Was the Missing 2013 WASDE Missed?

Was the Missing 2013 WASDE Missed? Abstract Government crop data have been shown to contribute to the efficient operation of agricultural commodity markets. In 2013, the USDA curtailed its crop report publication for the first time in decades due to an appropriations lapse, thereby offering the chance to study the impact on markets of missing government data. As expected, derivatives markets for corn and soybeans did not display characteristic short-run patterns in terms of uncertainty resolution and price changes that are normally observed around scheduled USDA release times. We are unable to detect evidence of a prolonged period of heightened uncertainty, realized volatility around the missing report, or abnormal pricing errors in the absence of government data. However, an unsurprisingly large 2013 corn and soybean crop could confound that attempt. Announcement effects, crop production, futures, grain stocks, implied volatility, options, shutdown, USDA, WASDE Many studies show that domestic futures and options traders react to the publication of important government reports about the situation and outlook for major commodities (see, e.g., Isengildina-Massa et al. 2008b; Adjemian 2012). Under the assumption that these markets are efficient, price changes at USDA crop report announcement time represent a realignment of trader expectations based on new information.1 Whether or not the government’s information and forecasts correspond exactly to the beliefs of the average trader, statistically significant changes in forward commodity prices upon report announcement indicate newsworthiness and informational value (Adjemian and Smith 2012). Ultimately, more accurate information about market conditions improves resource allocation decisions throughout the supply chain. To generate these reports, the USDA commits substantial resources to the collection and analysis of survey information from across the growing region. Some researchers have argued that, in the absence of government action, private firms would step in and provide commodity situation and outlook information to market participants (Just 1983; Salin et al. 1998). Leaving aside questions of the accuracy (which may be affected by scale economies) or sufficiency of the service that may be offered by private alternatives, or even the welfare impact of increased costs on smaller consumers resulting from such a transition, we explore the counterfactual: what is the empirical effect of no longer providing government reports in the absence of additional alternative private sources of information to take their place? Declining federal budgets put this question into sharp relief. In 2011, the USDA National Agricultural Statistics Service (NASS) canceled several longstanding commodity reports due to financial constraints, including those depicting the national soybean crush, fats and oils stocks, cotton stocks and processing, and a quarterly wheat millings survey (Micik 2015). Although each of these reports resumed publication by 2015, in many cases due to direct calls from market participants, a steadily growing federal debt leaves in question the future stability of government support for the collection and dissemination of agricultural market data. In 2013, after an appropriations lapse, the U.S. government ceased routine operations from October 1st through October 16th. Due to an inability to collect necessary data and convene government specialists to discuss their implications, the USDA canceled two reports that had been previously scheduled for simultaneous release on the morning of October 11th: the World Agricultural Supply and Demand Estimates (WASDE)—USDA’s premier situation and outlook report—and the supporting NASS Crop Production report (Abbott 2013).2 Previously, these reports had routinely provided traders, market participants, and observers with the government’s balance-sheet-view of supply and demand conditions for major U.S. agricultural commodities. The absence of these reports broke a streak of consecutive publications stretching back to inception dates of 1980 and 1866, respectively. USDA production forecasts from 2013 are shown in figure 1; the October values are notably missing. Figure 1 View largeDownload slide USDA production forecasts for 2013/14 in millions of bushels Figure 1 View largeDownload slide USDA production forecasts for 2013/14 in millions of bushels Because they contain domestic and international production, trade, and consumption estimates, forecasts of end-of-year stock and average farm-price levels, as well as descriptions of important market developments, WASDE announcements are closely watched by market observers and have been shown to generate shocks to expected commodity prices and volatility levels (Isengildina-Massa et al. 2008b; Adjemian 2012). The October 2013 shutdown provides a natural experiment for studying the effects of the absence of government information about major domestic crops. Using historical Chicago Mercantile Exchange (CME) futures and options data for corn and soybeans, we conduct an event study to test for market impacts (or the absence of impacts) on price levels and implied volatility around the time of the scheduled release of the missing October 2013 WASDE. In doing so, we search for evidence of pricing errors and uncertainty that might be attributed to lower information quality about the first harvest following the drought conditions and tight stocks of 2012/2013. Because our estimation strategy could be confounded by a general increase in uncertainty generated by the government shutdown, we include a control for the Chicago Board Options Exchange Volatility Index (VIX). It is important to distinguish between informational impacts of reports on (a) observed realized volatility (futures price movements following report releases) and on (b) implied volatility derived from options prices. Extant literature has shown that the revelation of information contained in scheduled reports leads to an increase in realized futures price volatility as market expectations are realigned in the immediate aftermath of report release, but that implied volatilities—which are a forward-looking measure of volatility and reflect the expected volatility over the remaining life of an options contract—will fall in the aftermath of a “newsworthy” report. This is because reports contain valuable information and because uncertainty is removed from the market as expectations are realigned (Ederington and Lee 1996; McNew an Espinosa 1994; McKenzie, Thomsen, and Phelan 2007). We show that typical corn and soybean futures price and implied volatility responses observed around report release dates did not occur during the government shutdown and absence of the October 2013 WASDE report. Failure to discover these systematic pricing patterns provides supporting evidence about the important role played by government agricultural reports in guiding markets to more efficient price and uncertainty equilibria. We further quantify the impact of the missing report by demonstrating that it contributed to heightened uncertainty in both the corn and soybean markets compared to the levels that likely would have been observed had historical information patterns held (at that time of the crop year). This indicates an increased cost of insuring against adverse price changes in the absence of government crop information. However, we are unable to demonstrate any conclusive evidence of pricing errors due to the canceled report, or any enhanced effect of the November WASDE—the first government report released after the shutdown—conditional on the size of the production surprise for either commodity. This result could be confounded by the fact that the 2013 crop was large, and the November WASDE that year was viewed as unsurprising (Good 2013). Background The USDA began issuing monthly crop reports in 1866 to inform producers and users of commodities about market conditions. Though the initial reports covered only cotton and tobacco, over time many commodities have been added or removed to these publications. Nevertheless, the reporting schedule remained consistent for 147 years.3 The collection of supply, demand, and price estimates published in the WASDE report is generated by a consensus process that draws together data and experts from several USDA agencies (Vogel and Bange 1999). Due to its sensitivity, each report is generated under high-security “lockup” conditions, and released at noon in the Eastern U.S. Time Zone (ET), during active trading hours.4 The WASDE reports can be organized according to a forecasting cycle: the May report contains the first pre-harvest estimates for the coming marketing year. The August report includes the first NASS production forecasts for corn and soybeans based on detailed, farm-level surveys; these continue through November. Final production estimates for these commodities are reported in January, alongside a quarterly NASS Grain Stocks report, which contains government data regarding on- and off-farm inventory levels. Other Grain Stocks reports appear in late March, June, and September. The first and second of these are published contemporaneously with NASS Planting Intentions and Acreage reports. The WASDE balance sheets reported in May through January are influenced by these NASS reports. From February through April, WASDE reports remain relatively unchanged, except for small refinements.5 Prior research has demonstrated significant market impacts on commodity futures prices of crop production reports (Gorham 1978; Miller 1979; Hoffman 1980; Sumner and Mueller 1989; Colling and Irwin 1990; Grunewald, McNulty, and Biere 1993; Garcia et al. 1997; Isengildina, Irwin, and Good 2006). McKenzie (2008) showed that corn traders would be willing to pay for advance access to the August crop production report. A growing body of literature focuses specifically on the impact of WASDE. Fortenberry and Sumner (1993) do not detect a price reaction to WASDE from the mid- to late-1980s, and wonder whether private forecasts had improved enough to anticipate government announcements In contrast, Isengildina-Massa et al. (2008a; 2008b) find that corn and soybean markets exhibit significant announcement effects around WASDE publication, and that the effect is particularly pronounced during those months when the report includes NASS crop production forecasts. Isengildina, Irwin, and Good (2006) also show that WASDE affects cattle and hog prices. Adjemian (2012) finds that that WASDE generates announcement effects in soybean, wheat, and that existing market conditions (e.g., low inventory levels) can amplify these effects. Kauffman (2013), Lehecka, Wang, and Garcia (2014), and Adjemian and Irwin (2016) all show that most of the information provided by WASDE is absorbed by each market in a matter of minutes. To proxy for market expectations, several researchers use private forecast data to condition the market reaction to the “surprise” component of the USDA forecast (see, e.g., Colling and Irwin 1990; Grunewald, McNulty, and Biere 1993; Garcia et al. 1997; Egelkraut et al. 2003). The 2013 harvest was a bumper crop for both corn and soybeans. Following the drought-plagued prior year, which saw the lowest corn production level since 2006, stocks were very low in 2013 and the U.S. 2013 crop reversed a three-year trend of declining production for both commodities. It was also a year of relatively unsurprising news about corn and soybean production. Table 1 compares the average surprise of the USDA crop production forecasts from 1995–2015 to the 2012 and 2013 individual crop years; surprises are calculated as the absolute percentage difference between the consensus private forecast—published a few days ahead of WASDE—and the official USDA numbers.6 The consensus private forecast comprises an average of pre-report production forecasts made by a variety of private firms and corresponding to each monthly report release date over the sample period. For both commodities, 2013 was an average year as far as USDA surprises go: the mean corn and soybean production figure that USDA published that year was 1.17% and 1.3% different than the private pre-release consensus, respectively—which is very close to the mean over all the years in the sample. Further, both are well below the prior year’s average corn surprise of 1.65% and soybean surprise of 2.02%. Similarly, the market surprise in reaction to the November 2013 USDA crop forecast—the first one that the department released post-shutdown—fits in with the historical surprises experienced in past Novembers, and are each