ECON 721: Lecture Notes on Duration Analysis. Petra E. Todd

Size: px
Start display at page:

Download "ECON 721: Lecture Notes on Duration Analysis. Petra E. Todd"

Transcription

1 ECON 721: Lecture Notes on Duration Analysis Petra E. Todd Fall, 213

2 2

3 Contents 1 Two state Model, possible non-stationary Hazard function Examples Expected duration The Exponential Distribution Show that a distribution with any form of hazard function can be tranformed into a constant hazard Introducing Covariates Variety of Parametric Hazard Models How would you estimate? Multi-state models Stationary Models Alternative way of computing the likelihood if you only have information on number of spells (not on length of spell) Nonstationary Models Applications (with nonstationarity) The Problem of Left-Censored Spells (Nickell, 1979) Examples Cox s Partial MLE 21 4 Nonparametric Identification - Heckman and Singer (1981) result 23 5 Mixed Proportional Hazard Models 25 i

4 ii CONTENTS 6 Nonparametric Estimation of the Survivor Function: The Kaplan-Mier Estimator 27 7 Competing risks model 31

5 Chapter 1 Two state Model, possible non-stationary reference: Lancaster s Econometric Analysis of Transition Data Assume people can be in one of two possible states (e.g. married or unmarried, alive or dead, employed or unemployed, pregnant or not) Probability of leaving between t, t + dt having survived to t Pr(t apple T apple t + dt T t) 1.1 Hazard function Define the hazard function as By Bayes rule h(t) = lim dt! P (t apple T apple t + dt T t) dt P (t apple T apple t + dt T t) = = P (t apple T apple t + dt, T t) P (T t) F (t + dt) F (t) 1 F (t) =S(t) 1

6 2 CHAPTER 1. TWO STATE MODEL, POSSIBLE NON-STATIONARY The denominator, S(t), is called the survivor function. of exit times, and F (t) istheassociatedcdf. f(t) is the density F (t + dt) h(t) = lim dt! dt F (t) 1 S(t) = F (t) S(t) = f(t) S(t) Therefore, and Z t h(t) = h(s)ds = Z t d log(s(t)) dt d log(s(t)) ds + C ds = logs(s) t + C = logs(t) log S() + C = logs(t)+c

7 1.2. EXAMPLES 3 The initial condition S() = 1 (everyone survives in the beginning), implies that C =. The survivor function can be written as: S(t) = 1 F (t) =exp{ Z t h(s)ds} Using that h(t) = f(t) S(t) we get f(t) = h(t)s(t) =h(t)exp{. Z t h(s)ds} If exit is certain, then lim S(t) =, t!1 which implies Z t lim t!1 h(s)ds = 1, if not, then h(t) iscalledadefective hazard. 1.2 Examples

8 4 CHAPTER 1. TWO STATE MODEL, POSSIBLE NON-STATIONARY 1.3 Expected duration What is the expected total duration in a state conditional on surviving to time s? g(t t e(s) = s) = f(t) S(s) Z 1 s p.d.f. t f(t) S(s) dt = 1 S(s) Z 1 s tf(t)dt Integrate by parts: lim m!1 Z m s tf(t)dt = lim tf (t) m s m!1 Z m s F (t)dt Z m = lim {mf (m) sf (s) F (t)dt m + s + m!1 Z m = lim { ms(m)+ss(s)+ S(t)dt} m!1 = ss(s)+ Z 1 s S(t)dt s s Z m s dt}

9 1.4. THE EXPONENTIAL DISTRIBUTION 5 Thus, expected durations as of time s and as of time are: e(s) = s + 1 S(s) e() = Z 1 S(t)dt Z 1 s S(t)dt Result: Mean duration is the integral of the survivor function. 1.4 The Exponential Distribution When the hazard is independent of how long the state has been occupied, the integral is Z t h(s)ds = t The survivor function is S(t) = exp{ t}, >,t and the pdf is f(t) = exp{ t} This is the exponential probability density function. distributed exponental with parameter. We can say that T is

