FARMLAND:ISIT CURRENTLY PRICEDASAN ATTRACTIVE …€¦ · farmland portfolio of Westchester, a large farmland realtor and investment company with properties throughout the United

FARMLAND:IS IT CURRENTLY PRICED AS AN

ATTRACTIVE INVESTMENT?

TIMOTHY G. BAKER, MICHAEL D. BOEHLJE, AND MICHAEL R. LANGEMEIER

Farmland prices have risen dramatically in recent years, which has attracted interest from thebroader investment community. At the same time, concern is being expressed regarding anotherbubble in farmland prices. This paper studies and compares the farmland price to cash rent ratio(P/rent) with the price to earnings (P/E) ratio of stocks. We find that the farmland P/rent ratiohas reached historical highs and is currently at the level of the P/E ratio of the S&P 500 duringthe tech bubble. Data from 1911 to 2012 are used to estimate the beta of farmland, a measureof the risk that farmland adds to a diversified portfolio. The beta is found to be very low overthis period. Farmland returns are also regressed against expected and unexpected inflation, andwe find that farmland moves relatively close to one-to-one with inflation. We also report 10- and20-year holding period returns for farmland and find that the relationship between return and thecyclically adjusted P/rent ratio is strongly negative. Moreover, the current cyclically adjusted P/rentratio is extremely high, indicating a reason for caution when investors are considering farmlandpurchases.

Key words: Beta, cyclically adjusted P/E ratio, farmland prices, inflation hedge, S&P 500 returns.

JEL codes: Q14.

Farmland comprises the vast majority offarmers’ asset bases and personal wealth;USDA balance sheet data indicate that in2012 the value of farmland accounted for82% of total assets in production agricul-ture (USDA 2014). This percentage hasbeen increasing during the past decade inlarge part because of the dramatic increasein farmland prices. Farmland prices in theUnited States have increased 37% duringthe last 5 years, and in real terms, priceshave risen 7–10% annually in Corn Beltstates like Iowa and Illinois (Gloy 2012).The recent dramatic increase in farm realestate prices has attracted interest from thebroader investment community in farmland

Timothy G. Baker, Michael D. Boehlje, and Michael R. Lange-meier are all professors in the Department of AgriculturalEconomics at Purdue University. This project was supported bythe USDA National Institute of Food and Agriculture, Hatchproject number IND010590, and by the Center for CommercialAgriculture, Department of Agricultural Economics, PurdueUniversity. The authors acknowledge the helpful commentsof AJAE editor JunJie Wu, anonymous AJAE reviewers, andthe contribution of Sarah A. Stutzman, who performed theregressions contained in the paper.

This article was subjected to an expedited peer-review pro-cess that encourages contributions that frame emerging andpriority issues for the profession, as well as methodologicaland theoretical contributions.

as a component of their investment portfolio,as illustrated by financial services companyTIAA-CREF’s recent acquisition of thefarmland portfolio of Westchester, a largefarmland realtor and investment companywith properties throughout the United States.Similar investment interest is reflected bynumerous articles on farmland investingfound on banking and financial websites(Gustke 2013; Robinson 2013; Forbes 2013;Lube 2013).

Concern is being expressed by manyinvestment analysts that farmland priceswill become higher than justified by the fun-damentals, and will result in what we willlater recognize as a bubble (Bowman 2013;Watts 2013). One justification for this concernis that previous research has established thetendency of the farmland market to over-shoot (Burt 1986; Featherstone and Baker,1987; Featherstone and Baker, 1988). Thus,from the standpoint of the literature and ofhistory, another bubble in farmland priceswould not be a surprise.

Numerous previous studies of farmlandprices and values have been completed assummarized by Moss and Katchova (2005).This article builds on and extends earlier

Amer. J. Agr. Econ. 96(5): 1321–1333; doi: 10.1093/ajae/aau037Published online July 2, 2014

© The Author (2014). Published by Oxford University Press on behalf of the Agricultural and Applied EconomicsAssociation. All rights reserved. For permissions, please e-mail: [email protected]

at :: on October 30, 2014

http://ajae.oxfordjournals.org/D

ownloaded from

http://ajae.oxfordjournals.org/

1322 October 2014 Amer. J. Agr. Econ.

work by positioning the farmland investmentdecision as an investment portfolio choice,with a focus on the financial attractive-ness of farmland as an investment. We willnot attempt to assess the numerous non-economic arguments often made by farmers(and others) to justify the purchase of aparticular parcel of farmland. Thus, ourdiscussion will emphasize the risk, return,portfolio, and inflation hedge characteristicsof farmland compared to other commonfinancial investments that one might make.It is also important to note that we will notfocus on the operational details of man-aging and maintaining farmland, both ofwhich present significant challenges (andopportunities) that require specialized farmmanagement expertise.

To frame farmland as an investment choiceanalysis, the commonly accepted income cap-italization model of asset valuation is used.The constant growth present value modelprovides the theory behind this analysis pro-cedure. In this model the return to an assetat the current time (R0) is expected to growat rate g indefinitely, and the required rateof return is r (also constant into perpetu-ity), leading to the following present valueequation:

(1) Value =∞∑

n=1

R0(1 + g)n(1 + r)−n.