less than half the size of the previous year’s November production surprises. Indeed, of all the NASS crop reports over the marketing year, those published in November carry the lowest average surprise for both commodities from 1995–2015. Table 1 USDA Market Surprises for Corn and Soybean Production in Absolute Terms, 1995–2015   All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)                All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)              Note: Standard deviations are shown in parentheses. Table 1 USDA Market Surprises for Corn and Soybean Production in Absolute Terms, 1995–2015   All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)                All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)              Note: Standard deviations are shown in parentheses. Both of these facts work against the likelihood of observing significant market impacts of a missing USDA report. If a harvest is relatively predictable, and a missing report occurs at a relatively calm point in the forecasting cycle, historically, the government’s role in aligning expectations is diminished—as is the researcher’s ability to measure the effect of a missing report. Data Historical government crop reports including production forecasts are maintained online by Cornell’s Mann Library, and can be accessed freely. We associate each report with the proper close-to-close change in implied volatility and absolute returns of the nearest-to-deliver futures contract not in the expiry month at the time the report was published (for a discussion of USDA's changing report publication schedule, see Adjemian 2012). We purchased daily corn futures and options data from January 1995 to October 2015 from Bloomberg data service, who calculate annualized implied volatilities for active futures contracts.7 We focus attention on nearest-to-deliver futures and options contracts as these are the most heavily traded, and as such likely incorporate market information and reflect market expectations more efficiently than deferred contracts. The Chicago Board Options Exchange maintains the VIX to capture daily market expectations about near-term volatility of the S&P 500 Index. Intraday futures and options prices were purchased from the CME and include both electronic and regular session trades; we calculated corresponding intraday implied volatilities according to Black (1976) using out-of-the money put and call options with strike prices that were no more than 10% greater or less than the current futures prices. To form a comprehensive time series of private forecast expectations, we draw data from several sources. For the 1970 to 2000 period, we use a simple average of the pre-report production forecasts made by Informa Economics (formerly Sparks Commodities, Inc.) and Conrad Leslie. From 2001 to 2005, we average the private expectations published by Informa Economics and the Dow Jones Newswire survey. The latter survey alone is used for crop years from 2006 to 2012. From 2013 to 2015, we use the private commodity production forecasts published in the Bloomberg survey. The WASDE and private crop production forecasts tend to be highly correlated. This fact is noted in extant literature (e.g., Garcia et al. 1997; Egelkraut et al. 2003; Good and Irwin 2006; McKenzie 2008). However, it is possible for private forecasts to reflect information not contained in WASDE reports and vice versa. With this in mind, McKenzie (2008) investigated the relative pricing impacts and influences on market agents’ price expectations of the two types of information. This author showed that although private production forecasts are at least as accurate as USDA crop production forecasts, the USDA numbers contribute to agents’ price expectations and influence futures price movements. Methods Traders use government commodity information to inform and adjust their expectations about market fundamentals. If private sources of such information are unable to approximate the government’s role—at least over the short-term—missing USDA information could lead to heightened uncertainty, higher volatility, and the absence of systematic price reactions that would otherwise occur at scheduled report release times. Then, once government information resumes, greater certainty and price shocks may be observed if the market’s fundamental expectations were less accurate, or less credible, in the absence of official forecasts. Commodity prices and implied volatilities are noisy, so we search for empirical evidence supporting these potential outcomes at different observational frequencies, that is, monthly, daily, and intraday. At the monthly frequency, we calculate the monthly average of our daily data, specifically, end-of-day implied volatility and VIX levels, and absolute close-to-close futures returns.8,9 Rather than splitting these data according to calendar month, we break them up according to the WASDE release schedule. For example, we identify as “July” the period of time beginning with the first day that could be affected by the July WASDE report to the trading day immediately preceding the release of the August report. To identify any potential effects of the missing report at a monthly frequency, for each commodity j and month m we difference the series to determine the percentage change in the average IV and returns from one month to the next. We then model the change in each as a function of independent variables. For example, the monthly IV model is stated as:   ΔIV-j,m=βcons+ β1ΔIV-j,m-1+β2ΔRet-j,m-1+β3ΔVIX-m+β4ΔVIX-m-1+βtime'[δtime]+ βm'[δm]+βSep13δSep13+βOct13δOct13+βNov13δNov13+ɛj,m. (1) In equation (1), the percentage change in average monthly implied IV is modeled as a function of the contemporaneous change in the average VIX (to capture any economy-wide changes in volatility), the lagged average IV, and the lagged value of the other dependent variable of interest (e.g., returns in the case of IV), to directly account for autocorrelation. We also include a vector of monthly dummies to capture seasonality, δm, and another to control for linear and quadratic time trends, δtime. Finally, besides a constant, we include three dummy variables for the months around the missing report, δSep13-δNov13, to identify any periods of abnormally heightened or dampened uncertainty or market returns.10 Our daily models are built in a similar way, but focus on the daily percentage change in implied volatility or absolute change in futures prices. For each commodity j and day t, we specify the percentage change in implied volatility as   ΔIV-j,t=γcons+ γ1ΔIV-j,t-1+γ2ΔRet-j,t-1+γ3ΔVIX-t+γ4ΔVIX-t-1+γtime'θtime+ γm'θm+γdow'θdow+γRpts'θRptsγOctWin'θOctWin+γOct2013Win'θOct2013Win+γNovWin'θNovWin+γNov2013Win'θNov2013Win+uj,t. (2) In addition to a regression constant, the specification in equation (2) includes lagged values of commodity j’s implied and realized market volatility, same day and lagged values of the VIX, and dummy vectors for calendar time (these enter as linear and quadratic values in θtime), seasonality ( θm), and day-of-week effects ( θdow). Indicator variables for important report publication dates are also included in the vector θRpts, specific to each WASDE and Stocks report over the calendar year (e.g., August WASDE, or Grain Stocks) outside of October and November; the coefficient on each of these indicators represents the average impact of these reports on IV levels or price volatility. We specify indicator variables, θOctWin and θNovWin, respectively, for the window around normal October and November reporting periods, that is, non-2013, October and November WASDE’s (using seven-day event windows that include three trading days before and after the publication of each report). It is important to include pre-report trading days, as the private forecasts are typically released several days prior to the USDA reports and market agents’ price expectations are influenced by information contained in these private reports. The impact of the missing report is studied by including similar windows for the 2013 versions of these reports in θOct2013Win and θNov2013Win. For the intraday analysis, we transform the futures price data into natural logarithms and construct minute-by-minute percentage returns for a two-hour event window around each of the scheduled release times of all October WASDE reports over the period 1995 to 2013. For the IVs, we first create an IV index by normalizing minute-by-minute IV’s with respect to the average IV observed across each event window for every year in the sample period, since systemic volatility varied considerably over the period of interest.11 To isolate the event window, we begin by calculating implied volatilities and returns 60 minutes before announcement, starting from the minute t=−60 through minute t=−1, and 60 minutes after announcement, from minute t = 1 through minute t = 60. Based on Kauffman (2013), Lehecka, Wang, and Garcia (2014), and Adjemian and Irwin (2016), we determine that a window of 60 minutes prior to the release of the report and 60 minutes after the release is sufficient to capture any intraday market absorption of news, as traders form positions and prices adjust to the new information contained in each report.12 Minute t = 0 represents the time of report release, minute t = 1 is the first trading minute after the new information in the report was released, while minute t=−1 is the last trading minute before the report was released. The data provide 18 distinct pre-2013 observations for each minute in the event window, representing the time series of corn and soybeans IV and futures prices on scheduled October report days. Given the small sample size, we use a Gaussian kernel estimator to generate smoothed multivariate empirical distributions that simulate the “normal” behavior of futures and IV during the event window around the scheduled publication time. Previous literature has established that futures returns are not well characterized by a normal distribution (e.g., Venkateswaran, Brorsen and Hall 1993), and we argue that our multivariate empirical distribution estimates better capture the historical price movements and correlation structure between the commodity returns. We narrow the window to the 21 trading minutes around the scheduled publication time to focus attention on the intraday period most affected by USDA announcements: our results are displayed on a per-minute basis for futures returns. Since options are far more lightly traded in the sample and IVs are calculated over differing levels of moneyness, we aggregate IV levels for the ten trading minutes preceding and following the moment of scheduled USDA WASDE announcement. For each unit of time in the analysis (21 for futures, and two for IV per each commodity), we perform 1,000 iterations of 18 futures returns and IV draws, each representing a pre-2013 year in the sample. By drawing from multivariate distributions we preserve the historical cross-commodity correlation structure at announcement time, as commodity prices changes are positively correlated in the immediate aftermath of an October WASDE.13 Then, we rank simulated futures returns (after converting them to absolute value) and normalized IVs by commodity from smallest to largest value. We thus generate simulated confidence bands at the 10% and 90% levels using the simulated cumulative distribution functions of absolute returns and normalized IVs, and plot the time path of the futures return and normalized IV that occurred around the scheduled release of the 2013 October WASDE. Results Using Monthly Averages Table 2 displays results from the portion of our analysis that relies on monthly data. A positive and significant coefficient on the Oct2013 indicator variable when IV is the dependent variable would indicate that the change in IV from the trading month following the September 2013 WASDE to the trading month that would have followed the October 2013 WASDE was less uncertainty-resolving than normal. In models where absolute return is the dependent variable, such a finding could instead indicate general under- or overreaction—in terms of realized volatility—compared to normal market conditions. Our results in table 2 do not provide any evidence for an abnormal change in the average level of annualized implied or realized volatility around the missing October 2013 crop report, compared to what would be normally expected in an ordinary October. No statistical significance is assigned to the coefficient of any 2013 indictor variable in any of the models in the table. We also observe that the VIX is positively correlated with IV for