10 6 CHAPTER 1. TWO STATE MODEL, POSSIBLE NON-STATIONARY 1.5 Show that a distribution with any form of hazard function can be tranformed into a constant hazard Pr(T t) =S(t) =exp{ Z t h(s)ds} Let Z(T ) = Z T h(s)ds Because h(s) >, Z(T )isamonotonictransformation. Pr(T t) =Pr(Z(T ) Z(t)) = exp{ Z(t)}, This expression is the survivor function for a random variable at point Z(t), so we can conclude that Z has an exponential distribution. This shows that the integrated hazard is a unit exponential variate. A distribution with any form of the hazard function can be transformed into a constant hazard (i.e. exponential form) by a suitable transformation of the time scale. 1.6 Introducing Covariates time-invariant: sex, race, perhaps education time-varying: age, business cycle e ects, month

11 1.7. VARIETY OF PARAMETRIC HAZARD MODELS 7 h(t; x) = = P (t apple T apple t + dt T t, x) dt f(t; x) S(t; x) and S(t; x) = exp{ Z t h(s; x)ds} We could do all the estimation within x cells. Alternatively, we could specify a functional form for how the hazard function depends on observables x, suchas S(t; x) = exp{ Z t h(s; x)ds} =exp{ x } 1.7 Variety of Parametric Hazard Models Would like to use economic theory to guide in selection of appropriate functional forms, depending on context Weibull family f(t) = t 1 exp{ ( t) } h(t) = t 1 S(t) = exp{ ( t) } = exp{ x} This hazard depends on time (unless =1)butisrestrictedtobemonotonic. =1givestheexponentialhazard.

12 8 CHAPTER 1. TWO STATE MODEL, POSSIBLE NON-STATIONARY Proportional hazard h(x, t) = k 1 (x)k 2 (t) This form implies that hazards for two people with x = x 1 and x = x 2 are in the same ratio for all t, but not necessarily monotonic. k 2 is called the baseline hazard. Akeyadvantageofthismodelisthatwecanestimatek 1 (x) without having to specify k 2 (t).(moreonthislater) If k 1 (x) =k 1 (x(t)), then only get proportional hazard model if covariates vary same way for all. Box-Cox Model h(t) = k 1 (x) exp{ t ( ) }

13 1.7. VARIETY OF PARAMETRIC HAZARD MODELS 9 Box-Cox transformation: t ( ) = t 1 t (1) = t 1 t 1 e ln t 1 lim = lim!! e ln t ln t = lim (by l hopital srule)! 1 = lnt As!, we get h(t) = k 1 (x) exp{ t ( ) } = k 1 (x) exp{ ln t} = k 1 (x) t Weibull The Box-Cox generates Weibull as a special case. Still, monotonic though dt dt = t 1, which is of constant sign Generalized Box-Cox Flinn and Heckman (1982) suggest using instead h(t) =k 1 (x) exp{ 1 t ( 1) + 2 t ( 2) } (for example, 1 =1and 2 =2). This form now allows for a nonmonotonic hazard.

14 1CHAPTER 1. TWO STATE MODEL, POSSIBLE NON-STATIONARY 1.8 How would you estimate? If you observe all exit times, the likelihood is L(t; 1, 2,, 1, 2) = n i=1[1 S(t i )] = n i=1[1 exp{ Z t h(s, x)ds}] If there is no closed form solution for the survival function, then evaluating the likelihood requires numerical integration.

15 Chapter 2 Multi-state models reference: Amemiya, Chapter 11 i jk(t) t =Prob(personiobservedinstatekattimet + t in state j at time t) i If jk (t) = i jk, then the model is stationary. Will start with stationary models, then do nonstationary. 2.1 Stationary Models The probability that a person stays in state j in period (,t)andthenmoves to k in period (t, t + t) (eventa) isgivenby P (A) = (1 j t) t/ t jk t (treating t/ t) asaninteger), where is the probability of exiting j. j = MX k=1 jk jj 11

16 12 CHAPTER 2. MULTI-STATE MODELS Note that lim (1 1 n!1 n )n = e 1 Now, using stationarity, one can show that lim t! (1 j t) t/ t = exp{ jt} Thus, P (A) = exp{ jt} jk t Because t does not depend on parameters, we can drop it and regard exp{ jt} jk as the likelihood. Example 1: Suppose M=3 and event history is state 1 in period (,t 1 ) state 2 in period (t 1,t 1 + t 2 ) state 3 in period t 1 + t 2 to t 1 + t 2 + t 3 then back to state 1 The likelihood in this case is given by L = exp( 1t 1 ) 12 exp( 2t 2 ) 23 exp( 3t 3 ) 31