The solution to the infinite series of equation(1) yields the constant growth model (2):

(2) Value = R1

[1

r − g

].

From equation (2) we can derive the com-mon investment analysis metrics of cap rate(R1/Value) and the inverse of the cap rate,which is the value (or price) to earnings(P/E) ratio. In equation (2), the capitalizationrate is the difference between the requiredrate of return (r) and the anticipated con-stant growth rate (g). The P/E multiple willincrease as the difference between r − gdeclines, which means that reducing therequired rate of return, or increasing theexpected long-term growth in earnings, arethe two factors that will increase the multiple.

The P/E ratios for stocks are comparedto the price to earnings multiple (P/rent)for farmland in this paper. The P/E ratio is

computed by dividing market value per sharefor a particular stock or group of stocks bythe appropriate earnings per share (EPS).Historical or expected earnings per share canbe used in the computation. The reportedP/E ratios typically use historical earningsper share to compute the ratio. The averagemarket P/E ratio for stocks is 15 to 20; how-ever, it is important to note that the averageP/E ratio does vary across industries. In addi-tion, we create a cyclically adjusted P/E ratiofor farmland and compare this to RobertShiller’s data for the S&P 500. These resultsare augmented by econometric estimatesof the beta (β), a fundamental relative riskmetric for farmland, gold, and housing assets,as well as the relationship between farm-land values and expected and unexpectedinflation.

Data

We used the following 12 data series: owner-operator returns for Tippecanoe County,Indiana (1960 to 2013); farmland prices forWest Central Indiana (1960 to 2013); farm-land cash rent for West Central Indiana(1960 to 2013); ten-year treasury interestrates (1960 to 2013); S&P 500 P/E ratio(1960 to 2013); the general price level asmeasured by the implicit price deflator forpersonal consumption expenditures (PCE)(1960 to 2012); Iowa farmland prices (1911to 2012); Iowa farmland cash rent (1911 to2012); S&P returns (1911 to 2012); consumerprice index (CPI) (1911 to 2012); gold prices(1911 to 2012); and housing prices (1911 to2012).

In the first section of our analysis we usedata on owner-operator returns, cash rent,and farmland prices. While our paper focusesmore on the farmland price to cash rent mul-tiple, we think it is important to include inour analysis owner-operator returns, whichis the most important factor driving the cashrental market, and perhaps the best indica-tor of the return to land. We have access toowner-operator returns starting with 1960.Owner-operator returns from 1960–1986were created by Featherstone and Baker(1988), and these return calculations havebeen extended several times for the yearsup to 2013. A 50-50 corn-soybean rotationis assumed. Owner-operator returns arebudgeted returns using Tippecanoe County,



ownloaded from


Baker, Boehlje, and Langemeier Farmland: Is It Currently Priced as an Attractive Investment? 1323

Indiana average corn and soybean yieldseach year, season average prices of corn andsoybeans, as well as costs taken from thePurdue Crop Budget (e.g., Dobbins et al.2013) that are updated annually. Governmentpayments and set-aside are included for earlyyears according to the details of the farmprograms, and in later years (EnvironmentalWorking Group 2013) are used to estimategovernment payments.

Tippecanoe County is located in WestCentral Indiana, and we chose the farmlandvalue and cash rent data that geographicallymatches best the owner-operator returns byusing data for West Central Indiana fromthe annual Purdue farmland survey; themost recent survey is reported in Dobbinsand Cook (2013). Since 1974, in June ofeach year Purdue University has conductedits statewide survey of opinions regardingfarmland prices and cash rents for farmland(tillable, bare land) of different productiv-ity levels (top, average, poor), as well as theprice of transition farmland (moving outof agricultural production). Average sur-vey values are published for six regions ofthe state. The survey respondents include“…rural appraisers, agricultural loan offi-cers, FSA personnel, farm managers, andfarmers. The results of the survey provideinformation about the general level and trendin farmland values,” (Dobbins and Cook2013). To obtain cash rent and farmlandvalue from 1960 to 1973, the 1974 Purduesurvey numbers were indexed backwardsusing the percentage change in USDA farm-land value and cash rent data for the state ofIndiana.

The USDA farm real estate value per acrefor the State of Indiana is highly correlated(correlation coefficient of 0.989) with the Pur-due survey’s West Central Indiana farmlandvalue over the 1974–2012 period. Since theUSDA changed to the June Area Survey in1994, the correlation has been even higher at0.991. The Purdue farmland values for WestCentral Indiana are higher than the USDAvalues for most years. Our results would notchange greatly if we used USDA survey data,but we think that the West Central Indianafarmland and cash rent data are the mostconsistent with our owner-operator data.

