Adverse Selection in Developing Country Factor Markets: The Case of Fertilizers in Cambodia

Transcription

1 Adverse Seection in Deveoping Country Factor Markets: The Case of Fertiizers in Cambodia Günter Schame 1 and Friederike Höngen 2 May 2003 Abstract: We anayze the presence and potentia impact of ow quaity fertiizers, inadequate access to credit and market information on fertiizer market participation and appication rates. We expain in theory and show empiricay that farmers chose to abstain from the fertiizer market atogether or decrease their appication rates beow recommended eves because bad quaity deters a buyers not wiing to pay the market price for the average quaity fertiizer avaiabe. Key words: asymmetric information, adverse seection, quaity JEL codes: O12, D82, L15, 1 Assistant Professor and 2 Former MSc Student, Institute of Agricutura Economics and Socia Sciences, Humbodt-University Berin, Germany (contact emai: g.schame@rz.hu-berin.de) We acknowedge the Car-Duisberg Geseschaft for providing financia support to this project. Seected paper submission prepared for the 2003 AAEA Annua Meeting to be hed in Montrea, Canada, Juy 27-30, Copyright 2003 by Günter Schame. A rights reserved. Readers may make verbatim copies of this document for non-commercia purposes by any means, provided that this copyright notice appears on a such copies. However, do not cite before contacting the author(s).

2 Adverse Seection in Deveoping Country Factor Markets: The Case of Fertiizers in Cambodia Deveoping country factor markets are often fragmented and inadequatey refect suppy and demand conditions. Many studies have anayzed agricutura factor markets in deveoping countries and information deficiencies have been cited as an important reason for factor market inefficiencies. In this study, we anayze the effects of aduterated fertiizer suppies, insufficient access to credit, and imited market information on fertiizer appication rates in different regions of Cambodia. In 1993, Cambodia became a democratic monarchy and has since experienced a significant degree of market iberaization characterized by the emergence of private traders meeting growing private demand. In the eary years, the private sector fertiizer market grew rapidy from 200t in 1993 to 55,000t in 1996 whie the share of the pubic market decined steadiy (Yaptenco 1996). FAO evauated the performance of the pubic sector distribution system concuding that fertiizer stocks and distribution channes are poory managed, overstaffed, and without proper technica skis eading to higher operating costs compared to private sector traders (FAO, 1995). Provincia distribution pans are impemented poory and unreiabe suppies do not meet actua demand. Many farmers prefer buying from private sector traders, even at higher prices, because of their reative reiabiity and accessibiity of suppy and service. The whoesaer/retaier network is reativey dense, ensuring proximity of service to need (Yaptenco 1996). However, there are aso probems within the private market usuay in terms of aduterations such as the sae of underweight bags, incorrect abes, or fertiizers mixed with sand or ash. Other probems hampering the private sector incude informa road fees and imited and/or preferentia access to credit ony for seected farmers (Höngen and Suon, 1999). As a resut, a significant amount of ow quaity fertiizers has surfaced in private markets that are either tainted or od and deteriorated. Among other reasons, the presence of ow quaity fertiizers has made many farmers suspicious buyers or to abstain from the market atogether and has contributed to fertiizer appication rates we beow recommended eves (Höngen, 2000). The phenomenon just described is an exampe of adverse seection, which characterizes transactions in which the seer has reevant information that the buyer does not have, or vice versa. 2