17 2.1. STATIONARY MODELS 13 exp( 1t 1 )istheprobabilityofsurvivingt 1 periods in state is the probability of transiting to state 2 exp( 2t 2 )istheprobabilityofsurvivinginstate2 23 is the probability of transiting to state 3, etc... If instead we observe the person leaving state 3 but do not know where the person went, then replace 31 by 3 L = exp( 1t 1 ) 12 exp( 2t 2 ) 23 exp( 3t 3 ) 3 Now suppose that we terminate observation at time t 1 + t 2 + t 3 without knowing whether a person continues to stay in state 3 or not (right censoring), then would drop term 3 altogether. Example 2: Two State model M =2, 1 = 12, 2 = 21 state #1 is unemployment, state #2 is employment Suppose an individual experiences r completed unemployment spells of length t 1,t 2,...,t r. Then L = r 1e T (contribution of r unemp spells to the likelihood) T = rx j=1 t j For overall likelihood, would also need the corresponding part for employment spells.

18 14 CHAPTER 2. MULTI-STATE MODELS Here, F (t) = 1 P (T >t)=1 e t (cdf of a r.v. signifying duration of a spell) t f(t) = e (density of observed duration spell) = f(t) 1 F (t) = e t e t hazard rate t = f(t) t 1 F (t) = Prob(leaves unemployment in t, t + t has not left up to t) is the hazard rate. Example #3: Suppose we observe one completed unemployment spell of duration t i for the ith individual: L = N i=1f i (t i )= N i=1 i exp( i t i ) where i is the probability of exiting unemployment and exp( i t i )isthe probability of being unemployed t i periods. Now suppose that individuals 1..n complete their unemployment spells of duration t i but that individuals n +1..N are right censored at time t i. The likelihood is then given by L = n i=1f i (t i ) N i=n+1[1 F i (t i )] Note how the duration model with right censoring is similar to a standard Tobit model. Previously, the likelihood depended on the observed duration t 1,t 2,...,t r only through r (number of spells) and T (total length of time in state). This implies that r and T are su cient statistics, which is a property of a stationary model.

19 2.1. STATIONARY MODELS Alternative way of computing the likelihood if you only have information on number of spells (not on length of spell) Assume that there are two completed unemployment spells and that the third spell is incomplete. T = total unemployment time Probability of observing two completed spells and one censored spell in total time T is P ( apple t 1 <t, <t 2 apple T t 1,t 3 T t 1 t 2 ) = = Z T Z T f(z 1 ) ( T ) 2 e T 2 z1 applez 1 f(z 2 ) T z 1 f(z 3 )dz 3 z 2 dz 2 dz 1 The probability of observing r completed spells in total time T is Pr(r, T )= The likelihood is given by ( T ) r e T r!, which is poisson L = N i=1 r i i exp{ it i } Now, assume that depends on individuals characteristics x i i =exp{ + x i } Amemiya, Chapter 11, derives MLE estimators for,.

20 16 CHAPTER 2. MULTI-STATE MODELS 2.2 Nonstationary Models Relax assumption that i jk (t) = i jk (constant hazard rate) Distribution function of duration under a nonstationary model F (t) = 1 exp[ f(t) = (t)exp[ Z t Z t (z)dz] (z)dz] We can write the likelihood function in the integral representation as before. Suppose that i (t) = g(x it ; ) We need to specify x it as a continuous function of t Applications (with nonstationarity) Tuma, Hannan and Groeneveld (1979) Study marriage duration. Divide sample period into 4 subperiods and assume that the hazard rate is constant between subperiods. i (t) = P x i, t 2 T p p =1, 2, 3, 4 T p is the pth subperiod

21 2.2. NONSTATIONARY MODELS 17 Lancaster (1979) Studies unemployment duration F (t) = 1 exp( t ) (Weibull distribution, nonstationary because it depends on t) (t) = t 1 (hazard function) > = < > (increasing hazard) = (constanthazard,=exponential) < (decreasing hazard) Lancaster introduced covariates by specifying i (t) = t 1 exp( x i ), x i is constant over time Lancaster found a pattern of decreasing bazard, but he would that the hazard decreased less when more covariates were included. Thus, he was concerned that the finding of negative duration dependence might be due to omitted unobservables. This led him to consider an alternative specification for the hazard rate that explicitly incorporated unobservables. µ i (t) = v i i (t), v i unobservable, assumed to be iid gamma(1, 2 ) v i is a proxy for unobservable, exogenous variables Heckman and Borjas (198) Study of unemployment duration l =lth unemployment spell experienced by an individual il (t) = t 1 exp( lx il + v i ), where v i is an unobservable that needs to be integrated out to obtain the marginal distribution function of duration. Flinn and Heckman (1982)