We deflate the West Central Indiana cashrent, owner-operator returns, and farm-land prices with the PCE deflator (the PCEdeflator and interest rates on 10-year Trea-suries are gathered from the Federal Reserve

Bank of St. Louis www.research.stlouisfed.org/fred2/.). We believe that using an implicitprice deflator or consumer price index ismore appropriate than the producer priceindex (PPI) to deflate income streams suchas cash rent and owner operator returns.The PPI is primarily used to deflate revenueto measure real growth in output, whichis not our purpose. The CPI and PPI arenot terribly different; for the most part thegoods are substantially overlapping. The PPImeasures what businesses receive for goods,while the CPI measures what consumers pay.The CPI does not include capital goods (asdoes the PPI) but the PPI does not includethe prices of services. The PPI includes onlydomestic goods, whereas the CPI includesconsumer goods regardless of where theywere produced.

When it is available, such as it is from1960 to 2013, we use the PCE deflator ratherthan the CPI because we feel it is the bettermeasure.

As Chairman Bernanke said, thePCE index is generally thought tobe “the single most comprehen-sive and theoretically compellingmeasure of consumer prices.” Atthe same time, Bernanke said that“no single measure of inflation isperfect, and the Committee willcontinue to monitor a range of mea-sures when forming its view aboutinflation prospects,” (Hakkio 2008).

For our long-run (101 year) analysis, we usewhat we think are the best data available; thePCE deflator is not available for that period.Instead of the PCE, we use the annual gen-eral price level data readily available onRobert Shiller’s website, which is primar-ily based on the CPI. The S&P 500 returns(1911–2012) and P/E10, defined later, for theS&P 500 (1960–2013) are also from RobertShiller’s website.

The USDA has surveyed state-level farm-land and cash rent data since 1911 andearlier. We found gaps and inconsistenciesin the cash rent data for Indiana, and thedata for Iowa are the most consistent of theMidwest states from 1911–2012. Thus, weuse Iowa farmland price and cash rent datafrom USDA survey reports and from IowaState extension. For a point of comparisonto farmland over the 101-year period, we usedata for housing prices from Robert Shiller’s



ownloaded from



0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

Actual P/rent Average P/rent

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

Figure 1. Farmland price to cash rent muliple for West Central Indiana, 1960 to 2013

website, and gold prices from BloombergLondon gold fixing price from the Londonbullion market.1

Results

The P/rent ratio for West Central Indianahas an average value of 17.9 over the 53-yearperiod from 1960 to 2013, with a high of 31.8reached in the last year of data (2013) anda low of 11.1 in 1986, which was perhaps thebottom of the valley after the bubble of the1970s and 1980s (figure 1). At the peak of thisbubble, the P/rent multiple reached a highof just over 20 from 1977 through 1979. TheP/rent multiple subsequently dropped to theteens in the early 1980s, and reached its lowin 1986. The rise from around 15 in 1976 intothe 20s and down to 11.1 in 1986 correspondsexactly to what is viewed as the bubble infarmland prices, and one of the more difficultperiods for agriculture in modern history. Itis within this historical context that the rise inthe P/rent ratio from the upper teens in thelate 1990s to the 2013 survey value of 31.8 isof alarm.

Interest Rates

Falling interest rates help explain the recentrise in the P/rent ratio. Ten-year U.S. Treasury

1 London gold fixing price from the London bullion marketwere retrieved from the Bloomberg database, Purdue University,Parrish Library, December 3, 2013.

average annual interest rates have fallensteadily from a peak of 13.1% in 1981 toapproximately 2.0% in 2012 and 2013. Therequired rate of return by farmland investorswould be higher than the 10-year U.S. Trea-sury interest rate due to a risk premium,which is expected to be at least as large as thepremium of the farm borrowing rate over the10-year U.S. Treasury rate, which is currently1.99%.2 If farmland market participants haverequired rates of return for farmland that fol-low the 10-year treasury rate (plus a constantrisk premium), then the time series patternshown by the 10-year treasury rate providesan indication of the direction of change inthe discount rate in the constant growthmodel.

The reciprocal of the 10-year treasuryinterest rate tracks the increase in the farm-land P/rent ratio from 1985 to 2010 (figure 2).From 1960 to 1985 there is no obvious con-nection between the P/rent ratio and the10-year treasury interest rate. Also, in the lastseveral years the reciprocal of the 10-yeartreasury rate has risen much faster than theP/rent ratio. If the recent very high P/rentratios are caused by the recent extremely lowinterest rates (which results in a high recipro-cal), then the implication is that the marketexpects relatively low rates to continue overthe long term. Interest rate futures marketsand a positive slope in the treasury yieldcurve have been predicting rising interestrates for the last two years, and even though

2 The average difference between the U.S. Federal ReserveBank of Kansas City’s reported average interest rate paid onreal estate loans and 10-year U.S. Treasuries from 1988 to 2012.



ownloaded from



0.0

10.0

20.0

30.0

40.0

50.0

60.0

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

Farmland P/rent Ratio Reciprocal 10-Year Treasury

Figure 2. Farmland P/rent ratio and the reciprocal of ten-year treasuries, 1960 to 2013

