26 CHAPTER 4
in LGAs, we expect some form of correlation among households in any given LGA. To account for this correlation, we estimated the bootstrapped standard errors of ATT with an option of clusters at the LGA level (Stata 2007). Given that we expect a strong correlation of the outcome in a given cluster (LGA), we expect that clustered standard errors will be larger than is the case for standard errors estimated without clustering. Hence statistical inferences from clustered standard errors are expected to be conservative.12 Using the matched samples, we also analyzed the impact of Fadama II on demand for advisory services. In that analysis, we compared the type and rate of adoption of production and postproduction technologies of Fadama II benefi- ciaries and nonbeneficiaries. We also asked the respondents using each tech- nology whether they asked for that technology. We then compared the type of technologies demanded by Fadama II beneficiaries and nonbeneficiaries. Because agriculture was among the major sectors supported by Fadama II and the adoption of improved production and marketing technologies is among the strategies that beneficiaries could use to increase their incomes, we analyzed the determinants of adoption decisions. To determine the impact of the program on adoption decisions, participation in Fadama II was included as one of the covariates. Because participation in Fadama II is an endogenous variable, conventional methods (such as fixed-effect methods with panel data) will produce biased estimates. We address this problem using a two- stage procedure, in which the estimated PSMs are used as weights in the regression model; the PSM weighting removes the bias stemming from any correlation between covariates and participation in Fadama II (Imbens and Wooldridge 2008). The two-stage weighted regression is specified as
∆Yi = β0Y0 + βi∆X + τFII, (4)
where Yi is outcome i (income or value of assets), i = 1, 2; Y0 is the initial value of the outcome of interest, X is the vector of covariates that determine
outcome Yi; and τ is a coefficient that measures the impact of FII. The vec- tor X includes the same variables used for calculating PSM, because PSM is
estimated using covariates that simultaneously affect both participation in the Fadama II and the outcomes of the program.
12 Note that the estimation of standard errors using clusters affects only the standard errors and not the coefficients.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93