corn, implying that higher general economic uncertainty contributes to higher corn price uncertainty. Table 2 Impact of Missing October 2013 Report on the Change in Monthly Average Implied and Realized Volatility (absolute returns) in Each Commodity’s nearby Contract, 1995–2015   Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%    Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%  Note: Standard errors are shown in parentheses. Monthly dummies and linear and quadratic time trends were estimated, but their coefficients are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Table 2 Impact of Missing October 2013 Report on the Change in Monthly Average Implied and Realized Volatility (absolute returns) in Each Commodity’s nearby Contract, 1995–2015   Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%    Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%  Note: Standard errors are shown in parentheses. Monthly dummies and linear and quadratic time trends were estimated, but their coefficients are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Using Daily Data Table 3 results demonstrate that VIX plays a greater role in influencing daily (as opposed to monthly) changes in implied and realized volatility. Our IV and absolute returns measures using the nearest delivery contract in both commodity markets are positively correlated with the VIX, albeit at low magnitude. Our daily models support the empirical finding that the following USDA reports have significant announcement effects for one or both commodities, in terms of lowering uncertainty and updating price expectations: January WASDE/Grain Stocks; March Grain Stocks; June Grain Stocks; and the August–November WASDEs. Panel 3b of table 3 shows that normal October WASDEs are associated with statistically significant decreases in implied volatility on the day of release in both the corn and soybean market, and the day following release in the corn market.14 Out of 20 October WASDE report days from 1995 to 2015, 18 and 17 of those days saw lower close-of-day implied volatility than the day prior for corn and soybeans, respectively. In contrast, on the day scheduled for the October 2013 report, implied volatility ticked up in both markets, although not significantly. At the same time, realized volatility normally increases as this report updates traders’ expectations about expected fundamentals. In 2013, because the USDA did not publish an October WASDE, the market did not exhibit the normal patterns of uncertainty reduction or adjustment in price expectations associated with a WASDE. Furthermore, in addition to not being significant, the coefficients on the “Report” indicator variable in the October 2013 window bear the wrong sign for both commodities: being positive, they do not provide evidence that uncertainty about commodity conditions fell as would be expected had the curtailed report been released on that day as originally scheduled. Taken together, our daily results establish abnormal IV and absolute return behavior around the missing October report. That is, because the government shutdown prevented the writing and publication of a WASDE in October 2013, it is likely that market uncertainty—and therefore options prices—were higher than they would have been in the presence of fresh government crop information. Table 3 USDA Announcement Effects on the Change in Daily Implied and Realized Volatility (absolute returns), 1995–2015    Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)     Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)  Note: Standard errors are shown in parentheses. Dummy variables to capture day-of-week and seasonality effects, as well as linear and quadratic time trends were estimated, but are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Table 3 USDA Announcement Effects on the Change in Daily Implied and Realized Volatility (absolute returns), 1995–2015    Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)     Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)  Note: Standard errors are shown in parentheses. Dummy variables to capture day-of-week and seasonality effects, as well as linear and quadratic time trends were estimated, but are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. On the other hand, panel 3c of table 3 does not show evidence of any uncharacteristic snap-backs that could be associated with a stronger-than-normal reaction to USDA news. Indeed, outside of the reduction in corn market implied volatility, none of the other report-day announcement effects are judged to be significant—although it is important to note that they all bear the proper sign when compared to what is normally observed for the November WASDE report.15 To account for the possibility that the market reaction to the November 2013 report was stronger relative to the amount of information that USDA provided that month, figure 2 plots the change in corn and soybean futures prices against the difference in the government and private production forecasts. Predictably, the government’s production surprise is inversely correlated with the market price response, that is, a larger than expected government production forecast is generally followed by a decrease in commodity prices. In the figures, neither the November 2013 production surprise nor the price reaction to that surprise was noticeably larger when compared to previous crop production reports, or previous November reports specifically—echoing our table 3 non-finding of evidence about larger than normal market snap-back. Figure 2 View largeDownload slide Reaction to USDA news conditioned on the USDA production surprise, 1995–2015 Figure 2 View largeDownload slide Reaction to USDA news conditioned on the USDA production surprise, 1995–2015 As a natural experiment, the timing of the appropriations lapse limits the generalizability of this result: recall that 2013 was generally a quiet year in terms of changes to USDA harvest expectations. If the shutdown had occurred the prior year amidst a historic drought, or if USDA’s reporting capacity had been limited for a longer stretch of time, it is possible if not likely that the first post-shutdown report would carry a larger than average announcement effect. Using Intraday Data Figure 3 shows the results of our intraday simulation of the distribution of normalized implied volatility in the 20 minutes surrounding October WASDE announcements from 1995–2012. For both commodities, but particularly for soybeans, the implied volatility level usually decreases immediately following the USDA announcement, indicating a reduction in uncertainty about commodity price expectations. The relative magnitude of these changes between commodities match well with the results from our models that use daily data: the average October report seems to have a larger effect on the soybean harvest contract. On October 11th, 2013—the day the October WASDE was originally scheduled—the intraday corn IV level did not decrease at the scheduled announcement time (but stayed within the 90% confidence bands predicted by the simulation). For soybeans, the average intraday IV level that day actually increased substantially in the two ten-minute periods bracketing the release time, to a level well outside that predicted by the 90% confidence level of the simulated normalized distribution. Figure 3 View largeDownload slide Simulated average normalized intraday implied volatility around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Note: Implied volatility index values are shown on the vertical axis of both panels. At a value of 1, the normalized implied volatility is equal to the average IV level observed in the twenty trading minutes around the announcement time that of that year’s report. Figure 3 View largeDownload slide Simulated average normalized intraday implied volatility around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Note: Implied volatility index values are shown on the vertical axis of both panels. At a value of 1, the normalized implied volatility is equal to the average IV level observed in the twenty trading minutes around the announcement time that of that year’s report. Likewise, figure 4 presents our intraday simulation results for absolute futures returns. As found in prior research (e.g., Adjemian and Irwin 2016) we demonstrate that, ordinarily, the moment of scheduled publication for a USDA report is characterized by a spike in realized price volatility. On October 11th, 2013, however (the date of the missing WASDE), futures returns for corn and soybeans exhibited no such spike. That is, corn and soybean markets did not react to a report that did not exist, as expected. Figure 4 View largeDownload slide Simulated absolute futures returns around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Figure 4 View largeDownload slide Simulated absolute futures returns around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Conclusion By disseminating survey results about farm plantings, acreage levels, inventories, and production, the USDA can improve the decisions and plans of market participants. As a public good, the information is provided free of charge, but the process of collecting it is costly. An established body of literature validates the importance of USDA situation and outlook information to the efficient operation of agricultural commodity markets by searching for announcement effects, or anomalous changes in option-implied volatility or futures prices that coincide with the publication of a report. But traders cannot adjust their price or uncertainty expectations to information that they do not have. In 2013, a U.S. government shutdown that curtailed the release of the first USDA crop report since the nineteenth century offers the chance—for the first time—to observe the operation of commodity markets in the absence of government information. Using both daily and intraday data, we find that corn and soybean markets did not display characteristic patterns in terms of uncertainty resolution and price changes that are normally observed around scheduled USDA release times, meaning that options prices (and therefore the price of hedging) were higher than they likely would have been had a WASDE report come out. In this context we can say that the 2013 October WASDE was missed. However, we cannot establish that these effects persisted for a prolonged period (using monthly averages), or that the first report that USDA issued following the shutdown carried any enhanced effects. And so with this in mind, the 2013 October WASDE was not missed “too much”. Because 2013 was a relatively unsurprising harvest in terms of production news—a fact that was confirmed by the November WASDE (Good 2013)—efforts to detect pricing errors or heightened uncertainty in the absence of government data are likely obscured. A shutdown or curtailed report that occurs in a less predictable news environment, such as during a hurricane or a drought, or even one that occurs earlier in the crop year when less information is known about the crop, could further destabilize commodity markets by barring their access to official government harvest statistics at a time when they are most necessary. Likewise, a prolonged cancelation of USDA crop reports, unlike 2013 when only one WASDE was terminated, could more easily lead to higher expected and realized market volatility. 1Weak-form efficient markets react to new public information. 2Due to shared information and contemporaneous publication, we refer to the WASDE alone since it is the headline report. In addition to the October WASDE and Crop Production reports, a Cotton Ginnings report and two weekly crop progress reports that had been scheduled during the government shutdown were canceled, while the USDA postponed a Cattle on Feed and peanut report until after the resumption of normal operations (USDA 2013). The department suspended all of its market news reports during the shutdown (Joint Economic Committee of the U.S. Congress 2013). 3Report delays have been observed, however. For example, the WASDE report originally scheduled for publication on September 12th, 2001, was actually issued two days later, following the terrorist attacks. 4From 1994-2012, the report was published at 8:30am ET in advance of market opening; previously, the release time was 3:30pm ET, after the close of daytime trading sessions. 5The location of the shutdown in the crop forecasting cycle is notable. October WASDEs include crop production survey information that improve their appeal to corn and soybean market participants. Were the shutdown to have occurred in say, February or March, we may not have been able to detect a similarly significant effect since those reports are not normally market-moving. 