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

S&P 500 P/E Ratio Farmland P/rent Ratio

Figure 3. Farmland P/rent ratio and S&P 500 P/E ratio, 1960 to 2013

this has not occurred, it is reasonable to ques-tion the long-run persistence of low interestrates required as the fundamental to justifyhigh P/rent ratios.3

Equity Investment Comparisons

We compare the P/rent ratio to stock marketindices to gain insight into the comparativeattractiveness of farmland as an investment.Figure 3 shows the P/E ratio for the S&P500 and the P/rent ratio. The average P/Eratio for the S&P 500 for the period at 18.2is relatively close to the 17.9 average for the

3 At the time of writing, interest rates on 10-year and longerU.S. Treasuries had started to increase.

P/rent ratio for farmland. With the exceptionof 1995 and 1997, the P/E ratio was higherthan the P/rent ratio from 1986 to 2003. Since2003, except for 2009 which exhibited a veryhigh P/E ratio for stocks,4 the P/rent ratio forfarmland has been higher than the P/E ratio.In addition to being relatively high, the P/rentratio has exhibited an upward trend in thelast ten years. The current P/rent ratio of 31.8is well above the average P/E multiple, andvery unattractive compared to the 2004 to2013 P/E ratio.

There are shortcomings in comparing theP/rent and P/E ratios. Cash flow received bya stock investor is the dividend on the stock,

4 The high P/E ratio for 2009 was an anomaly caused by verylow S&P earnings during the financial crisis.



ownloaded from



-20%

-15%

-10%

-5%

0%

5%

10%

15%

20%

25%

30%

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

Current 5-Year Moving Average

Figure 4. Current growth rate and five-year moving average of cash rent in West CentralIndiana, 1960 to 2013

not the earnings. The percentage of earningsper share (EPS) paid out as dividends variesfrom company to company, ranging fromzero to over 100%. Historically, the dividendpayment ratio for the S&P 500 has averagedabout 55%. In recent years, this ratio hasbeen closer to 30%. An alternative metricwould be to compare the P/rent ratio forfarmland to the price/dividends (P/D) ratiofor stocks. Dividends are of course smallerthan earnings, so price/dividend (P/D) ratiosare higher than P/E ratios. The prices ofshares of stock in companies reflect both thedividend payout and the effect of retentionon expected growth. The effect of changingpayout and retention rates is masked whenlooking at the P/E ratio at different times andcomparing different companies. Because ofthese confounding factors, comparing P/Dratios of individual companies or even stockindices to farmland is probably of limitedvalue because of the extreme variation individend payout policies.

Growth in Farmland Returns

As previously indicated, the expected long-run future growth rate in farmland returnsis one of the key variables in the constantgrowth model and is a major determinantof the P/rent ratio. Farmland market par-ticipants likely examine past growth in rentwhen anticipating future growth because it ishuman nature to use past experience whenassessing the future. Thus, one would expectparticipants in the farmland market to look

at past growth in returns, along with currentinformation about drivers of that growth toform their expectations.

Growth in cash rent (year-over-year con-tinuous percentage change)5 and the 5-yearmoving average growth rate are shown infigure 4. The mean growth in returns to landover the last 50 years has been 4.6%. The 5-year and 10-year moving averages of growthrates in cash rent have movements that aresimilar to the patterns shown in the farm-land P/rent ratio (figure 5). Extremely highgrowth rates in cash rent were experiencedin the mid 1970s, but fell through the mid1980s. The 5-year moving average smoothsthe year-to-year changes in annual growthand lags substantial increases and decreasesin growth, but generally follows the same pat-tern as the current growth rate in cash rent.While the 5-year moving average might be agood indicator of the optimism or pessimismof those in agriculture, it is hard to believethat farmland market participants expectedlong-term growth in the −5.0% range in thelate 1980s, but surely the anticipated growthin the 1980s was lower than what it was in themid 1970s. If market participants did haveexpectations following a moving average itwould explain the market’s tendency to overand under shoot. If recent growth in renthas been high, the moving average increases,and if the expectation of higher growth ratein rent follows, then the P/rent ratio would

5 The continuously compounded rate of change is calculated byconsidering the difference in the logarithms of consecutive cashrents. The continuous rate is a better measure when summingacross positive and negative growth rates.



ownloaded from



0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

-10%

-5%

0%

5%

10%

15%

20%

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

5-Year Moving Average (left axis) 10-Year Moving Average (left axis)

Farmland P/rent Ratio (right axis)

Figure 5. Farmland P/rent ratio and five- and ten-year moving average growth rates for cashrent inWest Central Indiana, 1960 to 2013

increase. The reverse would happen whenrents fall. Thus, land values would be increas-ing or decreasing both because rents increaseor decrease, and because the P/rent multipleincreases or decreases.

The logic of farmland market participants’expectations of growth following a movingaverage pattern is troublesome. When returnsgrow it is generally due to high crop prices.Long-run supply response might suggest thatslower growth would follow high growth,which is a contrarian view of growth, ratherthan believing that good times beget furthergood times.