22 18 CHAPTER 2. MULTI-STATE MODELS Use a modified version of the Box-Cox hazard: il (t) =exp[ t lx il (t)+c l v i If you set 1 =and 2 =, get a Weibull model Here, x il (t) isassumedtobeexogenous. 1 t ] The Problem of Left-Censored Spells (Nickell, 1979) The problem occurs if individuals are not observed at the start of their unemployment spells. 2.3 Examples Three Cases considered in the literature: (i) s observed, t not observed (ii) both s and t observed (iii) t is observed but s is not observed Case (i): Analyzed by Nickell (1979) in studying unemployment Assume that the beginning of the spell (s) is observed, end of the spell (t) not observed. We observe that individual is unemployed at time t. Also, assume that the P [U started in ( s s, s)] does not depend on s (constant entry rate). For su ciently small s, g(s) s = P [U started in ( s s, s) U at ] = P [U at U started in ( s s, s)] sp [U started in ( s s, s)] (Numerator)ds = Pr(U at ) R 1

23 2.3. EXAMPLES 19 -s Start of spell Time of interview t End of spell The denominator integrates the numerator over all possible dates when the spell could have started. Assuming that P [U started in ( s s, s)] does not depend on s: = = g(s) = P [U at U started in ( s s, s)] s R 1 (Numerator)ds [1 F (s)] s [1 F (s)] s = [1 F (s)]ds ES = R 1 (show by integration by parts) sf(s)ds R 1 so, [1 F (s)]. ES Case (ii): Lancaster (1979) Assume that both s and t are observed Need the joint density g(s, t) = g(t s)g(s). The density g(s) wasderivedabove.

24 2 CHAPTER 2. MULTI-STATE MODELS Let X denote total unemployment duration. First, evaluate P (X >s+ t X >s) = = = P (X >s+ t,x >s) P (X >s) P (X >s+ t) P (X >s) 1 F (s + t). 1 F (s) (*) Therefore, Pr(X <s+ t x >s)=1 1 F (s + t) 1 F (s) = F (s + t) F (s) 1 F (s) Di erentiating with respect to t gives g(t s) = f(s + t) 1 F (s), which is the Pr(spell ends at time s+t given that it started at time s). Combining with the earlier results, we get g(s, t) = f(s + t) ES. Case (iii): Flinn and Heckman (1982): t is observed but s is not observed obtain g(t) byintegratingg(t, s) withrespecttos : g(t) = 1 ES Z 1 = 1 F (t). ES f(s + t)ds

25 Chapter 3 Cox s Partial MLE Consider proportional hazard model of the form i (t) = (t)exp{ x i } and assume that data are right censored (observe start of spell but do not observe length of spell for some individuals). Let t i,i =1, 2,...,n be completed durations and let t i,i= n +1,n+2,n+ 3,...,N be censored durations. The likelihood is given by Z t L = n i=1 exp{ x i } (t i )exp[ exp( x i ) N i=n+1 exp[ exp( x i ) Z t (z)dz]. (z)dz] Through some algebraic manipulation, can obtain n i=1 exp( x i ) (t i )exp{ Z 1 [ X h2r(t) exp( x h )] (t)dt}, where R(t) ={i t i t}. 21

26 22 CHAPTER 3. COX S PARTIAL MLE Cox (1975) suggested and Tsiatis (1981) proved that the likelihood could be decomposed into two components: L = L 1 L " 2 # = n exp( x i ) i=1 P h2r(t) exp( x h ) 2 3 X exp( x h ) (t i ) 5 exp 4 n i=1 Z 1 [ X h2r(t) 1 exp( x h )] (t)dta can be ob- and that a consistent and asymptotically normal estimator for tained by maximizing L 1 (the partial MLE). This means that can be estimated without specifying (t). This is a remarkable result, given that L 1 is not a proper likelihood. See Amemiya, Ch. 11, for an interpretation of the di erent components.

