With only cross-sections over time rather than panel data
(Todd, 2007), TOT PSM can be written as:
TOTDDPSM= 1/NT2 (Yi2T- (i, j ) YCj2 )- )- 1/NT1 (
Yi1T - (i, j ) -Yj1C)
For NT2 = iT2 & jC2
For NT1 = iT1 & j C1
Here, YitT and YjtC , t = cfw_1,2 are the o

measured similarly across the two groups and thereby
reflect the same concepts. Second, a representative
sample survey of eligible nonparticipants as well as
participants can greatly improve the precision of the
propensity score. Also, the larger the samp

error term. Although observations are dropped to achieve
the common support, PSM increases the likelihood of
sensible comparisons across treated and matched control
units, potentially lowering bias in the program impact.
This outcome is true, however, onl

PSM approach tries to capture the effects of different
observed covariates X on participation in a single
propensity score or index. Then, outcomes of participating
and non-participating households with similar propensity
scores are compared to obtain the

the matched companies in control. The results concerning
the effects of the RDG subsidy were therefore mixed: a
positive effect with respect to employment but no effect
concerning return on total assets. This result is in
accordance with previous results

Heckman, Ichimura, and Todd (1997) encourage dropping
treatment observations with weak common support. Only
in the area of common support can inferences be made
about causality.
If conditional independence holds, and if there is a sizable
overlap in P(X )

constructed from a probit or logit model of program
participation. Caliendo and Kopeinig (2008) also provide
examples of estimations of the participation equation with
a non-binary treatment variable, based on work by
Bryson, Dorsett, and Purdon (2002); I

the balancing test was performed by dividing each
comparison and treatment group into two strata, ordered
by probability propensity scores. Within each stratum, a ttest of equality of means across participants and matched
nonparticipants was conducted for

This situation indicates that the sampled participant
farmers tend to have lower profitability under the
participation but higher profitability outside participation.
This is an unlikely scenario because it implies that
farmers who do not have a comparati

constitutes a blue print for collection, measurement and
analysis of data. Various types of research design
approaches exist to answer different research problems or
questions, the commonly used ones include that of
explanatory, exploratory, casual-compar

subset of the population as is the case here, many of the
methods suggest turning to non-exposed units (nonparticipants) in search of the missing information. They
also specify circumstances under which the use of such
information yields reliable estimate

the treated (TOT). Typically, researchers and evaluators
can ensure only internal as opposed to external validity of
the sample, so only the TOT can be estimated.
Weaker assumptions of conditional independence as well
as common support apply to estimating

2.1.1 Milk import policies
In order to promote the local dairy industry, African
policy makers tend to discourage the importation of milk
and dairy products. The import situation could be
worsened in subsequent years as the WTO globalization
policies are

misleading. For this stage of PSM, causality is not of as
much interest as the correlation of X with T.
As for the relevant covariates X, PSM will be biased if
covariates that determine participation are not included in
the participation equation for othe

without the participation. We also compare farmers'
expected performance under the participation and without
the participation.
2.3.1
Indicators for Premiums of participation
Based on equations (1) to (6), three indicators can be
constructed to compare fa

of dimensions on which to match participants and
comparison units. Nevertheless, consistent OLS estimates
of the ATE can be calculated under the assumption of
conditional exogeneity. One approach suggested by
Hirano, Imbens, and Ridder (2003) is to estima

nearby in the propensity score distribution (Heckman,
LaLonde, and Smith 1999).Specifically, the effectiveness
of PSM also depends on having a large and roughly equal
number of participant and nonparticipant observations so
that a substantial region of co

beyond the normal sampling variation will cause the
standard errors to be estimated incorrectly (see Heckman,
Ichimura, and Todd 1998).
One solution is to use bootstrapping (Efron and Tibshirani
1993; Horowitz 2003), where repeated samples are drawn
from

note that a minimal milk supply level is required for
profitable operation of such units. Encouragement of
formal and informal markets for milk is a common policy
area looked upon (D'Haese et al 2005; Leksmono et al
2006). Marketing policies are most conv

YiC Ti /Xi | .
Conditional independence is a strong assumption and is
not a directly testable criterion; it depends on specific
features of the program itself. If unobserved
characteristics determine program participation,
conditional independence will be

data records from the processors (e.g. KCC, Brookside),
the Kenya dairy board (KDB), and cooling plants records
and from farmers own kept records on milk quantities,
feed quantities, prices offered by different processors at
different times and the costs

groups came from a similar economic environment: 80
percent of Trabajar workers came from the poorest 20
percent of the population, and the study used a sample of
about 2,800 Trabajar participants along with
nonparticipants from a large national survey.
K

nonparticipant sample; these differences should be
monitored carefully to help interpret the treatment effect.
Balancing tests can also be conducted to check whether,
within each quantile of the propensity score distribution,
the average propensity score

two ways: first of all, by government regulations on milk
imports through restricted import licensing, prohibition of
fresh milk imports and imposing of specific import duties
on dairy products (Ngwoko 1986), and secondly, after the
devaluation of the Nig

within the common support and thus construct the
counterfactual outcome. Nonparametric matching
estimators such as kernel matching and LLM use a
weighted average of all nonparticipants to construct the
counterfactual match for each participant. If Pi is t

participate tend to have higher profitability than those
who did.
1_1 , 0_1 , 0_0 , and 1_0 measure farmers selection bias
on participation. There are four patterns.
(1) 1_1 > 0 ; 1_0 < 0 and 0_1 > 0; 0_0 < 0
This situation indicates that the sampled farm

provide unbiased estimation of treatment-effects. The
possibility of "bias" arises here because the effectiveness
of a treatment may depend on characteristics that are
associated with whether or not a participant in an
observational study chooses, or is c

2000; Per and Marc 2002 Missing). Most formal credit
institutions are reluctant to provide loans to dairy farmers
because they often don't have good sureties and are
susceptible to epidemics which could lead to inability to
pay debts. For easy management

common support) was used to construct a weighted match
for each participant, applying a nonparametric kernel
regression method proposed by Heckman, Ichimura, and
Todd (1998).
Gadd,Hansson and Manson (2009) while evaluating the
impact of firm subsidy using

each matched participant- nonparticipant set. As discussed
below, the choice of a particular matching technique may
therefore affect the resulting program estimate through the
weights assigned:
Nearest-neighbor matching. One of the most frequently
used m

participation minus his/her expected performance without
participation. The mean of 1 measures the sample hub
farmers average profitability premiums from
participation
(3) 0 = yc1_0i yc0_0i
According to equations (5) and (6), 0 is equal to a
sample non- p

private services may also be good, since public services
are hardly regular (Swai et al 1993). The impact of
provision of such services could be measured in several
ways. In Kenya, for example, training of farmers led to a
reduction of calf mortality from

1.1 Data Analysis and Estimation of Parameters
This study will adopt different approaches both
econometric and qualitative for testing the postulated
hypothesis concerning milk industry in the county and
inferring the happenings to the overall milk indust

estimator has the same form as the kernel-matching
estimator, except for the weighting function:
(i, j )LLR=Kij Kik(Pk-Pi)2-( Kij(Pj-Pi) Kik(Pk-Pi)
Kij Kik(Pk-Pi)2-( Kij(Pj-Pi)
Kik(Pk-Pi)
Difference-in-difference matching. With data on
participant and co

outcomes and for estimating the TOT (such as Yi = +
Ti + Xi + i ) and applying weights on the basis of the
propensity score to the matched comparison group. It can
also allow one to control for selection on unobserved
characteristics, again assuming these