Cyclically Adjusted P/Rent

Shiller (2005) uses a 10-year moving aver-age for earnings in the P/E ratio (oftenlabeled either P/E10 or cyclically adjustedP/E (CAPE)) to remove the effect of theeconomic cycle on the P/E ratio. When earn-ings collapse in recessions, stock prices oftendo not fall as much as earnings, and theP/E ratios based on the low current earn-ings sometimes become very large (e.g., in2009). Similarly, in good economic times P/Eratios can fall and stocks look cheap, simplybecause the very high current earnings arenot expected to last, so stock prices do notincrease as much as earnings do. By using a10-year moving average of earnings in thedenominator of the P/E ratio, Shiller (2005)has smoothed out the business cycle bydeflating both earnings and prices to remove

the effects of inflation. Shiller (2005) uses theCAPE to determine if there is informationin the CAPE in relation to future rates ofreturn. That is, when the CAPE is high, dosubsequent returns turn out to be low, andvice versa? Shiller (2005) finds a negativerelationship between CAPE and resulting 10-and 20-year cumulative returns for the S&P500.

Similar to stock earnings, farmland returnsare subject to cycles. If farmland returns arecyclical, high rents are likely to eventuallybe followed by lower rents and vice versa.If expectations over-respond to changesin returns, either directly as higher/lowerexpected returns or indirectly as a changein the expected growth rate of returns, thenthere would be a relationship between thefarmland CAPE and resulting returns onfarmland investment.

The P/rent ratios reported thus far are thecurrent year’s farmland price divided by cashrent for the same year. The P/rent10 is mod-eled after Shiller’s (2005) cyclically adjustedP/E ratio. Cash rent and farmland prices aredeflated, and then 10-year moving averagesof real cash rent are calculated. The P/rent10ratio is computed by dividing the real farm-land price by the 10-year moving average realcash rent. A similar computation is done for10-year owner-operator returns (P/00-10).

Figure 6 presents real land prices dividedby 10-year moving average real cash rentsand real owner operator returns, as well asShiller’s P/E10 ratio. The P/00-10 fell throughthe first half of the 1970s when real returnsgrew faster than land values, increased from



ownloaded from



0

10

20

30

40

50

60

P/rent10 P/00-10 P/E10

1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012

Figure 6. Ten-year moving average of cyclcally adjusted P/rent, P/00, and P/E ratios, 1960 to2013

the high teens in the mid 1970s to 28.2 in1977, and then fell to 6.8 in 1987. The P/00-10then increased steadily until it reached 37.6 in2013. In 2011 and 2012 the P/rent10 ratio rosesubstantially above the P/00-10 ratio. The fol-lowing two points are evident from figure 6.First, the P/rent10 ratio in 2013 exceeded thepeak of the S&P 500 P/E-10 ratio during thedot-com bubble. Second, the relationshipbetween the P/rent10 ratio and the P/00-10ratio suggests that producers are not biddingall of the increases in owner/operator returnsinto cash rents. Producers may be expectingowner/operator returns to decline, whichwould make it difficult to maintain high cashrents. However, this relationship could alsobe explained if one expects cash rents toadjust slowly to changes in operator returns.Historically, there have been times when cashrents were slow to adjust.

Shiller (2005) shows the relationshipbetween the P/E10 ratio and the annual-ized rate of return from holding S&P 500stocks for long periods. In general, his resultsshow that the higher the P/E10 ratio at thetime of purchase, the lower the resulting mul-tiple year returns. The West Central Indianafarmland and cash rent data from 1960 to2013 are used to compute 10- and 20-yearannualized rates of return (computed as thesum of the average of cash rent as a frac-tion of the farmland price each year, plusthe annualized price appreciation over theholding period). The results for farmland

show a negative relationship similar to thatexhibited in Shiller’s stock data. The 10-yearholding period returns for farmland show astrong negative relationship (figure 7). Thatis, the higher the P/rent10 (farmland pricedivided by 10-year moving average of cashrent) at the time of purchase, the lower theresulting 10-year rate of return. The 10-yearholding returns range from a slightly negativerate of return to 20%. The 20-year holdingperiod returns also exhibit a strong negativerelationship with the P/rent10 ratio (figure 8).The 20-year holding returns range from6–14%.

The highest historical P/rent10 in our datafor which a 10-year holding period return canbe calculated is 30 in the late 1970s, resultingin the only negative 10-year holding periodreturn in our data. The P/rent10 levels in2009 through 2012 have grown to values wellabove 30, which is literally off the chart (hor-izontal axis of figure 7). In this recent period,cash rents have increased substantially, butfarmland prices have increased more. Farm-land prices in 2013 were at a historically highmultiple of moving average rent, even higherthan the level seen in the late 1970s prior tothe agricultural crisis of the 1980s.

