           I have used SAS in my work since 2003. In 2011, I passed both Base and Advanced Programmer exams offered by SAS Global Certification Program   Project 1 : GMM parameter  estimation for the Double Exponential Jump-Diffusion model This is one of my term projects where I employed GMM for parameter estimation for DEJD model proposed by Ramezani, Zeng and Kou. Below you can download:   Project 2 : Mutual Fund Performance Evaluation: Bootstrap Regression in SAS The input dataset contains monthly returns for five US equity mutual funds and four Carhart factors (downloaded from CRSP). One way to obtain a relatively robust estimate of the performance measure, alpha, is to bootstrap the corresponding multiple regression. Bootstrapping assumes that both the response and covariates are random. For comparison, parametric-based quantities are also reported. More information is provided in the SAS code.     Project 3: CRSP Mutual Fund Database: Solving the Shareclass Problem Please click here   Project 4: Multivariate time series bootstrap This project is the computational part of my white paper released in Jan 2011. The input dataset includes monthly historical information about the returns of a certain hedge fund and some macroeconomic variables. The goal is to test whether the mean of a next period's return is greater than zero. For each macroeconomic predictor, the hypothesis is tested separately, and here I present the results for one such predictor, the default spread. For the reasons elaborated in the paper, the t-statistic of interest is obtained via resampling w/o replacement, a.k.a. randomization. The underlying model is VAR(2), so the residuals are resampled in pairs.    Project 5: Multiple Imputation for Panel Data Analysis Panel data can be seen as multivariate time series where some parameters are assumed the same for all cross-sections. For instance, Sufian et al. use panel regression to model macroeconomic data where “country” is a space (cross-sectional) variable and “year” is the time variable. Consider the case when there are missing observations in the input data set, and multiple imputation is necessary. In SAS, this is handled by applying PROC MI, PROC X, and PROC MIANALYZE sequentially, where X stands for the procedure that defines the core model. A common question that is asked on is “how do I make PROC MIANALYZE work with PROC X?” One can find working examples online for X = REG, but for X = PANEL the situation is not that straightforward. Unfortunately, the interface of PROC MIANALYZE is far from user friendly. Let me ask you a question: suppose MIANALYZE works fine with the input dataset named in_data1. If you create a copy of this dataset called in_data2 as   data in_data2;       set in_data1; run; will MIANALYZE work with in_data2 as well? Read my and use this sample SAS to find out.   P.S. Please let me know if you are aware of a more elegant way to interface PROC MIANALYZE with PROC PANEL. 