27 Chapter 4 Nonparametric Identification - Heckman and Singer (1981) result Ask what features of hazard functions can be identified from the raw data (i.e. G(t x)). Denote unobservables by. Would like to infer properties of G(t x, ) without having to impose strong parametric assumptions, either on µ( ) orh(t x, ). Assume that x(t) is constant. Heckman and Singer (1981) show that if G(t x) exhibits positive duration dependence, then it must be that h(t x, ) also exhibits positive duration dependence over some interval of values in those intervals of t. Also, show that omitting omitted variables leads to tendency towards negative duration dependence. Consider hazard of the form h(t x, ) = (t) (x), 23

28 24CHAPTER 4. NONPARAMETRIC IDENTIFICATION - HECKMAN AND SINGER (1981 (time- which gives the proportional hazard model with (t) =, x(t) =x invariant unobserved heterogeneity and time invariant regressors). Let h(t x) betheconditionalhazard,notcontrollingforunobservables. Let F (t x, ) andf (t x) betheconditionaldistributions. h(t x) = = R f(t x, )dµ( ) R [1 F (t x, )]dµ( ) R h(t x, )(1 F (t x, ))dµ( ) R [1 F (t x, )]dµ( = x, (1 F (t x, ))dµ( ) R [1 F (t x, )]dµ( ) +nonpositiveterm. This shows that ignoring unobservables will bias the hazard downwards.

29 Chapter 5 Mixed Proportional Hazard Models Consider the Cox Proportional Hazard model: h(t X;, )= (t)exp( x) Cox (1972) observed that it is possible to estimate parameters without specifying (t) using the partial likelihood approach discussed earlier. It is a semi parametric estimation approach, because (t) isleftunspecified. Now, suppose we want to introduce unobservables and want to be flexible about the way in which regressors enter. Let the unobservables be distributed, which is unknown to the econometrician. The model that includes unobservables is called the Mixed Proportional Hazard Model. h(t X, U;, ) = (t)exp(z(x, ))exp(u) Elbers and Ridder (1982) and Heckman and Singer (1984) Showed that this model, with regressors, is non parametrically identified under some restrictions on. Elbers and Ridder (1982) assumed that E(exp(U)) < 1 (Heckman and Singer (1984) consider alternative assumptions). 25

30 26 CHAPTER 5. MIXED PROPORTIONAL HAZARD MODELS Honore (1993) Studied a multi-spell generalization where T 1 and T 2 are the length of multiple spells. He showed the model can be identified under weaker conditions on. Hahn (1994) derives the semi parametric e ciency bound, whichprovidesaboundonthe attainable statistical accuracy (it was first introduced by Stein (1956) and has been derived for a number of di erent models). Because semi parametric estimation must be at least as di cult as any parametric sub model (model allowed under the semi parametric model), it follows that the asymptotic variance of any p N-consistent estimator is no smaller than the supremum of the Cramer-Rao lower bounds for all parametric sub models. The infimum of the information matrix for,theinverseofthe Cramer-Rao lower bound, gives the semi parametric version of the information matrix, called the semi parametric information bound. Hahn (1994) shows that for the single-spell Weibull mixed proportional hazard model, the information matrix is singular, which implies that there cannot exist a p N-consistent estimator. He also shows this to be the case for the multi-spell version of the model. Thus, even though the model is identified, there is no p N-consistent estimator. Ridder and Woutersen (23) Present new conditions for the mixed proportional hazard model under which parameters are identified and under which the information matrix is nonsingular. The paper also presents an estimator that converges at a p N rate. The key additional assumption is that the baseline hazard needs to be bounded away from and 1 near t =.