6Surprises are measured for those corn and soybean reports based on detailed, farm-level surveys conducted by NASS and released in August-November, as well as the final USDA estimate published in January. 7The options data are quite sparse until 1995, so we begin our analysis in that year. 8We preserve the directional implied volatility changes since USDA reports are assumed to be uncertainty-resolving, leading to lower IV levels. 9Absolute returns are used since they represent realized market volatility. Directional returns are not very informative for our purposes, since USDA information could just as easily lead to higher or lower prices; regressing directional price changes on report publications would cancel out the magnitude of government information shocks. 10Because no report was published in October 2013, we use the day it was originally scheduled (October 11th, 2013) to divide the periods Sep2013 and Oct2013. 11An additional issue is that our intraday IV series reflects the IV recovered from the most recent OTM option with a strike within 10% of the underlying futures price. It is well-known that IV can vary systematically over moneyness, giving rise to “smile” patterns where IVs from options further from the money are higher than those nearer to the money or to “smirks” where OTM puts trade at a premium relative to similarly OTM calls, or vice versa. This introduces unwanted noise into our IV time series because variation is driven, in part, by the fact that options are being drawn from different points on the IV surface. The problem is compounded by the fact that trading in options was thin during some portions of our study period. For this reason, we average the IV series over blocks of the pre- and post-event windows to assure that IV changes or lack thereof can be detected accurately. 12Using more trading minutes could potentially lead to the problem of other information influencing market prices and decreasing our ability to measure the response of the market to the October report. 13Although wheat is not a focus of the current study, we include it in this portion of the analysis for completeness since it is among the most liquid of the agricultural derivatives markets. 14For corn, the null hypothesis of no announcement effect is rejected at the 6% level for both these findings. 15Once again, significance for this coefficient is established at the 6% level. References Abbott C. 2013. U.S. Cancels October Crop Report, First Miss in 147 years. Reuters. Available at: http://www.reuters.com/article/us-usa-fiscal-cropreports-idUSBRE99G11T20131017. Adjemian M.K. 2012. Quantifying the WASDE Announcement Effect. American Journal of Agricultural Economics  94 ( 1): 238– 56. http://dx.doi.org/10.1093/ajae/aar131 Google Scholar CrossRef Search ADS   Adjemian M.K., Irwin S.H. 2016. USDA Announcement Effects in Real Time. Washington DC: U.S. Department of Agriculture, Economic Research Service. Available at: http://farmdoc.illinois.edu/irwin/research/USDA_Announcement_Effects_in_Real_Time.pdf. Adjemian M.K., Smith A. 2012. Using USDA Forecasts to Estimate the Price Flexibility of Demand for Agricultural Commodities. American Journal of Agricultural Economics  94 ( 4): 978– 95. http://dx.doi.org/10.1093/ajae/aas032 Google Scholar CrossRef Search ADS   Black F. 1976. The Pricing of Commodity Contracts. Journal of Financial Economics  3 ( 1-2): 167– 79. Google Scholar CrossRef Search ADS   Colling P.L., Irwin S.H. 1990. The Reaction of Live Hog Futures Prices to USDA Hogs and Pigs Reports. American Journal of Agricultural Economics  72 ( 1): 84– 94. http://dx.doi.org/10.2307/1243147 Google Scholar CrossRef Search ADS   Ederington L.H., Lee J.H. 1996. The Creation and Resolution of Market Uncertainty: The Impact of Information Releases on Implied Volatility. Journal of Financial and Quantitative Analysis  31 ( 4): 513– 39. Google Scholar CrossRef Search ADS   Egelkraut T.M., Garcia P., Irwin S.H., Good D.L. 2003. An Evaluation of Crop Forecast Accuracy for Corn and Soybeans: USDA and Private Information Agencies. Journal of Agricultural and Applied Economics  35 ( 1): 79– 95. Google Scholar CrossRef Search ADS   Fortenberry T.R., Sumner D.A. 1993. The Effects of USDA Reports in Futures and Options Markets. The Journal of Futures Markets  13 ( 2): 157– 73. http://dx.doi.org/10.1002/fut.3990130204 Google Scholar CrossRef Search ADS   Garcia P., Irwin S., Leuthold R., Yang L. 1997. The Value of Public Information in Commodity Futures Markets. Journal of Economic Behavior and Organization  32 ( 4): 559– 70. http://dx.doi.org/10.1016/S0167-2681(97)00013-9 Google Scholar CrossRef Search ADS   Good D. 2013. Large Corn and Soybean Crops Confirmed. Farmdoc Daily. University of Illinois at Urbana-Champaign. Available at: http://farmdocdaily.illinois.edu/2013/11/large-corn-soybean-crops-confirmed.html. Good D.L., Irwin S.H. 2006. Understanding USDA Corn and Soybean Production Forecasts: An Overview of Methods, Performance and Market Impact over 1970–2005. AgMAS Project Research Report 2006–01, Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign. Gorham M. 1978. Public and Private Sector Information in Agricultural Commodity Markets. Federal Reserve Bank of San Francisco Economic Review , Spring, pp. 30– 8. Grunewald O., McNulty M.S., Biere A.W. 1993. Live Cattle Futures Response to Cattle on Feed Reports. American Journal of Agricultural Economics  75 ( 1): 131– 7. http://dx.doi.org/10.2307/1242961 Google Scholar CrossRef Search ADS   Hoffman G. 1980. The Effect of Quarterly Livestock Reports on Cattle and Hog Prices. North Central Journal of Agricultural Economics  2 ( 2): 145– 50. Google Scholar CrossRef Search ADS   Isengildina-Massa O., Irwin S.H., Good D.L., Gomez J.K. 2008a. The Impact of Situation and Outlook Information in Corn and Soybean Futures Markets: Evidence from WASDE Reports. Journal of Agricultural and Applied Economics  40 ( 1): 89– 103. Google Scholar CrossRef Search ADS   Isengildina-Massa O., Irwin S.H., Good D.L., Gomez J.K. 2008b. Impact of WASDE Reports on Implied Volatility in Corn and Soybean Markets. Agribusiness  24 ( 4): 473– 90. Google Scholar CrossRef Search ADS   Isengildina O., Irwin S.H., Good D.L. 2006. The Value of USDA Situation and Outlook Information in Hog and Cattle Markets. Journal of Agricultural and Resource Economics  31 ( 2): 262– 82. Joint Economic Committee of the U.S. Congress. 2013. The Economic Consequences of the Federal Government Shutdown. Report prepared by Vice Chair Klobuchar's staff. October. Available at: https://www.jec.senate.gov/public/_cache/files/e29f1102-765b-4676-9b92-952a5f75d099/the-economic-consequences-of-the-federal-government-shutdown.pdf. Just R.E. 1983. The Impact of Less Data on the Agricultural Economy and Society. American Journal of Agricultural Economics  65 ( 5): 872– 81. Google Scholar CrossRef Search ADS   Kauffman N. 2013. Have Extended Trading Hours Made Agricultural Commodity Markets Riskier? Economic Review of the Kansas City Federal Reserve Bank  ( QIII): 67– 94, Quarter 3: pp. 5–32. Lehecka G., Wang X., Garcia P. 2014. Gone in Ten Minutes: Intraday Evidence of Announcement Effects in the Electronic Corn Futures Market. Applied Economic Perspectives and Policy  36 ( 3): 504– 26. Google Scholar CrossRef Search ADS   McKenzie A.M. 2008. Pre-Harvest Price Expectations for Corn: The Information Content of USDA Reports and New Crop Futures. American Journal of Agricultural Economics  90 ( 2): 351– 66. http://dx.doi.org/10.1111/j.1467-8276.2007.01117.x Google Scholar CrossRef Search ADS   McKenzie A.M., Thomsen M.R., Phelan J.B. 2007. How Do You Straddle Hogs and Pigs? – Ask the Greeks! Applied Financial Economics  17 ( 7): 511– 20. Google Scholar CrossRef Search ADS   McNew K.P., Espinsosa J.A. 1994. The Informational Content of USDA Crop Reports: Impacts on Uncertainty and Expectations in Grain Futures Markets. The Journal of Futures Markets  14 ( 4): 475– 92. Google Scholar CrossRef Search ADS   Micik K. 2015. USDA Restarts Soybean Crush Reports. The Progressive Farmer. October 1. Available at: https://greatamericancrop.com/news-resources/article/2015/10/01/usda-restarts-soybean-crush-reports. Miller S.E. 1979. The Response of Futures Prices to New Market Information: The Case of Live Hogs. Southern Journal of Agricultural Economics  11 ( 1): 67– 70. Google Scholar CrossRef Search ADS   Salin V., Thurow A.P., Smith K.R., Elmer N. 1998. Exploring the Market for Agricultural Economics Information: Views of Private Sector Analysts. Review of Agricultural Economics  20 ( 1): 114– 24. Google Scholar CrossRef Search ADS   Sumner D.A., Mueller R.A.E. 1989. Are Harvest Forecasts News? USDA Announcements and Futures Market Reactions. American Journal of Agricultural Economics  71 ( 1): 1– 8. http://dx.doi.org/10.2307/1241769 Google Scholar CrossRef Search ADS   U.S. Department of Agriculture. 2013. USDA Announces Cancellation and Postponement of Selected Reports Impacted by the Lapse in Federal Funding. Department Press Release No. 0194.13, October 17th. Available at: https://www.usda.gov/media/press-releases/2013/10/17/usda-announces-cancellation-and-postponement-selected-reports. Venkateswaran Mi., Brorsen B. Wade, Hall J.A. 1993. The Distribution of Standardized Futures Price Changes. Journal of Futures Markets  13 ( 3): 279– 98. http://dx.doi.org/10.1002/fut.3990130305 Google Scholar CrossRef Search ADS   Vogel F.A., Bange G.A. 1999. Understanding USDA Crop Forecasts . Washington DC: U.S. Department of Agriculture, National Agricultural Statistics Service and Office of the Chief Economist, World Agricultural Outlook Board. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by US Government employees and is in the public domain in the US. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Economic Perspectives and Policy Oxford University Press

Loading next page...
 
/lp/ou_press/was-the-missing-2013-wasde-missed-Gpny4geD5a
Publisher
Oxford University Press
Copyright
Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by US Government employees and is in the public domain in the US.
ISSN
2040-5790
eISSN
2040-5804
D.O.I.
10.1093/aepp/ppx049
Publisher site
See Article on Publisher Site

Abstract

Abstract Government crop data have been shown to contribute to the efficient operation of agricultural commodity markets. In 2013, the USDA curtailed its crop report publication for the first time in decades due to an appropriations lapse, thereby offering the chance to study the impact on markets of missing government data. As expected, derivatives markets for corn and soybeans did not display characteristic short-run patterns in terms of uncertainty resolution and price changes that are normally observed around scheduled USDA release times. We are unable to detect evidence of a prolonged period of heightened uncertainty, realized volatility around the missing report, or abnormal pricing errors in the absence of government data. However, an unsurprisingly large 2013 corn and soybean crop could confound that attempt. Announcement effects, crop production, futures, grain stocks, implied volatility, options, shutdown, USDA, WASDE Many studies show that domestic futures and options traders react to the publication of important government reports about the situation and outlook for major commodities (see, e.g., Isengildina-Massa et al. 2008b; Adjemian 2012). Under the assumption that these markets are efficient, price changes at USDA crop report announcement time represent a realignment of trader expectations based on new information.1 Whether or not the government’s information and forecasts correspond exactly to the beliefs of the average trader, statistically significant changes in forward commodity prices upon report announcement indicate newsworthiness and informational value (Adjemian and Smith 2012). Ultimately, more accurate information about market conditions improves resource allocation decisions throughout the supply chain. To generate these reports, the USDA commits substantial resources to the collection and analysis of survey information from across the growing region. Some researchers have argued that, in the absence of government action, private firms would step in and provide commodity situation and outlook information to market participants (Just 1983; Salin et al. 1998). Leaving aside questions of the accuracy (which may be affected by scale economies) or sufficiency of the service that may be offered by private alternatives, or even the welfare impact of increased costs on smaller