3 The main objective of the paper is to empiricay investigate the presence and potentia impact of adverse seection on fertiizer use in addition to other impeding factors such as insufficient access to credit and imited market information. The data used in the anaysis was gathered from group interviews questioning 82 farmers in seven different regions in Cambodia participating in the Cambodia-IRRI-Austraia Project (CIAP). We are abe to show that farmers choose to abstain from fertiizer markets or decrease their appication rates we beow recommended eves because they have obtained bad quaity stocks in the past, which in turn adversey affects a current buyers. Granuated urea is the most common nitrogen fertiizer in wet rice production. In deveoping countries, it is typicay sod in kg bags. Granuated urea can be converted to ammonia onsite. The conversion rate depends on the nitrogen content. An important advantage of granuated urea is simpe handing for deivery and storage. A disadvantage reative to iquid forms is the need for de-mineraized water to convert to ammonia. Other fertiizers appied incude DAP (diammonium phosphate) and ess importanty potash. The recommended appication eves and times are purey technica in nature and are based on soi types and rice varieties determined through mutiyear soi tests for the various rice growing regions (White et a, 1997). In genera, the recommendation eves increase with soi fertiity and the suitabiity to grow high-yied varieties. The recommendations have been pubished in a handbook used in training courses for farm advisors through the Ministry of Agricuture and are thus known even in remote areas (Heer et a, 1998). According to the data, ony about 70% of a farmers choose to appy urea, whie about 30% abstain from such practices (see aso Tabe 1). 1 Farmers that choose to appy urea use on average ony 66% of the recommended appication rate per hectare. Between regions, urea recommendations vary by a factor of three (or more than 100 kg/ha) whie average appication rates differ by a factor of ten (or more than 90 kg/ha ranging from 13% - 84% of recommendations). In absoute terms, Takeo and Kanda have the highest average urea appication rates and the smaest discrepancy between recommendation and usage. Srey Veng has the owest urea appication rate, but the highest difference between recommendation and use. We empoy simpe market participation and 1 Ony 48% of a farmers choose to appy DAP (13% appy potash). 3

4 average appication rates to draw concusions with respect to the presence of adverse seection effects. This seems appropriate because of the arge variation in the absoute and reative difference between recommended and actua urea appication rates. In the next section, we present a game theoretica mode based on a simpe adverse seection argument (Varian, 1999; Gintis, 2000). With a simpe quaity choice mode, we expain how ow quaity fertiizer saes affect appication rates because of uncertainty and information deficiencies. Farmers judge the quaity of fertiizers sod in the market by the quaity of past deiveries. Because current fertiizer quaity is not easiy observabe, having obtained bad quaity stocks in the past adversey affects a current buyers. A significant amount of distrust among farmers exists that they may obtain tainted ow quaity fertiizers in the marketpace. It deters a those buyers not wiing to pay the market price for the average quaity fertiizer avaiabe from purchases, which in turn wi decrease the appication rates beow their recommended eves. In accordance with the theoretica mode, we observe a significant ack of adequate higher quaity fertiizer suppies in the market. The Mode Consider a market, in which seers (traders) know a about the vaue of their product (e.g. fertiizer), but the buyers (farmers) ony know the distribution of product quaity. For simpicity, suppose that a agents are risk neutra and the (quaity determined) fertiizer vaue for one unit v is a random variabe uniformy distributed between [v, v h ], where 0 < v < v h. Moreover, assume that the reservation price for traders is f v, that 0 < f < 1, that f is the same for a traders and that a farmers know f. For farmers, the choice is then to offer a price p. For traders, it is a best response to accept any price p f v. If we assume perfect competition, each farmer wi offer a price p=e[v p] or the expected fertiizer vaue assuming the trader is wiing to se at that price. In such a market, Pareto-optimaity is achieved when a fertiizer vaues [v, v h ] are sod, which requires p > f v h. In this case, the expected fertiizer vaue is E[v p] = (v h + v )/2 which shoud be equa to p : [1] vh + v 1+ v / vh p = f vh f

5 However, we are more interested in what fraction of fertiizer vaues is sod in genera. 2 If equation [1] fais, the range of fertiizer vaues on offer wi shrink. To see this, et V(v') be the event "true fertiizer vaue v' " and S(p) be the event "trader wiing to se fertiizer for price p." Then, for the range of reservation prices f v p < f v h, we have Pr[V( v ) S( p )] = = = = Pr[V( v )& S( p )] Pr[ S( p )] Pr[ S( p )V( v )] Pr[V( v )] Pr[ S( p )] Pr[ S( p )V( v )] ( v ( p Pr[ S( p )V( v )] ( v ( p f ) /( v f ) h ) /( v ) ) h ) The ast expression defines the uniform distribution on the interva [v, p/f ] : [2] Pr[ S( p )V( v )] ( v ( p f ) ) = ( v 0 ) /( p f ) for for v > p f f v v p f. Thus, in case equation [1] fais, i.e. f > (1+ v /v h )/2, farmers wi purchase fertiizer if and ony if p f + v p and the ony possibe Nash equiibrium yieds a price 2 p f + v p = or 2 v p =. 2 1 f Hence, the range of fertiizer vaues on offer wi be [v, v /(2f 1)], which impies that as f 1, the interva shrinks in ength to zero. This resut woud be Gresham s Law in action: bad vaue fertiizer drives out the good. 3 In the next step, we present our empirica argument in an effort to show the presence and potentia impact of adverse seection on market participation and fertiizer use in addition to other impeding factors such as insufficient access to credit and imited market information. During the data gathering investigation, we used quaitative and quantitative group interviews questioning 82 2 Notice that in order to have any fertiizer vaues sod at a, we must have p > v. 3 Named after Sir Thomas Gresham ( ) being the first to state the economic principe that bad money drives out the good, i.e. when depreciated, mutiated, or debased coinage (or currency) is in concurrent circuation with money of high vaue in terms of precious metas, the good money is withdrawn from circuation by hoarders. 5