The high P/rent10 in 2009–2012 could bepartially explained by market participantsincorporating the current high rents intofuture expectations faster than they areincorporated into a 10-year moving average.Biofuel demand appears to be a step up in



ownloaded from



-5%

0%

5%

10%

15%

20%

25%

0.0 5.0 10.0 15.0 20.0 25.0 30.0

Figure 7. Ten-year rate of return (left axis) and P/rent10 at the time of purchase, 1960 to2013

0%

2%

4%

6%

8%

10%

12%

14%

16%

5.0 10.0 15.0 20.0 25.0 30.0

Figure 8. Twenty-year rate of return (left axis) and P/rent10 at the time of purchase, 1960 to2013

demand that is not very likely to decline sub-stantially. Similarly, increased export demand,mainly soybean demand by China, could beseen as likely to hold rather than decline.However, even if one considers the averageof only the highest two years of cash rent,one still requires a combination of growthexpectations and cost of capital that yieldsa historically large P/rent ratio to justify thecurrent price; current extremely low interestrates combined with modest growth expec-tations must continue, or if interest ratesare expected to rise, higher growth expecta-tions are needed to offset it to maintain thehistorically high P/rent ratio.

Long-term Risk, Return, and Inflation Hedge

Finally, the risk, return, and inflation hedgecharacteristics of Iowa farmland are pre-sented to investigate the attractiveness offarmland as a portfolio investment. Themeans, range, and correlation of the keyvariables used in the econometric estima-tion are shown in table 1. Iowa farmlandhas a mean return of 10.7%, which is com-parable to the S&P 500 mean return of9.0%. The 1.7% higher mean rate of returnis not significant given that landlords nor-mally have to pay some expenses out of cashrent (landlords usually pay property taxes



ownloaded from



Table 1. Summary Statistics and Correlations, 1911–2012

Correlation Coefficients

Mean (%) Iowa Expected UnexpectedVariable (Min., Max.) Farmland S&P 500 Inflation Inflation Inflation

Iowa FarmlandReturna

10.7 1.000 0.166 0.541 0.335 0.438

(−24.9, 38.3)S&P 500 Returnb 9.0 1.000 0.570 −0.046 0.094

(−59.7, 44.7)Inflation 3.1 1.000 0.650 0.760

(−11.7, 17.9)Expected

Inflation3.2 1.000 0.000

(−7.1, 14.0)Unexpected

Inflation0.0 1.000

(−10.7, 14.0)

Notes: Iowa farmland and S&P 500 returns are in nominal U.S. dollars.aIowa farmland had a mean cash rent rate of return of 6.5%, and a mean capital gain of 4.2%.bThe S&P 500 had a mean dividend rate of return of 4.2% and a mean capital gain of 4.8%.

Table 2. Regression Results of Iowa Farmland Returns, Gold Price Percentage Changes, andHousing Price Percentage Changes against S&P 500 Returns, 1911-2012

Independent Variable

DependentVariable Intercept S&P 500 Return R2

Iowa Farmland 9.8207 0.1069 0.068Gold 4.6302∗ −0.0285 0.108Housing 0.0316∗ 0.0003 0.325

Notes: Yule-Walker estimates, correcting for autocorrelation, are provided.∗Indicates statistical difference from zero at the 1% level of significance.

and sometimes pay other expenses such astile upkeep and lime). The conventionalwisdom that farmland has a competi-tive rate of return comparable to stocksappears to be supported by the 101 years ofdata.

Annual returns for the S&P 500 rangeconsiderably lower on the negative side(−59.7%) than those for Iowa farmland(−24.9%). Inflation ranges from −11.7% to17.9%, with a mean of 3.1%. Correlations intable 1 show that farmland has a much highercorrelation with inflation (0.541) than theS&P 500 return has with expected inflation(−0.046).

To determine the risk of farmland whenadded to a well-diversified portfolio, weregressed Iowa farmland returns against S&P500 returns (table 2). For comparison pur-poses, gold and housing price changes werealso regressed against S&P 500 returns. The

results show a beta of 0.1069 for farmland.Barry (1980) published a similar regressionfor U.S. farmland with a larger but stillrelatively low beta of 0.19. Irwin, Forster,and Sherrick (1988) estimated one andtwo factor models for the sample periods1950–1977 and 1947–1984, and found simi-lar market betas to that of Barry, who usedthe data period 1950–1977. However, Irwin,Forster, and Sherrick’s market betas arelarger for the more recent period (.32 in theone-factor model and .25 in the two-factormodel). On the other hand, our sampleperiod starts considerably earlier and endsconsiderably later, and we find a beta smallerthan Barry’s. However, all of these betas arerelatively small, indicating a systematic riskmuch smaller than an average stock. Theseresults support the conventional wisdom thatfarmland adds little risk to a well-diversifiedinvestment portfolio, although the results



ownloaded from



Table 3. Regression of Nominal Returns against Expected and Unexpected Inflation,1911-2012

Dependent Expected UnexpectedVariable Intercept Inflation Inflation R2

Iowa Farmland Return 7.637 0.9585∗ 1.1761∗ 0.515S&P 500 Return 10.12 −0.27 0.4740 0.011

Notes: Yule-Walker estimates, correcting for autocorrelation, are provided.∗Indicates statistical difference from zero at the 1% level of significance.

indicate that gold and housing add even lessrisk to a well-diversified portfolio with betasof −0.0285 and 0.00025 for gold and housing,respectively.