consumers resulting from such a transition, we explore the counterfactual: what is the empirical effect of no longer providing government reports in the absence of additional alternative private sources of information to take their place? Declining federal budgets put this question into sharp relief. In 2011, the USDA National Agricultural Statistics Service (NASS) canceled several longstanding commodity reports due to financial constraints, including those depicting the national soybean crush, fats and oils stocks, cotton stocks and processing, and a quarterly wheat millings survey (Micik 2015). Although each of these reports resumed publication by 2015, in many cases due to direct calls from market participants, a steadily growing federal debt leaves in question the future stability of government support for the collection and dissemination of agricultural market data. In 2013, after an appropriations lapse, the U.S. government ceased routine operations from October 1st through October 16th. Due to an inability to collect necessary data and convene government specialists to discuss their implications, the USDA canceled two reports that had been previously scheduled for simultaneous release on the morning of October 11th: the World Agricultural Supply and Demand Estimates (WASDE)—USDA’s premier situation and outlook report—and the supporting NASS Crop Production report (Abbott 2013).2 Previously, these reports had routinely provided traders, market participants, and observers with the government’s balance-sheet-view of supply and demand conditions for major U.S. agricultural commodities. The absence of these reports broke a streak of consecutive publications stretching back to inception dates of 1980 and 1866, respectively. USDA production forecasts from 2013 are shown in figure 1; the October values are notably missing. Figure 1 View largeDownload slide USDA production forecasts for 2013/14 in millions of bushels Figure 1 View largeDownload slide USDA production forecasts for 2013/14 in millions of bushels Because they contain domestic and international production, trade, and consumption estimates, forecasts of end-of-year stock and average farm-price levels, as well as descriptions of important market developments, WASDE announcements are closely watched by market observers and have been shown to generate shocks to expected commodity prices and volatility levels (Isengildina-Massa et al. 2008b; Adjemian 2012). The October 2013 shutdown provides a natural experiment for studying the effects of the absence of government information about major domestic crops. Using historical Chicago Mercantile Exchange (CME) futures and options data for corn and soybeans, we conduct an event study to test for market impacts (or the absence of impacts) on price levels and implied volatility around the time of the scheduled release of the missing October 2013 WASDE. In doing so, we search for evidence of pricing errors and uncertainty that might be attributed to lower information quality about the first harvest following the drought conditions and tight stocks of 2012/2013. Because our estimation strategy could be confounded by a general increase in uncertainty generated by the government shutdown, we include a control for the Chicago Board Options Exchange Volatility Index (VIX). It is important to distinguish between informational impacts of reports on (a) observed realized volatility (futures price movements following report releases) and on (b) implied volatility derived from options prices. Extant literature has shown that the revelation of information contained in scheduled reports leads to an increase in realized futures price volatility as market expectations are realigned in the immediate aftermath of report release, but that implied volatilities—which are a forward-looking measure of volatility and reflect the expected volatility over the remaining life of an options contract—will fall in the aftermath of a “newsworthy” report. This is because reports contain valuable information and because uncertainty is removed from the market as expectations are realigned (Ederington and Lee 1996; McNew an Espinosa 1994; McKenzie, Thomsen, and Phelan 2007). We show that typical corn and soybean futures price and implied volatility responses observed around report release dates did not occur during the government shutdown and absence of the October 2013 WASDE report. Failure to discover these systematic pricing patterns provides supporting evidence about the important role played by government agricultural reports in guiding markets to more efficient price and uncertainty equilibria. We further quantify the impact of the missing report by demonstrating that it contributed to heightened uncertainty in both the corn and soybean markets compared to the levels that likely would have been observed had historical information patterns held (at that time of the crop year). This indicates an increased cost of insuring against adverse price changes in the absence of government crop information. However, we are unable to demonstrate any conclusive evidence of pricing errors due to the canceled report, or any enhanced effect of the November WASDE—the first government report released after the shutdown—conditional on the size of the production surprise for either commodity. This result could be confounded by the fact that the 2013 crop was large, and the November WASDE that year was viewed as unsurprising (Good 2013). Background The USDA began issuing monthly crop reports in 1866 to inform producers and users of commodities about market conditions. Though the initial reports covered only cotton and tobacco, over time many commodities have been added or removed to these publications. Nevertheless, the reporting schedule remained consistent for 147 years.3 The collection of supply, demand, and price estimates published in the WASDE report is generated by a consensus process that draws together data and experts from several USDA agencies (Vogel and Bange 1999). Due to its sensitivity, each report is generated under high-security “lockup” conditions, and released at noon in the Eastern U.S. Time Zone (ET), during active trading hours.4 The WASDE reports can be organized according to a forecasting cycle: the May report contains the first pre-harvest estimates for the coming marketing year. The August report includes the first NASS production forecasts for corn and soybeans based on detailed, farm-level surveys; these continue through November. Final production estimates for these commodities are reported in January, alongside a quarterly NASS Grain Stocks report, which contains government data regarding on- and off-farm inventory levels. Other Grain Stocks reports appear in late March, June, and September. The first and second of these are published contemporaneously with NASS Planting Intentions and Acreage reports. The WASDE balance sheets reported in May through January are influenced by these NASS reports. From February through April, WASDE reports remain relatively unchanged, except for small refinements.5 Prior research has demonstrated significant market impacts on commodity futures prices of crop production reports (Gorham 1978; Miller 1979; Hoffman 1980; Sumner and Mueller 1989; Colling and Irwin 1990; Grunewald, McNulty, and Biere 1993; Garcia et al. 1997; Isengildina, Irwin, and Good 2006). McKenzie (2008) showed that corn traders would be willing to pay for advance access to the August crop production report. A growing body of literature focuses specifically on the impact of WASDE. Fortenberry and Sumner (1993) do not detect a price reaction to WASDE from the mid- to late-1980s, and wonder whether private forecasts had improved enough to anticipate government announcements In contrast, Isengildina-Massa et al. (2008a; 2008b) find that corn and soybean markets exhibit significant announcement effects around WASDE publication, and that the effect is particularly pronounced during those months when the report includes NASS crop production forecasts. Isengildina, Irwin, and Good (2006) also show that WASDE affects cattle and hog prices. Adjemian (2012) finds that that WASDE generates announcement effects in soybean, wheat, and that existing market conditions (e.g., low inventory levels) can amplify these effects. Kauffman (2013), Lehecka, Wang, and Garcia (2014), and Adjemian and Irwin (2016) all show that most of the information provided by WASDE is absorbed by each market in a matter of minutes. To proxy for market expectations, several researchers use private forecast data to condition the market reaction to the “surprise” component of the USDA forecast (see, e.g., Colling and Irwin 1990; Grunewald, McNulty, and Biere 1993; Garcia et al. 1997; Egelkraut et al. 2003). The 2013 harvest was a bumper crop for both corn and soybeans. Following the drought-plagued prior year, which saw the lowest corn production level since 2006, stocks were very low in 2013 and the U.S. 2013 crop reversed a three-year trend of declining production for both commodities. It was also a year of relatively unsurprising news about corn and soybean production. Table 1 compares the average surprise of the USDA crop production forecasts from 1995–2015 to the 2012 and 2013 individual crop years; surprises are calculated as the absolute percentage difference between the consensus private forecast—published a few days ahead of WASDE—and the official USDA numbers.6 The consensus private forecast comprises an average of pre-report production forecasts made by a variety of private firms and corresponding to each monthly report release date over the sample period. For both commodities, 2013 was an average year as far as USDA surprises go: the mean corn and soybean production figure that USDA published that year was 1.17% and 1.3% different than the private pre-release consensus, respectively—which is very close to the mean over all the years in the sample. Further, both are well below the prior year’s average corn surprise of 1.65% and soybean surprise of 2.02%. Similarly, the market surprise in reaction to the November 2013 USDA crop forecast—the first one that the department released post-shutdown—fits in with the historical surprises experienced in past Novembers, and are each less than half the size of the previous year’s November production surprises. Indeed, of all the NASS crop reports over the marketing year, those published in November carry the lowest average surprise for both commodities from 1995–2015. Table 1 USDA Market Surprises for Corn and Soybean Production in Absolute Terms, 1995–2015   All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)                All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)              Note: Standard deviations are shown in parentheses. Table 1 USDA Market Surprises for Corn and Soybean Production in Absolute Terms, 1995–2015   All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)                All Years Avg.   2013 Crop Year   2012 Crop Year     All Months  November  All Months  November  All Months  November  Corn  1.01%  0.50%  1.17%  0.29%  1.65%  0.90%    (0.86%)  (0.29%)              Soybeans  1.38%  0.92%  1.30%  1.34%  2.02%  2.77%    (1.17%)  (0.71%)              Note: Standard deviations are shown in parentheses. Both of these facts work against the likelihood of observing significant market impacts of a missing USDA report. If a harvest is relatively predictable, and a missing report occurs at a relatively calm point in the forecasting cycle, historically, the government’s role in aligning expectations is diminished—as is the researcher’s ability to measure the effect of a missing report. Data Historical government crop reports including production forecasts are maintained online by Cornell’s Mann Library, and can be accessed freely. We associate each report with the proper close-to-close change in implied volatility and absolute returns of the nearest-to-deliver futures contract not in the expiry month at the time the report was published (for a discussion of USDA's changing report publication schedule, see Adjemian 2012). We purchased daily corn futures and options data from January 1995 to October 2015 from Bloomberg data service, who calculate annualized implied volatilities for active futures contracts.7 We focus attention on nearest-to-deliver futures and options contracts as these are the most heavily traded, and as such likely incorporate market information and reflect market expectations more efficiently than deferred contracts. The Chicago Board Options Exchange maintains the VIX to capture daily market expectations about near-term volatility of the