6 farmers in seven different regions that participated in this Cambodia-IRRI-Austraia Project (CIAP). From the interviews, we obtained the foowing data: whether the farmer appies artificia fertiizer [abstain: no = 0, yes = 1], the amount of fertiizers appied per hectare, simpe quaitative indicators for having obtained ow quaity fertiizers during the past two years [aduteration: no = 0, yes = 1] and having access to market information in addition to the technica recommendations [market info: no = 0, yes = 1], as we as a very simpe normative indicator for having access to credit [credit: no credit = 0, some credit = 1, enough credit = 2]. We aso obtained information on the recommended eves of fertiizer appication [recommend. (kg/ha)] and cacuated the market participation ratio [participate (%)] for each region. Tabe 1 ists more detaied regiona characteristics incuding yieds and the percentage of urea recommendations which was actuay appied by farmers. The data was first used to estimate a simpe ogistic regression mode. The discrete dependent variabe is whether or not farmers appy urea fertiizer [abstain]. The independent variabes are whether farmers have obtained ow quaity fertiizers in the past [aduteration], the normative indicator for how easy they have access to credit [credit], and the market participation ratio [Participate] as an indicator for how ikey farmers in a particuar region are to use urea fertiizer. We expect a positive impact of aduteration variabe because having had a bad experience in the past shoud favor the current decision to abstain from the urea market. We woud anticipate a negative impact on the decision to abstain if many other farmers in a given region aready use urea fertiizers which woud mean that the traders in that region are reativey trustworthy. We aso expect a negative impact if access to credit is reativey easy for farmers. The regression resuts confirm these expectations and indicate a statisticay significant reationship between the decision to abstain and the aduteration variabe conveying the adverse seection information, as we as the indicators for ease of credit access and regiona market participation (Tabe 2). A ikeihood-ratio and Wad-test statistics are significant. The predictive power of the ogistic mode as shown in the cassification tabe is good. The estimated odds ratios indicate that abstaining from the market becomes much more ikey given a past aduteration event, whie it becomes very unikey given easy access to credit and a high market participation ratio in the region. 6

7 In a second, mutivariate regression, we estimate the impact of the indicators for aduteration and credit access, and whether farmers have access to market information [market info] as we as a categorica regiona dummy variabe on urea appication rates per hectare. For this regression, we estimate two modes. The first incudes a farmers in the sampe whie the second ony incudes those that do not abstain from the artificia fertiizer market and appy at east some urea. The resuts of this regression indicate a statisticay significant reationship between urea consumption and aduteration variabe conveying the adverse seection information (Tabe 3). Moreover, we aso obtain statisticay significant reationships between urea appication and having adequate access to credit and market information, as we as for a except one of the regiona dummy variabes. The signs of the regression coefficients are as expected, i.e. adequate access to credit and adequate market information both increase urea appication (by about 29 and 32 kg per ha, respectivey) whereas having obtained bad quaity fertiizers in the past (aduteration) decreases urea appication rates (about 22 kg per ha). The regiona impacts amount to decreases in urea consumption between 40 and 60 kg per ha reative to the base region (Kanda). For the second mode incuding ony urea users, the resuts are simiar but the aduteration variabe becomes insignificant. Concusions In this paper, we anayze the presence and potentia impact of ow quaity fertiizers, inadequate access to credit and market information on fertiizer market participation and appication rates. We show empiricay that farmers chose to abstain from the market atogether or decrease their appication rates beow recommended eves because bad quaity deters a buyers not wiing to pay the market price for the average quaity fertiizer avaiabe. Adequate access to credit expains most of the variation of urea consumption. Moreover, the resuts of the quaitative group interviews aso show that aduteration, the ack of credit and a ack of adequate market information decreases urea appication mainy for poor smahoders in remote areas who are not invoved in any training or credit programs. Given the regiona disaggregation, a cear shortcoming of the empirica anaysis is the reativey sma sampe size. Comments on this first draft are anticipated and very wecome. 7