31 Chapter 6 Nonparametric Estimation of the Survivor Function: The Kaplan-Mier Estimator Allow for right censored exit times No regressors, but could accommodate regressors by doing everything within x cells. Does not allow for unobservables N possibly right-censored exit times M apple N distinct exit times t (1), t (2),t (3),...,t (M), where multiple people can exit at the same time n j =number leaving at time t j. Ŝ(t) = 1 = 1 number leaving before t N n 1 + n n k, k =max such that t j <t N j 27

32 28CHAPTER 6. NONPARAMETRIC ESTIMATION OF THE SURVIVOR FUNCTION: TH We can write Ŝ(t) = N n 1 n 2.. n k N N n1 N n1 n 2 N n1 n 2 n 3 = N N n 1 N n 1 n 2 N n1 n 2... n n N n 1... n k 1 n 2 n 3 = n 1 1 N N n 1 n n N n 1... n k 1, N n 1 n 2... where the ratios represent the number leaving of those who survive (the risk set). Thus, the hazard function is: ˆ j = n j N n 1... n j 1 The survivor function a time t (which would be used to handle right censoring) can be written as: Ŝ(t) = tj <t(1 ˆ j ) (called a product limit estimator) The term ˆ j is the hazard rate. up to that date.) (prob of leaving at date t j given survived The Kaplan Mier estimator of the survivor function looks like a step function:

33 29 Kaplan-Mier Survivor Function 1. t

34 3CHAPTER 6. NONPARAMETRIC ESTIMATION OF THE SURVIVOR FUNCTION: TH

35 Chapter 7 Competing risks model J causes of failure 1..J T j latent failure time from cause j Observe duration to first failure and associated cause. That is, observe time of death and cause of death. (e.g. observe death from cancer or heart disease) (T,I) = {min j (T j ), argmin j (T j )} In application, there may be considerable content in models with regressors. For example, the goal may be to study how smoking, blood pressure and weight a ect the marginal distribution of time to death attributable to heart attack or cancer. In models without regressors, need to make functional form assumptions about the joint distribution of failure times. Heckman and Honore (1989) study identification in models with regressors. 31

36 32 CHAPTER 7. COMPETING RISKS MODEL For example, in Cox proportional hazard model: S(t x) = exp{ z(t)'(x)} ('(x) usuallye x ), we can assume that each potential failure time has a proportional hazard specification. Can specify joint survivor function of T 1,T 2 conditional on x. S(t 1,t 2 x) =K[exp{ Z 1 (t 1 )' 1 (x)}, exp{ Z 2 (t 2 )' 2 (x)}] Theorem: Assume that (T 1,T 2 )hasthejointsurvivorfunctionasgivenabove. Then ' 1,' 2,Z 1 and Z 2 are identified from the observed minimum of (T 1,T 2 )under the following assumptions: (i) K is c 1 with partial derivatives K 1 and K 2 and for i =1, 2thelimitas n!1of K( 1n, 2n ) is finite for all sequences 1n! 1, 2n! 1forn!1. K is strictly increasing in each of its arguments in all of [ 1] [ 1] (ii) Z 1 (1) = 1,Z 2 (1) = 1,' 1 (x )=1,' 2 (x ) = 1 for some fixed point in the support of x. (iii) The support of {' 1 (x),' 2 (x)} is ( 1) ( 1) (iv) Z 1 and Z 2 are nonnegative, di erentiable, strictly increasing functions, except that we allow them to be 1 for finite t. Sketch of Proof: Define: Q 1 (t) = pr(t 1 >t,t 2 >T 1 ) (die from cause 1 by time T 1 ) Q 2 (t) = pr(t 2 >t,t 1 >T 2 ) (die from cause 2 by time T 2 ) Q 1(t) = K 1 [exp{ Z 1 (t)' 1 (x)} exp{ Z 2 (t 2 )' 2 (x)}]exp{ Z 1 (t)' 1 (x)}z 1(t)' 1 (x)

37 The ratio of Q 1 at two points x and x (in the support of X) andusingthe assumptions on x (assumption (iii)) gives: 33 = K 1 [exp{ Z 1 (t)' 1 (x)} exp{ Z 2 (t 2 )' 2 (x)}]exp{ Z 1 (t)' 1 (x)}z 1(t)' 1 (x) K 1 [exp{ Z 1 (t)' 1 (x )} exp{ Z 2 (t 2 )' 2 (x )}]exp{ Z 1 (t)' 1 (x )}Z 1(t)' 1 (x ) Take lim above expression=' 1 (x) t! By a symmetric argument, we can identify ' 2 (x). Heckman and Honore also provide approaches to identify K, Z 1 (t),z 2 (t).