S&P 500 Index. Intraday futures and options prices were purchased from the CME and include both electronic and regular session trades; we calculated corresponding intraday implied volatilities according to Black (1976) using out-of-the money put and call options with strike prices that were no more than 10% greater or less than the current futures prices. To form a comprehensive time series of private forecast expectations, we draw data from several sources. For the 1970 to 2000 period, we use a simple average of the pre-report production forecasts made by Informa Economics (formerly Sparks Commodities, Inc.) and Conrad Leslie. From 2001 to 2005, we average the private expectations published by Informa Economics and the Dow Jones Newswire survey. The latter survey alone is used for crop years from 2006 to 2012. From 2013 to 2015, we use the private commodity production forecasts published in the Bloomberg survey. The WASDE and private crop production forecasts tend to be highly correlated. This fact is noted in extant literature (e.g., Garcia et al. 1997; Egelkraut et al. 2003; Good and Irwin 2006; McKenzie 2008). However, it is possible for private forecasts to reflect information not contained in WASDE reports and vice versa. With this in mind, McKenzie (2008) investigated the relative pricing impacts and influences on market agents’ price expectations of the two types of information. This author showed that although private production forecasts are at least as accurate as USDA crop production forecasts, the USDA numbers contribute to agents’ price expectations and influence futures price movements. Methods Traders use government commodity information to inform and adjust their expectations about market fundamentals. If private sources of such information are unable to approximate the government’s role—at least over the short-term—missing USDA information could lead to heightened uncertainty, higher volatility, and the absence of systematic price reactions that would otherwise occur at scheduled report release times. Then, once government information resumes, greater certainty and price shocks may be observed if the market’s fundamental expectations were less accurate, or less credible, in the absence of official forecasts. Commodity prices and implied volatilities are noisy, so we search for empirical evidence supporting these potential outcomes at different observational frequencies, that is, monthly, daily, and intraday. At the monthly frequency, we calculate the monthly average of our daily data, specifically, end-of-day implied volatility and VIX levels, and absolute close-to-close futures returns.8,9 Rather than splitting these data according to calendar month, we break them up according to the WASDE release schedule. For example, we identify as “July” the period of time beginning with the first day that could be affected by the July WASDE report to the trading day immediately preceding the release of the August report. To identify any potential effects of the missing report at a monthly frequency, for each commodity j and month m we difference the series to determine the percentage change in the average IV and returns from one month to the next. We then model the change in each as a function of independent variables. For example, the monthly IV model is stated as:   ΔIV-j,m=βcons+ β1ΔIV-j,m-1+β2ΔRet-j,m-1+β3ΔVIX-m+β4ΔVIX-m-1+βtime'[δtime]+ βm'[δm]+βSep13δSep13+βOct13δOct13+βNov13δNov13+ɛj,m. (1) In equation (1), the percentage change in average monthly implied IV is modeled as a function of the contemporaneous change in the average VIX (to capture any economy-wide changes in volatility), the lagged average IV, and the lagged value of the other dependent variable of interest (e.g., returns in the case of IV), to directly account for autocorrelation. We also include a vector of monthly dummies to capture seasonality, δm, and another to control for linear and quadratic time trends, δtime. Finally, besides a constant, we include three dummy variables for the months around the missing report, δSep13-δNov13, to identify any periods of abnormally heightened or dampened uncertainty or market returns.10 Our daily models are built in a similar way, but focus on the daily percentage change in implied volatility or absolute change in futures prices. For each commodity j and day t, we specify the percentage change in implied volatility as   ΔIV-j,t=γcons+ γ1ΔIV-j,t-1+γ2ΔRet-j,t-1+γ3ΔVIX-t+γ4ΔVIX-t-1+γtime'θtime+ γm'θm+γdow'θdow+γRpts'θRptsγOctWin'θOctWin+γOct2013Win'θOct2013Win+γNovWin'θNovWin+γNov2013Win'θNov2013Win+uj,t. (2) In addition to a regression constant, the specification in equation (2) includes lagged values of commodity j’s implied and realized market volatility, same day and lagged values of the VIX, and dummy vectors for calendar time (these enter as linear and quadratic values in θtime), seasonality ( θm), and day-of-week effects ( θdow). Indicator variables for important report publication dates are also included in the vector θRpts, specific to each WASDE and Stocks report over the calendar year (e.g., August WASDE, or Grain Stocks) outside of October and November; the coefficient on each of these indicators represents the average impact of these reports on IV levels or price volatility. We specify indicator variables, θOctWin and θNovWin, respectively, for the window around normal October and November reporting periods, that is, non-2013, October and November WASDE’s (using seven-day event windows that include three trading days before and after the publication of each report). It is important to include pre-report trading days, as the private forecasts are typically released several days prior to the USDA reports and market agents’ price expectations are influenced by information contained in these private reports. The impact of the missing report is studied by including similar windows for the 2013 versions of these reports in θOct2013Win and θNov2013Win. For the intraday analysis, we transform the futures price data into natural logarithms and construct minute-by-minute percentage returns for a two-hour event window around each of the scheduled release times of all October WASDE reports over the period 1995 to 2013. For the IVs, we first create an IV index by normalizing minute-by-minute IV’s with respect to the average IV observed across each event window for every year in the sample period, since systemic volatility varied considerably over the period of interest.11 To isolate the event window, we begin by calculating implied volatilities and returns 60 minutes before announcement, starting from the minute t=−60 through minute t=−1, and 60 minutes after announcement, from minute t = 1 through minute t = 60. Based on Kauffman (2013), Lehecka, Wang, and Garcia (2014), and Adjemian and Irwin (2016), we determine that a window of 60 minutes prior to the release of the report and 60 minutes after the release is sufficient to capture any intraday market absorption of news, as traders form positions and prices adjust to the new information contained in each report.12 Minute t = 0 represents the time of report release, minute t = 1 is the first trading minute after the new information in the report was released, while minute t=−1 is the last trading minute before the report was released. The data provide 18 distinct pre-2013 observations for each minute in the event window, representing the time series of corn and soybeans IV and futures prices on scheduled October report days. Given the small sample size, we use a Gaussian kernel estimator to generate smoothed multivariate empirical distributions that simulate the “normal” behavior of futures and IV during the event window around the scheduled publication time. Previous literature has established that futures returns are not well characterized by a normal distribution (e.g., Venkateswaran, Brorsen and Hall 1993), and we argue that our multivariate empirical distribution estimates better capture the historical price movements and correlation structure between the commodity returns. We narrow the window to the 21 trading minutes around the scheduled publication time to focus attention on the intraday period most affected by USDA announcements: our results are displayed on a per-minute basis for futures returns. Since options are far more lightly traded in the sample and IVs are calculated over differing levels of moneyness, we aggregate IV levels for the ten trading minutes preceding and following the moment of scheduled USDA WASDE announcement. For each unit of time in the analysis (21 for futures, and two for IV per each commodity), we perform 1,000 iterations of 18 futures returns and IV draws, each representing a pre-2013 year in the sample. By drawing from multivariate distributions we preserve the historical cross-commodity correlation structure at announcement time, as commodity prices changes are positively correlated in the immediate aftermath of an October WASDE.13 Then, we rank simulated futures returns (after converting them to absolute value) and normalized IVs by commodity from smallest to largest value. We thus generate simulated confidence bands at the 10% and 90% levels using the simulated cumulative distribution functions of absolute returns and normalized IVs, and plot the time path of the futures return and normalized IV that occurred around the scheduled release of the 2013 October WASDE. Results Using Monthly Averages Table 2 displays results from the portion of our analysis that relies on monthly data. A positive and significant coefficient on the Oct2013 indicator variable when IV is the dependent variable would indicate that the change in IV from the trading month following the September 2013 WASDE to the trading month that would have followed the October 2013 WASDE was less uncertainty-resolving than normal. In models where absolute return is the dependent variable, such a finding could instead indicate general under- or overreaction—in terms of realized volatility—compared to normal market conditions. Our results in table 2 do not provide any evidence for an abnormal change in the average level of annualized implied or realized volatility around the missing October 2013 crop report, compared to what would be normally expected in an ordinary October. No statistical significance is assigned to the coefficient of any 2013 indictor variable in any of the models in the table. We also observe that the VIX is positively correlated with IV for corn, implying that higher general economic uncertainty contributes to higher corn price uncertainty. Table 2 Impact of Missing October 2013 Report on the Change in Monthly Average Implied and Realized Volatility (absolute returns) in Each Commodity’s nearby Contract, 1995–2015   Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%    Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%  Note: Standard errors are shown in parentheses. Monthly dummies and linear and quadratic time trends were estimated, but their coefficients are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Table 2 Impact of Missing October 2013 Report on the Change in Monthly Average Implied and Realized Volatility (absolute returns) in Each Commodity’s nearby Contract, 1995–2015   Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%    Corn   Soybeans     (1)  (2)  (3)  (4)    Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Lag IV  −0.094  −0.10  −0.230***  −0.13     (0.071)  (0.24)  (0.069)  (0.23)  Lag Returns  0.044**  −0.27***  0.078***  −0.22***     (0.021)  (0.07)  (0.021)  (0.07)  VIX  0.083**  0.13  0.046  0.11     (0.041)  (0.14)  (0.039)  (0.13)  Lag VIX  0.037  0.02  0.020  0.01     (0.041)  (0.14)  (0.039)  (0.13)  Sep. 2013  −16.1  −27.2  −7.2  −26.2     (10.8)  (36.6)  (10.0)  (33.4)  Oct. 2013  5.8  −47.4  −9.1  −25.6     (10.6)  (35.8)  (10.0)  (33.3)  Nov. 2013  10.5  44.3  −4.4  −12.7     (10.6)  (35.7)  (10.0)  (33.4)  Constant  9.1**  −27.8**  5.2  −6.6     (3.7)  (12.4)  (3.4)  (11.3)  Observations  248  248  248  248  R-squared  33%  28%  23%  23%  Note: Standard errors are shown in parentheses. Monthly dummies and linear and quadratic time trends were estimated, but their coefficients are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Using Daily Data Table 3 results demonstrate that VIX plays a greater role in influencing daily (as opposed to monthly) changes in implied and realized volatility. Our IV and absolute returns measures using the nearest delivery contract in both commodity markets are positively correlated with the VIX, albeit at low magnitude. Our daily models support the empirical finding that the following USDA reports have significant