8 Tabe 1: Regiona Characteristics Urea Use observa- yied participate recommend. average % appied of Province tions (t/ha) (%) (kg/ha) (kg/ha) recommend. Kampong Cham % % Kampong Chnang % % Kanda % % Phnom Penh % % Prey Veng % % Srey Veng % % Takeo % % Tota % % Source: Höngen and Suon, Tabe 2: Logistic Mode [dept. variabe: abstain (0,1)] Maximum Likeihood Estimation Parameter Estimate Std. Error Wad-test Est. Odds ratio CONSTANT Credit Aduteration Participate Anaysis of Deviance Source Deviance d.f. P-Vaue Mode % Residua % Tota (corr.) McFadden Pseudo R 2 = 53.5% (adj. = 45.4%) Cassification Predicted Observed 0 1 % Correct % % Overa % 69.5% 30.5% 89.0% Likeihood Ratio Tests Factor Chi-Square d.f. P-Vaue Credit % Aduteration % Participate % Source: Own Cacuations 8

9 Tabe 3: Mutivariate Mode [dept. variabe: urea consumption (kg/ha)] A farmers Urea users ony Parameter Estimate t-statistic P-Vaue Estimate t-statistic P-Vaue Constant 57.2* % 51.8* % Credit 30.0* % 29.4* % Aduteration -22.8* % % Market Info 32.8** % 39.3** % Takeo % % Kampong Chnang -40.6* % -34.5** % Kampong Cham -49.5* % -31.2** % Prey Veng -59.7* % -45.3* % Phnom Penh -48.1* % -45.7* % Srey Veng -48.1* % -57.9** % n = 82 F-Ratio = 23.3 R 2 = 74.4% R 2 (adj. for d.f.) = 71.2% Std Error of Est. = 25.8 Mean absoute error = 18.8 DW statistic = 1.90 (P=0.095) * and ** indicate significance at the 1% and 5% eve, respectivey. Source: Own Cacuations. n = 58 F-Ratio = 10.5 R 2 = 66.4% R 2 (adj. for d.f.) = 60.1% Std Error of Est. = 26.9 Mean absoute error = 18.2 DW statistic = 2.00 (P=0.185) 9

10 References: Food and Agricutura Organisation Study of the Fertiizer Sector in the Kingdom of Cambodia. FAO/TCOR Mission, May-Juy 1995, Rome. Ginten H Game Theory Evoving: A Probem Centered Introduction to Modeing Strategic Interaction. Princeton University Press. Princeton. Heer, N., P. White, and S. Mot The Cambodian Agronomic Soi Cassification is being adopted by agricutura officers and Farmers in Cambodia. In Cambodian Journa of Agricuture Vo. II No. 1. Phnom Penh, Cambodia. Höngen, F Probemfaktoren des Privaten Düngermarktes in Kambodscha und deren Einfuß auf den Ureaverbrauch der Keinbauern. Unpubished M.Sc. Thesis. Humbodt University Berin. Höngen, F. and V. Suon A study of inorganic fertiizer marketing in Cambodia. Phnom Penh, Cambodia: Cambodia-IRRI-Austraia-Project. Varian, H Intermediate Microeconomics: A Modern Approach. 5 th Edition. W.W. Norton & Company. New York. White, P., T. Oberthür, and P. Sovuthy (Eds.) The sois used for rice production in Cambodia. Cambodia-IRRI-Austraia Project. Phnom Penh, Cambodia. Yaptenco, C Fertiizer Procurement and Distribution in Cambodia: Issues and Options. Center for Nationa Poicy. Washington D.C. 10