A substantial body of literature follow-ing Fama and Schwert (1977) measuresthe movement of returns for various assetclasses with inflation by regressing returnsagainst expected and unexpected inflation.The literature varies in the measures usedfor expected and unexpected inflation. Famaand Schwert (1977) used short-term inter-est rates as a proxy for expected inflation.However, since 2010 short-term governmentinterest rates have been approximately zeroand we do not believe that expected infla-tion has been zero. Also, with our data goingback to 1911, we include periods in whichexpected inflation may have been negative,whereas treasury interest rates do not go sig-nificantly negative. Thus, we follow anothermajor thread in the literature and followCaporale and Jung (1997), who assume thateconomic agents predict inflation based onpast values of inflation. Expected inflationis determined as the predicted value fromregressing current inflation on past inflationwith two lags.6 Unexpected inflation is theresidual from that regression.

Finally, we compared farmland to stocksas a hedge against inflation. Table 3 summa-rizes the results of regressing Iowa farmlandand S&P 500 returns against expected andunexpected inflation. Our results show thatIowa farmland returns move nearly one forone with both expected and unexpectedinflation. In contrast, S&P 500 returnsmove slightly in the opposite direction ofexpected inflation, and less than half (0.474times) of unexpected inflation. The con-ventional wisdom that farmland is a good

6 Thus, CPIt = 1.6016 + 0.7842 CPIt−1 − 0.2870 CPIt−2+eit ; allcoefficients are statistically significant at the 5% level, and theR squared is 0.43.

hedge against inflation is supported by theseresults.

Policy Implications

Although this analysis has focused onprivate-sector market behavior, two potentialpolicy implications are relevant. The firstpotential policy issue deals with the conse-quences of policies that enhance the demandfor agricultural products, as well as thosethat reduce the cost of capital for the sector.Biofuels policies have stimulated increaseduse of feed grains for energy production inthe form of ethanol and biodiesel, resultingin increased commodity prices and returnsto farmland. Changes in biofuels policiesthat alter that demand have potentially sig-nificant implications for commodity pricesand returns to farmland and farmland val-ues. Monetary policy focused on stimulatingeconomic growth through low interest rateshas reduced the cost of capital, resultingin a lower capitalization rate, which sup-ports higher land values. A policy shift toincreasing interest rates and consequentlycapitalization rates would be much less sup-portive of rising farmland values and mayprecipitate weakening land prices.

A second potential policy issue is the dom-inance of farmland in farmers’ investmentportfolios. This issue has two importantdimensions. First, this asset/investment con-centration means that farmers are morevulnerable in their retirement years to therisk of declining asset values, as well asthe income from those assets. It is difficultto mitigate this risk since farmland is anasset that farmers acquire as part of theirbusiness, and diverting funds to other invest-ments during their farming careers couldimpair the success of the farming business.Second, farmland is an illiquid asset, andmarkets to hedge against value changes suchas futures/options, or tradable instruments,



ownloaded from



are not readily available. A potential publicresponse to this concern would be to provideincentives for the private sector to developsuch instruments or markets through tax orother policies.

Conclusions

In this paper we have attempted to provideevidence and insight into the attractiveness offarmland as an investment portfolio choice—the long-run risk, return, and inflation hedgecharacteristics of farmland compared to otherfinancial/asset investments. Our analyses usestandard financial metrics such as P/E ratios,rates of return, and beta (β) to inform theseanalyses. The results suggest that the cur-rent P/rent ratio for farmland is substantiallyhigher than the historical ratio, and that thisratio is also high relative to the comparableP/E ratio on equities as measured by theS&P 500.

The analysis of historical rates of returnfor farmland indicates that it has returnscomparable to that of stock investments.The risk of farmland as a component of adiversified portfolio as measured by the beta(β) is very low, suggesting that it adds littlerisk to a diversified portfolio. The inflationhedge results also suggest that farmland is avery attractive inflation (both expected andunexpected) hedge investment, and is muchsuperior to stock investments.

We investigated historical cyclicallyadjusted P/rent ratios for farmland and founda negative relationship between 10- and20-year holding returns and the cyclicallyadjusted P/rent ratio at the time of purchase.Given the extremely high value of farm-land P/rent10 in 2013, we suggest caution.Even though our data confirms the con-ventional wisdom that farmland has highreturns, low risk, and is a good inflationhedge, the current P/rent10 ratio suggeststhat this is not a good time to buy. That is,those purchasing farmland today should notignore the very real possibility of “buyer’sremorse.”

References

Barry, P.J. 1980. Capital Asset Pricing andFarm Real Estate. American Journal ofAgricultural Economics 62: 549–553.

Bowman, A. 2013. Is Farmland Caughtin a Price Bubble That’s about toBurst? Retrieved from http://www.ag-professional.com/news/Farmland-bubble-faces-moment-of-truth-228809671.html.

Burt, O.R. 1986. Econometric Modeling ofthe Capitalization Formula for FarmlandPrices. American Journal of AgriculturalEconomics 68:10–26.

Caporale, T., and C. Jung. 1997. Inflation andReal Stock Rrices. Applied FinancialEconomics 7(3): 265–266.