announcement effects for one or both commodities, in terms of lowering uncertainty and updating price expectations: January WASDE/Grain Stocks; March Grain Stocks; June Grain Stocks; and the August–November WASDEs. Panel 3b of table 3 shows that normal October WASDEs are associated with statistically significant decreases in implied volatility on the day of release in both the corn and soybean market, and the day following release in the corn market.14 Out of 20 October WASDE report days from 1995 to 2015, 18 and 17 of those days saw lower close-of-day implied volatility than the day prior for corn and soybeans, respectively. In contrast, on the day scheduled for the October 2013 report, implied volatility ticked up in both markets, although not significantly. At the same time, realized volatility normally increases as this report updates traders’ expectations about expected fundamentals. In 2013, because the USDA did not publish an October WASDE, the market did not exhibit the normal patterns of uncertainty reduction or adjustment in price expectations associated with a WASDE. Furthermore, in addition to not being significant, the coefficients on the “Report” indicator variable in the October 2013 window bear the wrong sign for both commodities: being positive, they do not provide evidence that uncertainty about commodity conditions fell as would be expected had the curtailed report been released on that day as originally scheduled. Taken together, our daily results establish abnormal IV and absolute return behavior around the missing October report. That is, because the government shutdown prevented the writing and publication of a WASDE in October 2013, it is likely that market uncertainty—and therefore options prices—were higher than they would have been in the presence of fresh government crop information. Table 3 USDA Announcement Effects on the Change in Daily Implied and Realized Volatility (absolute returns), 1995–2015    Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)     Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)  Note: Standard errors are shown in parentheses. Dummy variables to capture day-of-week and seasonality effects, as well as linear and quadratic time trends were estimated, but are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. Table 3 USDA Announcement Effects on the Change in Daily Implied and Realized Volatility (absolute returns), 1995–2015    Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)     Corn   Soybeans      (1)  (2)  (3)  (4)     Implied Volatility  Absolute Returns  Implied Volatility  Absolute Returns  Panel 3a General Model  Lag IV  −0.22***  −0.0004  −0.29***  0.0005     (0.014)  (0.0010)  (0.014)  (0.0010)  Lag Returns  0.52***  0.096***  0.48**  0.084***     (0.150)  (0.014)  (0.190)  (0.014)  VIX  0.10***  0.008***  0.081***  0.009***     (0.027)  (0.002)  (0.030)  (0.002)  Lag VIX  0.007  0.002  −0.008  −0.000     (0.026)  (0.002)  (0.030)  (0.002)  Monthly USDA Reports  Jan WASDE  −5.75**  2.69***  −2.22  1.78***     (2.60)  (0.23)  (2.91)  (0.21)  Feb WASDE  −6.62**  0.17  −4.65  0.16     (2.67)  (0.23)  (2.99)  (0.21)  Mar WASDE  −1.20  0.064  −1.02  −0.12     (2.73)  (0.23)  (2.98)  (0.21)  Mar Stocks  −12.3***  1.82***  −13.3***  1.20***     (2.61)  (0.23)  (2.91)  (0.21)  Apr WASDE  −5.99**  −0.12  −5.18*  0.28     (2.61)  (0.23)  (2.91)  (0.21)  May WASDE  −0.12  0.36  −0.90  0.32     (2.60)  (0.23)  (2.91)  (0.21)  Jun WASDE  −7.93***  0.37  −6.91**  −0.34     (2.67)  (0.23)  (2.92)  (0.21)  Jun Stocks  −7.24***  2.44***  −8.73***  1.16***     (2.68)  (0.24)  (2.99)  (0.22)  Jul WASDE  −0.77  −0.14  −0.93  −0.10     (2.61)  (0.23)  (3.00)  (0.21)  Aug WASDE  −7.98***  1.51***  −6.76**  1.01***     (2.60)  (0.23)  (2.91)  (0.21)  Sep WASDE  −5.92**  1.22***  −5.32*  0.55**     (2.61)  (0.23)  (2.92)  (0.21)  Sep Stocks  −3.13  0.96***  −1.59  0.64***     (2.61)  (0.24)  (2.92)  (0.21)  Oct WASDE  See panel 3b  –  –  –                 Nov WASDE  See panel 3c  –  –  –                 Dec WASDE  −1.12  −0.094  −5.79*  0.16     (2.73)  (0.24)  (3.06)  (0.22)  Constant  −2.06  0.67***  −5.22*  0.29     (2.57)  (0.23)  (2.95)  (0.21)  Observations  4,856  5130  4,787  5,087  R-squared  10%  17%  12%  10%  Panel 3b October Window  Normal October Window  Report - 3  2.51  0.68***  1.25  −0.17     (2.69)  (0.24)  (3.02)  (0.22)  Report - 2  1.52  0.020  2.11  0.21     (2.69)  (0.24)  (3.02)  (0.22)  Report - 1  1.74  0.19  4.43  0.29     (2.69)  (0.24)  (3.01)  (0.22)  Report  −5.22*  1.55***  −16.1***  1.79***     (2.69)  (0.24)  (3.02)  (0.22)  Report + 1  −5.08*  0.49**  −3.64  0.35     (2.76)  (0.25)  (3.12)  (0.23)  Report + 2  0.22  0.11  −0.47  0.18     (2.75)  (0.25)  (3.09)  (0.23)  Report + 3  −0.64  −0.091  0.42  −0.22     (2.75)  (0.25)  (3.09)  (0.23)  October 2013 Window             Report - 3  −0.23  0.46  −2.84  −0.32     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  0.67  −0.74  −4.80  −0.75     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  2.28  0.28  −9.35  −0.71     (11.6)  (1.05)  (13.0)  (0.96)  Report  1.79  0.11  2.92  0.82     (11.6)  (1.05)  (13.0)  (0.96)  Report + 1  0.26  −0.43  −3.59  −0.67     (11.6)  (1.05)  (13.0)  (0.96)  Report + 2  2.56  0.22  −7.29  −0.59     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −1.18  −0.84  −6.03  0.07     (11.7)  (1.05)  (13.0)  (0.96)  Panel 3c November Window  Normal November Window  Report - 3  −0.01  0.089  −0.27  0.52**     (2.78)  (0.25)  (3.18)  (0.23)  Report - 2  1.01  −0.090  −0.42  −0.14     (2.78)  (0.25)  (3.10)  (0.23)  Report - 1  5.97**  −0.25  −0.11  0.00     (2.78)  (0.25)  (3.10)  (0.23)  Report  −19.9***  0.48*  −5.04  0.62***     (2.78)  (0.25)  (3.10)  (0.23)  Report + 1  −5.99**  0.01  −0.42  0.36     (2.79)  (0.25)  (3.10)  (0.23)  Report + 2  −2.45  −0.38  0.48  −0.072     (2.85)  (0.25)  (3.18)  (0.23)  Report + 3  11.5***  0.44*  1.25  0.12     (2.85)  (0.25)  (3.17)  (0.23)  November 2013 Window  Report - 3  1.97  −0.70  0.51  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 2  −1.65  −0.035  −1.58  −0.38     (11.6)  (1.05)  (13.0)  (0.96)  Report - 1  11.4  −0.93  0.73  −0.00     (11.6)  (1.05)  (13.0)  (0.96)  Report  −22.0*  0.57  −6.29  1.46     (11.7)  (1.05)  (13.0)  (0.96)  Report + 1  −16.5  0.63  −5.15  −0.72     (11.7)  (1.05)  (13.0)  (0.96)  Report + 2  −1.44  −0.59  −3.47  0.15     (11.6)  (1.05)  (13.0)  (0.96)  Report + 3  −5.14  −0.41  −3.02  −0.79     (11.6)  (1.05)  (13.0)  (0.96)  Note: Standard errors are shown in parentheses. Dummy variables to capture day-of-week and seasonality effects, as well as linear and quadratic time trends were estimated, but are not displayed to conserve space. Statistical significance is denoted by asterisks: *** = p < 0.01, ** = p < 0.05, and * = p < 0.1. On the other hand, panel 3c of table 3 does not show evidence of any uncharacteristic snap-backs that could be associated with a stronger-than-normal reaction to USDA news. Indeed, outside of the reduction in corn market implied volatility, none of the other report-day announcement effects are judged to be significant—although it is important to note that they all bear the proper sign when compared to what is normally observed for the November WASDE report.15 To account for the possibility that the market reaction to the November 2013 report was stronger relative to the amount of information that USDA provided that month, figure 2 plots the change in corn and soybean futures prices against the difference in the government and private production forecasts. Predictably, the government’s production surprise is inversely correlated with the market price response, that is, a larger than expected government production forecast is generally followed by a decrease in commodity prices. In the figures, neither the November 2013 production surprise nor the price reaction to that surprise was noticeably larger when compared to previous crop production reports, or previous November reports specifically—echoing our table 3 non-finding of evidence about larger than normal market snap-back. Figure 2 View largeDownload slide Reaction to USDA news conditioned on the USDA production surprise, 1995–2015 Figure 2 View largeDownload slide Reaction to USDA news conditioned on the USDA production surprise, 1995–2015 As a natural experiment, the timing of the appropriations lapse limits the generalizability of this result: recall that 2013 was generally a quiet year in terms of changes to USDA harvest expectations. If the shutdown had occurred the prior year amidst a historic drought, or if USDA’s reporting capacity had been limited for a longer stretch of time, it is possible if not likely that the first post-shutdown report would carry a larger than average announcement effect. Using Intraday Data Figure 3 shows the results of our intraday simulation of the distribution of normalized implied volatility in the 20 minutes surrounding October WASDE announcements from 1995–2012. For both commodities, but particularly for soybeans, the implied volatility level usually decreases immediately following the USDA announcement, indicating a reduction in uncertainty about commodity price expectations. The relative magnitude of these changes between commodities match well with the results from our models that use daily data: the average October report seems to have a larger effect on the soybean harvest contract. On October 11th, 2013—the day the October WASDE was originally scheduled—the intraday corn IV level did not decrease at the scheduled announcement time (but stayed within the 90% confidence bands predicted by the simulation). For soybeans, the average intraday IV level that day actually increased substantially in the two ten-minute periods bracketing the release time, to a level well outside that predicted by the 90% confidence level of the simulated normalized distribution. Figure 3 View largeDownload slide Simulated average normalized intraday implied volatility around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Note: Implied volatility index values are shown on the vertical axis of both panels. At a value of 1, the normalized implied volatility is equal to the average IV level observed in the twenty trading minutes around the announcement time that of that year’s report. Figure 3 View largeDownload slide Simulated average normalized intraday implied volatility around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Note: Implied volatility index values are shown on the vertical axis of both panels. At a value of 1, the normalized implied volatility is equal to the average IV level observed in the twenty trading minutes around the announcement time that of that year’s report. Likewise, figure 4 presents our intraday simulation results for absolute futures returns. As found in prior research (e.g., Adjemian and Irwin 2016) we demonstrate that, ordinarily, the moment of scheduled publication for a USDA report is characterized by a spike in realized price volatility. On October 11th, 2013, however (the date of the missing WASDE), futures returns for corn and soybeans exhibited no such spike. That is, corn and soybean markets did not react to a report that did not exist, as expected. Figure 4 View largeDownload slide Simulated absolute futures returns around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Figure 4 View largeDownload slide Simulated absolute futures returns around the announcement time of October WASDE reports, 1995–2012, compared to levels observed on the scheduled 2013 release day Conclusion By disseminating survey results about farm plantings, acreage levels, inventories, and production, the USDA can improve the decisions and plans of market participants. As a public good, the information is provided free of charge, but the process of collecting it is costly. An established body of literature validates the importance of USDA situation and outlook information to the efficient operation of agricultural commodity markets by searching for announcement effects, or anomalous changes in option-implied volatility or futures prices that coincide with the publication of a report. But traders cannot adjust their price or uncertainty expectations to information that they do not have. In 2013, a U.S. government shutdown that curtailed the release of the first USDA crop report since the nineteenth century offers the chance—for the first time—to observe the operation of commodity markets in the absence of government information. Using both daily and intraday data, we find that corn and soybean markets did not display characteristic patterns in terms of uncertainty resolution and price changes that are normally observed around scheduled USDA release times, meaning that options prices (and therefore the price of hedging) were higher than they likely would have been had a WASDE report come out. In this context we can say that the 2013 October WASDE was missed. However, we cannot establish that these effects persisted for a prolonged period (using monthly averages), or that the first report that USDA issued following the shutdown carried any enhanced effects. And so with this in mind, the 2013 October WASDE was not missed “too much”. Because 2013 was a relatively unsurprising harvest in terms of production news—a fact that was confirmed by the November WASDE (Good 2013)—efforts to detect pricing errors or heightened uncertainty in the absence of government data are likely obscured. A shutdown or curtailed report that occurs in a less predictable news environment, such as during a hurricane or a drought, or even one that occurs earlier in the crop year when less information is known about the crop, could further destabilize commodity markets by barring their access to official government harvest statistics at a time when they are most necessary. Likewise, a prolonged cancelation of USDA crop reports, unlike 2013 when only one WASDE was terminated, could more easily lead to higher expected and realized market volatility. 