Duration Analysis. Joan Llull

Duration Analysis. Joan Llull Duration Analysis Joan Llull Panel Data and Duration Models Barcelona GSE joan.llull [at] movebarcelona [dot] eu Introduction Duration Analysis 2 Duration analysis Duration data: how long has an individual

More information

Dynamic Models Part 1

Dynamic Models Part 1 Dynamic Models Part 1 Christopher Taber University of Wisconsin December 5, 2016 Survival analysis This is especially useful for variables of interest measured in lengths of time: Length of life after

More information

THE SINGULARITY OF THE INFORMATION MATRIX OF THE MIXED PROPORTIONAL HAZARD MODEL

THE SINGULARITY OF THE INFORMATION MATRIX OF THE MIXED PROPORTIONAL HAZARD MODEL Econometrica, Vol. 71, No. 5 (September, 2003), 1579 1589 THE SINGULARITY OF THE INFORMATION MATRIX OF THE MIXED PROPORTIONAL HAZARD MODEL BY GEERT RIDDER AND TIEMEN M. WOUTERSEN 1 This paper presents

More information

Lecture 22 Survival Analysis: An Introduction

Lecture 22 Survival Analysis: An Introduction University of Illinois Department of Economics Spring 2017 Econ 574 Roger Koenker Lecture 22 Survival Analysis: An Introduction There is considerable interest among economists in models of durations, which

More information

Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis

Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis Stock Sampling with Interval-Censored Elapsed Duration: A Monte Carlo Analysis Michael P. Babington and Javier Cano-Urbina August 31, 2018 Abstract Duration data obtained from a given stock of individuals

More information

In contrast, parametric techniques (fitting exponential or Weibull, for example) are more focussed, can handle general covariates, but require

In contrast, parametric techniques (fitting exponential or Weibull, for example) are more focussed, can handle general covariates, but require Chapter 5 modelling Semi parametric We have considered parametric and nonparametric techniques for comparing survival distributions between different treatment groups. Nonparametric techniques, such as

More information

Cox s proportional hazards model and Cox s partial likelihood

Cox s proportional hazards model and Cox s partial likelihood Cox s proportional hazards model and Cox s partial likelihood Rasmus Waagepetersen October 12, 2018 1 / 27 Non-parametric vs. parametric Suppose we want to estimate unknown function, e.g. survival function.

More information

Survival Analysis. Lu Tian and Richard Olshen Stanford University

Survival Analysis. Lu Tian and Richard Olshen Stanford University 1 Survival Analysis Lu Tian and Richard Olshen Stanford University 2 Survival Time/ Failure Time/Event Time We will introduce various statistical methods for analyzing survival outcomes What is the survival

More information

Introduction: structural econometrics. Jean-Marc Robin

Introduction: structural econometrics. Jean-Marc Robin Introduction: structural econometrics Jean-Marc Robin Abstract 1. Descriptive vs structural models 2. Correlation is not causality a. Simultaneity b. Heterogeneity c. Selectivity Descriptive models Consider

More information

Multistate Modeling and Applications

Multistate Modeling and Applications Multistate Modeling and Applications Yang Yang Department of Statistics University of Michigan, Ann Arbor IBM Research Graduate Student Workshop: Statistics for a Smarter Planet Yang Yang (UM, Ann Arbor)

More information

Estimating the Derivative Function and Counterfactuals in Duration Models with Heterogeneity

Estimating the Derivative Function and Counterfactuals in Duration Models with Heterogeneity Estimating the Derivative Function and Counterfactuals in Duration Models with Heterogeneity Jerry Hausman and Tiemen Woutersen MIT and University of Arizona February 2012 Abstract. This paper presents

More information

Survival Analysis. Stat 526. April 13, 2018

Survival Analysis. Stat 526. April 13, 2018 Survival Analysis Stat 526 April 13, 2018 1 Functions of Survival Time Let T be the survival time for a subject Then P [T < 0] = 0 and T is a continuous random variable The Survival function is defined

More information

Exercises. (a) Prove that m(t) =

Exercises. (a) Prove that m(t) = Exercises 1. Lack of memory. Verify that the exponential distribution has the lack of memory property, that is, if T is exponentially distributed with parameter λ > then so is T t given that T > t for

More information

UNIVERSITY OF CALIFORNIA, SAN DIEGO

UNIVERSITY OF CALIFORNIA, SAN DIEGO UNIVERSITY OF CALIFORNIA, SAN DIEGO Estimation of the primary hazard ratio in the presence of a secondary covariate with non-proportional hazards An undergraduate honors thesis submitted to the Department

