut. � If projected 74-5 machinery operating rates had If projected 74-5 machinery operating rates had � been used, the net returns per acre would have been used, the net returns per acre would have been about equal (Zepp Zepp 1975). 1975). been about equal (
Problems Problems � Previous studies calculated and compared Previous studies calculated and compared � the cost and returns from two technologies the cost and returns from two technologies for a single individual season. for a single individual season. � Dynamic decision-making analysis tools Dynamic decision-making analysis tools � often used in corporate finance: often used in corporate finance: Net present value (NPV) approach, Net present value (NPV) approach, Real options approach (ROA). Real options approach (ROA).
Objective of Our Study Objective of Our Study � Using NPV approach and ROA, we analyze the Using NPV approach and ROA, we analyze the � decision of the model sugarcane farmer (640 decision of the model sugarcane farmer (640 acres in total and 408 acres harvested) in Florida acres in total and 408 acres harvested) in Florida as to mechanization of harvesting for 72-3 as to mechanization of harvesting for 72-3 season. season. � Simulation: We also compute the adoption Simulation: We also compute the adoption � thresholds for the labor cost that should have thresholds for the labor cost that should have triggered investment in mechanical harvesting for triggered investment in mechanical harvesting for sugarcane in Florida. sugarcane in Florida. � Implication for mechanical harvesting for citrus. Implication for mechanical harvesting for citrus. �
NPV Approach NPV Approach � Compares discounted cash flow (DCF) less Compares discounted cash flow (DCF) less � the investment cost of two operations. the investment cost of two operations. � Important to forecast future free cash flow Important to forecast future free cash flow � (FCF) given information when the decision (FCF) given information when the decision is made. is made. � Also need to use appropriate discount rate Also need to use appropriate discount rate � (Weighted Average Cost of Capital). (Weighted Average Cost of Capital).
Estimated FCF Estimated FCF Estimated FCF from growing and harvesting sugarcane for the model farm ($ per 404 acres harvested) Hand cut Mechanical harvesting harvesting Season 72-3 72-3 Revenue 213,004.56 197,945.28 Labor cost 59,015.73 35,784.21 Other costs 95,995.25 104,256.41 Operating CF 57,993.59 57,904.66 Depreciation 10,855.60 13,569.20 EBIT 47,137.99 44,335.46 Tax on EBIT (29%) 13,670.02 12,857.28 CAPEX 10,855.60 13,569.20 FCF 33,467.97 31,478.18 Calculated from Zepp and Clayton (1975) and Walker (1972).
Forecasting Beyond 72-3 Season Forecasting Beyond 72-3 Season � Previous studies reported the revenue and Previous studies reported the revenue and � cost estimate only for 72-3 season. cost estimate only for 72-3 season. � Need to forecast for longer period for NPV Need to forecast for longer period for NPV � and ROA. and ROA. � A common approach is the Monte Carlo A common approach is the Monte Carlo � simulation in which all stochastic factors are simulation in which all stochastic factors are generated for future periods using the generated for future periods using the estimated parameters and distributions of estimated parameters and distributions of these series (Kobayashi 2003, Copeland these series (Kobayashi 2003, Copeland and Antikarov Antikarov 2003). 2003). and
Estimation of Stochastic Process Estimation of Stochastic Process � We estimate stochastic process for yield, price, We estimate stochastic process for yield, price, � labor cost, and other costs. labor cost, and other costs. st order � We make stationary series by taking 1 We make stationary series by taking 1 st order � difference of log of each series and subtracting difference of log of each series and subtracting st order difference. the mean of the 1 st order difference. the mean of the 1 � Test of independence between yield and price Test of independence between yield and price � and between costs was rejected. and between costs was rejected.
Vector Autoregressive Model Vector Autoregressive Model Since independence hypothesis was rejected, we estimate the VAR model for each combination. p is chosen that minimizes the bias-corrected version of the Akaike Information Criteria referred to as the AICC (Brockwell and Davis 2002): where L is likelihood function for bivariate normal distribution, n is number of obs, and m=2 is number of variables.
VAR Results VAR Results For yield and price X X 0 . 00 0 . 66 0 . 077 � � � � � � � � � � t 1 , 1 t 1 � = + � � � � � � � � X X 0 . 00 0 . 67 0 . 17 � � � � � � � � t 1 , 2 t 2 � X U 0 . 45 0 . 077 � � � � � � � t 2 , 1 t 1 � + + � � � � � � X U 0 . 60 0 . 40 � � � � � � � t 2 , 2 t 2 � U 0 0 . 0054 0 . 0021 � � � � � � � � � t 1 where ~ WN , and AICC -74.55 � � = � � � � � � � � U 0 0 . 0021 0 . 042 � � � � � � � � � t 2 For labor and other cost X 0 . 00 U � � � � � � t 1 t 1 = + � � � � � � X 0 . 00 U � � � � � � t 2 t 2 U 0 0 . 020 0 . 00052 � � � � � � � � � t 1 where ~ WN , and AICC -38.65 � � = � � � � � � � � U 0 0 . 00052 0 . 027 � � � � � � � � � t 2
Monte Carlo Simulation Monte Carlo Simulation Using the estimation results, we can generate future series as ˆ ˆ ˆ ˆ X ì Ö X Ö X U = + + + t 1 t 1 2 t 2 t � � where U is a white noise from estimated distributi on. t Since X t is the change in the de-trended (mean-subtracted) log of the original series, we can generate a sample of the original series by adding the trend, taking exponential and multiply it to the previous value. Repeating this nine times for each vector yields one sample of future nine-year pass for each variable. We generate 100,000 sets of the future paths of stochastic factors with which we calculate 100,000 sets of nine-year future FCF paths for each operation mode. The average of generated FCF is given next.
Recommend
More recommend