1Weak-form efficient markets react to new public information. 2Due to shared information and contemporaneous publication, we refer to the WASDE alone since it is the headline report. In addition to the October WASDE and Crop Production reports, a Cotton Ginnings report and two weekly crop progress reports that had been scheduled during the government shutdown were canceled, while the USDA postponed a Cattle on Feed and peanut report until after the resumption of normal operations (USDA 2013). The department suspended all of its market news reports during the shutdown (Joint Economic Committee of the U.S. Congress 2013). 3Report delays have been observed, however. For example, the WASDE report originally scheduled for publication on September 12th, 2001, was actually issued two days later, following the terrorist attacks. 4From 1994-2012, the report was published at 8:30am ET in advance of market opening; previously, the release time was 3:30pm ET, after the close of daytime trading sessions. 5The location of the shutdown in the crop forecasting cycle is notable. October WASDEs include crop production survey information that improve their appeal to corn and soybean market participants. Were the shutdown to have occurred in say, February or March, we may not have been able to detect a similarly significant effect since those reports are not normally market-moving. 6Surprises are measured for those corn and soybean reports based on detailed, farm-level surveys conducted by NASS and released in August-November, as well as the final USDA estimate published in January. 7The options data are quite sparse until 1995, so we begin our analysis in that year. 8We preserve the directional implied volatility changes since USDA reports are assumed to be uncertainty-resolving, leading to lower IV levels. 9Absolute returns are used since they represent realized market volatility. Directional returns are not very informative for our purposes, since USDA information could just as easily lead to higher or lower prices; regressing directional price changes on report publications would cancel out the magnitude of government information shocks. 10Because no report was published in October 2013, we use the day it was originally scheduled (October 11th, 2013) to divide the periods Sep2013 and Oct2013. 11An additional issue is that our intraday IV series reflects the IV recovered from the most recent OTM option with a strike within 10% of the underlying futures price. It is well-known that IV can vary systematically over moneyness, giving rise to “smile” patterns where IVs from options further from the money are higher than those nearer to the money or to “smirks” where OTM puts trade at a premium relative to similarly OTM calls, or vice versa. This introduces unwanted noise into our IV time series because variation is driven, in part, by the fact that options are being drawn from different points on the IV surface. The problem is compounded by the fact that trading in options was thin during some portions of our study period. For this reason, we average the IV series over blocks of the pre- and post-event windows to assure that IV changes or lack thereof can be detected accurately. 12Using more trading minutes could potentially lead to the problem of other information influencing market prices and decreasing our ability to measure the response of the market to the October report. 13Although wheat is not a focus of the current study, we include it in this portion of the analysis for completeness since it is among the most liquid of the agricultural derivatives markets. 14For corn, the null hypothesis of no announcement effect is rejected at the 6% level for both these findings. 15Once again, significance for this coefficient is established at the 6% level. References Abbott C. 2013. U.S. Cancels October Crop Report, First Miss in 147 years. Reuters. Available at: http://www.reuters.com/article/us-usa-fiscal-cropreports-idUSBRE99G11T20131017. Adjemian M.K. 2012. Quantifying the WASDE Announcement Effect. American Journal of Agricultural Economics  94 ( 1): 238– 56. http://dx.doi.org/10.1093/ajae/aar131 Google Scholar CrossRef Search ADS   Adjemian M.K., Irwin S.H. 2016. USDA Announcement Effects in Real Time. Washington DC: U.S. Department of Agriculture, Economic Research Service. Available at: http://farmdoc.illinois.edu/irwin/research/USDA_Announcement_Effects_in_Real_Time.pdf. Adjemian M.K., Smith A. 2012. Using USDA Forecasts to Estimate the Price Flexibility of Demand for Agricultural Commodities. American Journal of Agricultural Economics  94 ( 4): 978– 95. http://dx.doi.org/10.1093/ajae/aas032 Google Scholar CrossRef Search ADS   Black F. 1976. The Pricing of Commodity Contracts. Journal of Financial Economics  3 ( 1-2): 167– 79. Google Scholar CrossRef Search ADS   Colling P.L., Irwin S.H. 1990. The Reaction of Live Hog Futures Prices to USDA Hogs and Pigs Reports. American Journal of Agricultural Economics  72 ( 1): 84– 94. http://dx.doi.org/10.2307/1243147 Google Scholar CrossRef Search ADS   Ederington L.H., Lee J.H. 1996. The Creation and Resolution of Market Uncertainty: The Impact of Information Releases on Implied Volatility. Journal of Financial and Quantitative Analysis  31 ( 4): 513– 39. Google Scholar CrossRef Search ADS   Egelkraut T.M., Garcia P., Irwin S.H., Good D.L. 2003. An Evaluation of Crop Forecast Accuracy for Corn and Soybeans: USDA and Private Information Agencies. Journal of Agricultural and Applied Economics  35 ( 1): 79– 95. Google Scholar CrossRef Search ADS   Fortenberry T.R., Sumner D.A. 1993. The Effects of USDA Reports in Futures and Options Markets. The Journal of Futures Markets  13 ( 2): 157– 73. http://dx.doi.org/10.1002/fut.3990130204 Google Scholar CrossRef Search ADS   Garcia P., Irwin S., Leuthold R., Yang L. 1997. The Value of Public Information in Commodity Futures Markets. Journal of Economic Behavior and Organization  32 ( 4): 559– 70. http://dx.doi.org/10.1016/S0167-2681(97)00013-9 Google Scholar CrossRef Search ADS   Good D. 2013. Large Corn and Soybean Crops Confirmed. Farmdoc Daily. University of Illinois at Urbana-Champaign. Available at: http://farmdocdaily.illinois.edu/2013/11/large-corn-soybean-crops-confirmed.html. Good D.L., Irwin S.H. 2006. Understanding USDA Corn and Soybean Production Forecasts: An Overview of Methods, Performance and Market Impact over 1970–2005. AgMAS Project Research Report 2006–01, Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign. Gorham M. 1978. Public and Private Sector Information in Agricultural Commodity Markets. Federal Reserve Bank of San Francisco Economic Review , Spring, pp. 30– 8. Grunewald O., McNulty M.S., Biere A.W. 1993. Live Cattle Futures Response to Cattle on Feed Reports. American Journal of Agricultural Economics  75 ( 1): 131– 7. http://dx.doi.org/10.2307/1242961 Google Scholar CrossRef Search ADS   Hoffman G. 1980. The Effect of Quarterly Livestock Reports on Cattle and Hog Prices. North Central Journal of Agricultural Economics  2 ( 2): 145– 50. Google Scholar CrossRef Search ADS   Isengildina-Massa O., Irwin S.H., Good D.L., Gomez J.K. 2008a. The Impact of Situation and Outlook Information in Corn and Soybean Futures Markets: Evidence from WASDE Reports. Journal of Agricultural and Applied Economics  40 ( 1): 89– 103. Google Scholar CrossRef Search ADS   Isengildina-Massa O., Irwin S.H., Good D.L., Gomez J.K. 2008b. Impact of WASDE Reports on Implied Volatility in Corn and Soybean Markets. Agribusiness  24 ( 4): 473– 90. Google Scholar CrossRef Search ADS   Isengildina O., Irwin S.H., Good D.L. 2006. The Value of USDA Situation and Outlook Information in Hog and Cattle Markets. Journal of Agricultural and Resource Economics  31 ( 2): 262– 82. Joint Economic Committee of the U.S. Congress. 2013. The Economic Consequences of the Federal Government Shutdown. Report prepared by Vice Chair Klobuchar's staff. October. Available at: https://www.jec.senate.gov/public/_cache/files/e29f1102-765b-4676-9b92-952a5f75d099/the-economic-consequences-of-the-federal-government-shutdown.pdf. Just R.E. 1983. The Impact of Less Data on the Agricultural Economy and Society. American Journal of Agricultural Economics  65 ( 5): 872– 81. Google Scholar CrossRef Search ADS   Kauffman N. 2013. Have Extended Trading Hours Made Agricultural Commodity Markets Riskier? Economic Review of the Kansas City Federal Reserve Bank  ( QIII): 67– 94, Quarter 3: pp. 5–32. Lehecka G., Wang X., Garcia P. 2014. Gone in Ten Minutes: Intraday Evidence of Announcement Effects in the Electronic Corn Futures Market. Applied Economic Perspectives and Policy  36 ( 3): 504– 26. Google Scholar CrossRef Search ADS   McKenzie A.M. 2008. Pre-Harvest Price Expectations for Corn: The Information Content of USDA Reports and New Crop Futures. American Journal of Agricultural Economics  90 ( 2): 351– 66. http://dx.doi.org/10.1111/j.1467-8276.2007.01117.x Google Scholar CrossRef Search ADS   McKenzie A.M., Thomsen M.R., Phelan J.B. 2007. How Do You Straddle Hogs and Pigs? – Ask the Greeks! Applied Financial Economics  17 ( 7): 511– 20. Google Scholar CrossRef Search ADS   McNew K.P., Espinsosa J.A. 1994. The Informational Content of USDA Crop Reports: Impacts on Uncertainty and Expectations in Grain Futures Markets. The Journal of Futures Markets  14 ( 4): 475– 92. Google Scholar CrossRef Search ADS   Micik K. 2015. USDA Restarts Soybean Crush Reports. The Progressive Farmer. October 1. Available at: https://greatamericancrop.com/news-resources/article/2015/10/01/usda-restarts-soybean-crush-reports. Miller S.E. 1979. The Response of Futures Prices to New Market Information: The Case of Live Hogs. Southern Journal of Agricultural Economics  11 ( 1): 67– 70. Google Scholar CrossRef Search ADS   Salin V., Thurow A.P., Smith K.R., Elmer N. 1998. Exploring the Market for Agricultural Economics Information: Views of Private Sector Analysts. Review of Agricultural Economics  20 ( 1): 114– 24. Google Scholar CrossRef Search ADS   Sumner D.A., Mueller R.A.E. 1989. Are Harvest Forecasts News? USDA Announcements and Futures Market Reactions. American Journal of Agricultural Economics  71 ( 1): 1– 8. http://dx.doi.org/10.2307/1241769 Google Scholar CrossRef Search ADS   U.S. Department of Agriculture. 2013. USDA Announces Cancellation and Postponement of Selected Reports Impacted by the Lapse in Federal Funding. Department Press Release No. 0194.13, October 17th. Available at: https://www.usda.gov/media/press-releases/2013/10/17/usda-announces-cancellation-and-postponement-selected-reports. Venkateswaran Mi., Brorsen B. Wade, Hall J.A. 1993. The Distribution of Standardized Futures Price Changes. Journal of Futures Markets  13 ( 3): 279– 98. http://dx.doi.org/10.1002/fut.3990130305 Google Scholar CrossRef Search ADS   Vogel F.A., Bange G.A. 1999. Understanding USDA Crop Forecasts . Washington DC: U.S. Department of Agriculture, National Agricultural Statistics Service and Office of the Chief Economist, World Agricultural Outlook Board. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association 2017. This work is written by US Government employees and is in the public domain in the US.

Journal

Applied Economic Perspectives and PolicyOxford University Press

Published: Nov 9, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off