Dobbins, C.L., and K. Cook. 2013. Up Again:Indiana’s Farmland Market in 2013.Purdue Agricultural Economics Report,Purdue University, 1–9.

Dobbins, C.L., M.R. Langemeier, W.A.Miller, B. Nielsen, T.J. Vyn, S. Casteel,B. Johnson, and K. Wise. 2013. 2014Purdue Crop Cost & Return Guide. ID-166-W, Purdue University CooperativeExtension, September.

Environmental Working Group. 2013. Ret-rieved from http://www.ewg.org/?gclid=CNeg_ceMrr4CFXQ1MgodlmMAlw.

Fama, E.F., and Schwert, G.W. 1977. AssetReturns and Inflation. Journal ofFinancial Economics 5 (2): 115–146.

Featherstone, A.M., and T.G. Baker. 1987. AnExamination of Farm Sector Real AssetDynamics: 1910–85. American Journal ofAgricultural Economics 69: 532–546.

Featherstone, A.M., and T.G. Baker. 1988.Effects of Reduced Price and IncomeSupports on Farmland Rent and Value.North Central Journal of AgriculturalEconomics 10: 177–189.

Forbes, S. 2013. Steve Romick: Trade intothe Gold You Can Eat, Farmland.[Web page] Retrieved from http://www.forbes.com/sites/steveforbes/2013/04/23/steveromick-trade-into-the-gold-you-can-eatfarmland/.

Gloy, B. 2012. When do Farm Booms BecomeBubbles? Paper presented at “Is ThisFarm Boom Different?”, a symposiumsponsored by the Federal Reserve Bankof Kansas City. July 16–17.

Gustke, C. 2013. How to Invest in Farm-land. [Web page] Retrieved fromhttp://www.bankrate.com/finance/investing/how-to-invest-in-farmland.aspx.

Hakkio, C.S. 2008. PCE and CPI InflationDifferentials: Converting InflationForecasts. Federal Reserve Bank ofKansas City Economic Review Q1:51–68.



ownloaded from

http://www.agprofessional.com/news/Farmland-bubble-faces-moment-of-truth-228809671.html



http://www.ewg.org/?gclid=CNeg_ceMrr4CFXQ1MgodlmMAlw

http://www.ewg.org/?gclid=CNeg_ceMrr4CFXQ1MgodlmMAlw

http://www.forbes.com/sites/steveforbes/2013/04/23/steveromick-trade-into-the-gold-you-can-eatfarmland/




http://www.bankrate.com/finance/investing/how-to-invest-in-farmland.aspx

http://www.bankrate.com/finance/investing/how-to-invest-in-farmland.aspx



Irwin, S.H., D.L. Forster, and B.J. Sherrick.1988. Returns to Farm Real Estate Revis-ited. American Journal of AgriculturalEconomics 70: 580–587.

Moss. C., and A.L. Katchova, 2005. Farm-land Valuation and Asset Performance.Agricultural Finance Review 65:119–130.

Lube, M. 2013. The Ins and Outs of FarmlandInvesting. [Web page] Retrieved fromhttp://beta.fool.com/whichstockswork/2013/06/06/the-ins-and-outs-of-farmland-investing/36159/.

Robinson, J. 2013. Investing in Farmland:4 Ways to Play the Agricultural Boom.[Web page] Retrieved from http://www.financialsense.com/contributors/jerry-robinson/investing-farmland-four-ways-play-agricultural-boom.

Shiller, R. Irrational Exuberance [website].Retrieved from www.irrationalexhuberance.com.

Shiller, R.J. 2005. Irrational Exuberance. 2nd

Edition. New York: Crown Business.accessed September 18, 2013.

USDA. 2014. U.S. Farm Sector FinancialIndicators, 2010-2014F. [Web page]Retrieved from http://www.ers.usda.gov/data-products/farm-income-and-wealth-statistics/us-and-state-level-farm-income-and-wealth-statistics-%28includes-the-usfarm-income-forecast-for-2014%29.aspx#. U3TBPCjvCSp.

Watts. W. L. 2013. Farmland Bubble? 10-year Rise Raises Red Flags. [web site]Retrieved from http://www.marketwatch.com/story/farmland-bubble-10-year-riseraises-red-flags-2013-10-21.



ownloaded from

http://beta.fool.com/whichstockswork/2013/06/06/the-ins-and-outs-of-farmland-investing/36159/



http://www.financialsense.com/contributors/jerryrobinson/investing-farmland-four-waysplay-agricultural-boom




http://www.ers.usda.gov/data-products/farm-income-and-wealthstatistics/us-and-state-level-farm-incomeand-wealth-statistics-%28includes-the-usfarm-income-forecast-for-2014%29.aspx






http://www.marketwatch.com/story/farmland-bubble-10-year-riseraises-red-flags-2013-10-21




FARMLAND:ISIT CURRENTLY PRICEDASAN ATTRACTIVE …€¦ · farmland portfolio of Westchester, a large farmland realtor and investment company with properties throughout the United

Documents