Montecarlo Simulation LAB NOV 27 2009 ECON 4550. Montecarlo Simulations Monte Carlo simulation is a method of analysis based on artificially recreating.

Slides:



Advertisements
Similar presentations
The Simple Linear Regression Model Specification and Estimation Hill et al Chs 3 and 4.
Advertisements

Properties of Least Squares Regression Coefficients
The Simple Regression Model
Christopher Dougherty EC220 - Introduction to econometrics (chapter 2) Slideshow: a Monte Carlo experiment Original citation: Dougherty, C. (2012) EC220.
CmpE 104 SOFTWARE STATISTICAL TOOLS & METHODS MEASURING & ESTIMATING SOFTWARE SIZE AND RESOURCE & SCHEDULE ESTIMATING.
3.3 Omitted Variable Bias -When a valid variable is excluded, we UNDERSPECIFY THE MODEL and OLS estimates are biased -Consider the true population model:
Terminology A statistic is a number calculated from a sample of data. For each different sample, the value of the statistic is a uniquely determined number.
Objectives (BPS chapter 24)
Chapter 10: Sampling and Sampling Distributions
Lecture 8 Relationships between Scale variables: Regression Analysis
The Simple Linear Regression Model: Specification and Estimation
Econ Prof. Buckles1 Multiple Regression Analysis y =  0 +  1 x 1 +  2 x  k x k + u 1. Estimation.
Chapter 6 Introduction to Sampling Distributions
Fall 2006 – Fundamentals of Business Statistics 1 Chapter 6 Introduction to Sampling Distributions.
The Simple Regression Model
Experimental Evaluation
Ka-fu Wong © 2004 ECON1003: Analysis of Economic Data Lesson6-1 Lesson 6: Sampling Methods and the Central Limit Theorem.
1 A MONTE CARLO EXPERIMENT In the previous slideshow, we saw that the error term is responsible for the variations of b 2 around its fixed component 
Chapter 12 Section 1 Inference for Linear Regression.
Hypothesis Testing in Linear Regression Analysis
Chapter 11: Estimation Estimation Defined Confidence Levels
STA Lecture 161 STA 291 Lecture 16 Normal distributions: ( mean and SD ) use table or web page. The sampling distribution of and are both (approximately)
AP Statistics Chapter 9 Notes.
1 Theoretical Physics Experimental Physics Equipment, Observation Gambling: Cards, Dice Fast PCs Random- number generators Monte- Carlo methods Experimental.
Bootstrapping (And other statistical trickery). Reminder Of What We Do In Statistics Null Hypothesis Statistical Test Logic – Assume that the “no effect”
+ Chapter 12: Inference for Regression Inference for Linear Regression.
Managerial Economics Demand Estimation & Forecasting.
Inference for Regression Simple Linear Regression IPS Chapter 10.1 © 2009 W.H. Freeman and Company.
Inference for Regression Chapter 14. Linear Regression We can use least squares regression to estimate the linear relationship between two quantitative.
Stat 112: Notes 2 Today’s class: Section 3.3. –Full description of simple linear regression model. –Checking the assumptions of the simple linear regression.
Limits to Statistical Theory Bootstrap analysis ESM April 2006.
Business Statistics: A Decision-Making Approach, 6e © 2005 Prentice-Hall, Inc. Chap 13-1 Introduction to Regression Analysis Regression analysis is used.
The Simple Linear Regression Model: Specification and Estimation ECON 4550 Econometrics Memorial University of Newfoundland Adapted from Vera Tabakova’s.
AP STATISTICS LESSON 14 – 1 ( DAY 1 ) INFERENCE ABOUT THE MODEL.
Ka-fu Wong © 2003 Chap 6- 1 Dr. Ka-fu Wong ECON1003 Analysis of Economic Data.
Sampling and estimation Petter Mostad
Chapter 5 Sampling Distributions. The Concept of Sampling Distributions Parameter – numerical descriptive measure of a population. It is usually unknown.
Education 793 Class Notes Inference and Hypothesis Testing Using the Normal Distribution 8 October 2003.
Ka-fu Wong © 2007 ECON1003: Analysis of Economic Data Lesson0-1 Supplement 2: Comparing the two estimators of population variance by simulations.
Section 7.1 Sampling Distributions. Vocabulary Lesson Parameter A number that describes the population. This number is fixed. In reality, we do not know.
Sampling Distributions Chapter 18. Sampling Distributions A parameter is a number that describes the population. In statistical practice, the value of.
Chapter 9 Sampling Distributions 9.1 Sampling Distributions.
Chapter 4. The Normality Assumption: CLassical Normal Linear Regression Model (CNLRM)
Lecture 6 Feb. 2, 2015 ANNOUNCEMENT: Lab session will go from 4:20-5:20 based on the poll. (The majority indicated that it would not be a problem to chance,
Sampling Distributions
Chapter 4: Basic Estimation Techniques
Chapter 4 Basic Estimation Techniques
Ch. 2: The Simple Regression Model
The Simple Linear Regression Model: Specification and Estimation
Toward statistical inference
CHAPTER 12 More About Regression
Simple Linear Regression - Introduction
Ch. 2: The Simple Regression Model
Correlation and Simple Linear Regression
Confidence Intervals Tobias Econ 472.
Sampling Distribution Models
MATH 2311 Section 4.4.
Correlation and Simple Linear Regression
Undergraduated Econometrics
The Simple Linear Regression Model: Specification and Estimation
Section 7.1 Sampling Distributions
Sampling Distributions
Confidence Intervals Tobias Econ 472.
CHAPTER 12 More About Regression
Simple Linear Regression and Correlation
The Simple Regression Model
CHAPTER 12 More About Regression
Sampling Distributions
Correlation and Simple Linear Regression
Correlation and Simple Linear Regression
Presentation transcript:

Montecarlo Simulation LAB NOV ECON 4550

Montecarlo Simulations Monte Carlo simulation is a method of analysis based on artificially recreating a chance process (usually with a computer), running it many times, and directly observing the results We can use computers to draw large numbers of artificial random samples to evaluate the performance of a variety of sample-based statistics

Montecarlo Simulations Monte Carlo simulation” is a general term with many meanings “Simulation” means that we build an artificial model of a real system to study and understand the system The “Monte Carlo” part of the name comes from the capital of Monaco and reminds us of the randomness inherent in the analysis

Montecarlo Simulations We will conduct simple Monte Carlo simulations that will remind us about how the OLS performs It is like “test-driving” OLS

Montecarlo Simulations For example, in the single regression model, the parameter of interest is usually the slope parameter If the errors have zero mean, the explanatory variable is non-random and the model is correctly specified, the OLS estimator is an unbiased estimator of the slope parameter, which means that on average it will be equal to the parameter

Montecarlo Simulations If, additionally, the errors are normally distributed, the OLS estimator has a normal distribution With one given sample of observations from the population, OLS can be applied to obtain one estimate of the slope parameter That OLS estimate will be smaller or larger than the true parameter However, if estimates were obtained from a "large" number of random samples then the average estimate over all samples would equal the true parameter value

Montecarlo Simulations The above ideas can be illustrated with the help of computer simulation Repeated samples of data can be generated by artificially generating errors that follow a given distribution The properties of the OLS estimation rule can then be analyzed

Montecarlo Simulations Example m

Montecarlo Simulations Example mhttp://shazam.econ.ubc.ca/intro/mcarlo.ht m Using an amplified version of the dataset GHJ160.txt It contains data on food expenditures and income

Montecarlo Simulations Example mhttp://shazam.econ.ubc.ca/intro/mcarlo.ht m Using an amplified version of the dataset GHJ160.txt It contains data on food expenditures and income

Montecarlo Simulations We can use a simple OLS regression on the first 40 original observations to obtain: Yhat = INCOME

Montecarlo Simulations However, we will now assume that these results are not merely estimates but rather describe the true (usually deemed unobservable) model for household expenditure on food That is, let us assume that the true linear regression model is: Y = INCOME + e

Montecarlo Simulations The first order of business would be to say something about the error e We can assume it is distributed normally and homoskedastically We can find out from the original OLS regression its estimated variance

Montecarlo Simulations Now we assume that that estimated variance is actually the one driving the real process in the population That variance is , which you can obtain from SHAZAM

Montecarlo Simulations For a sample size of N=40 : Use a random number generator to generate a sample of independent and identically distributed errors with mean zero and variance The NOR function on the GENR command is used to generate normal random numbers.

Montecarlo Simulations For a sample size of N=40 : Calculate sample observations for the variable Y where INCOME is fixed Run OLS to obtain estimates of the intercept parameter and the slope parameter

Montecarlo Simulations For a sample size of N=40 : The above steps are repeated. At each replication a different set of expenditures Y is computed The number of replications is first set at 1000 The experiment then yields 1000 estimates of the slope parameter using a sample size of N=40 The sampling variability in the estimates can be summarized by plotting the empirical frequency distribution of the estimates in an histogram

Montecarlo Simulations For a sample size of N=80 : The above steps are repeated. At each replication a different set of expenditures Y is computed The number of replications is first set at 1000 The experiment then yields 1000 estimates of the slope parameter using a sample size of N=80 The sampling variability in the new set of estimates can be summarized by plotting the empirical frequency distribution of the estimates in an histogram

Montecarlo Simulations For a sample size of N=160 : The above steps are repeated. At each replication a different set of expenditures Y is computed The number of replications is first set at 1000 The experiment then yields 1000 estimates of the slope parameter using a sample size of N=160 The sampling variability in the new set of estimates can be summarized by plotting the empirical frequency distribution of the estimates in an histogram

Montecarlo Simulations Repeat the whole exercise for the case of only 200 replications and compare your results