More information

Survival Analysis: Weeks 2-3. Lu Tian and Richard Olshen Stanford University

Survival Analysis: Weeks 2-3. Lu Tian and Richard Olshen Stanford University Survival Analysis: Weeks 2-3 Lu Tian and Richard Olshen Stanford University 2 Kaplan-Meier(KM) Estimator Nonparametric estimation of the survival function S(t) = pr(t > t) The nonparametric estimation

More information

STAT331. Cox s Proportional Hazards Model

STAT331. Cox s Proportional Hazards Model STAT331 Cox s Proportional Hazards Model In this unit we introduce Cox s proportional hazards (Cox s PH) model, give a heuristic development of the partial likelihood function, and discuss adaptations

More information

MAS3301 / MAS8311 Biostatistics Part II: Survival

MAS3301 / MAS8311 Biostatistics Part II: Survival MAS3301 / MAS8311 Biostatistics Part II: Survival M. Farrow School of Mathematics and Statistics Newcastle University Semester 2, 2009-10 1 13 The Cox proportional hazards model 13.1 Introduction In the

More information

Other Survival Models. (1) Non-PH models. We briefly discussed the non-proportional hazards (non-ph) model

Other Survival Models. (1) Non-PH models. We briefly discussed the non-proportional hazards (non-ph) model Other Survival Models (1) Non-PH models We briefly discussed the non-proportional hazards (non-ph) model λ(t Z) = λ 0 (t) exp{β(t) Z}, where β(t) can be estimated by: piecewise constants (recall how);

More information

3003 Cure. F. P. Treasure

3003 Cure. F. P. Treasure 3003 Cure F. P. reasure November 8, 2000 Peter reasure / November 8, 2000/ Cure / 3003 1 Cure A Simple Cure Model he Concept of Cure A cure model is a survival model where a fraction of the population

More information

Lecture 3. Truncation, length-bias and prevalence sampling

Lecture 3. Truncation, length-bias and prevalence sampling Lecture 3. Truncation, length-bias and prevalence sampling 3.1 Prevalent sampling Statistical techniques for truncated data have been integrated into survival analysis in last two decades. Truncation in

More information

ECONOMETRICS FIELD EXAM Michigan State University May 9, 2008

ECONOMETRICS FIELD EXAM Michigan State University May 9, 2008 ECONOMETRICS FIELD EXAM Michigan State University May 9, 2008 Instructions: Answer all four (4) questions. Point totals for each question are given in parenthesis; there are 00 points possible. Within

More information

CIMAT Taller de Modelos de Capture y Recaptura Known Fate Survival Analysis

CIMAT Taller de Modelos de Capture y Recaptura Known Fate Survival Analysis CIMAT Taller de Modelos de Capture y Recaptura 2010 Known Fate urvival Analysis B D BALANCE MODEL implest population model N = λ t+ 1 N t Deeper understanding of dynamics can be gained by identifying variation

More information

Step-Stress Models and Associated Inference

Step-Stress Models and Associated Inference Department of Mathematics & Statistics Indian Institute of Technology Kanpur August 19, 2014 Outline Accelerated Life Test 1 Accelerated Life Test 2 3 4 5 6 7 Outline Accelerated Life Test 1 Accelerated

More information

Econometric Analysis of Cross Section and Panel Data

Econometric Analysis of Cross Section and Panel Data Econometric Analysis of Cross Section and Panel Data Jeffrey M. Wooldridge / The MIT Press Cambridge, Massachusetts London, England Contents Preface Acknowledgments xvii xxiii I INTRODUCTION AND BACKGROUND

More information

A Simple GMM Estimator for the Semiparametric Mixed Proportional Hazard Model. February 2013

A Simple GMM Estimator for the Semiparametric Mixed Proportional Hazard Model. February 2013 A Simple GMM Estimator for the Semiparametric Mixed Proportional Hazard Model G E. B, G R T W NIDI, U S C, U A February 213 A. Ridder and Woutersen 23) have shown that under a weak condition on the baseline

More information

STAT 6350 Analysis of Lifetime Data. Failure-time Regression Analysis

STAT 6350 Analysis of Lifetime Data. Failure-time Regression Analysis STAT 6350 Analysis of Lifetime Data Failure-time Regression Analysis Explanatory Variables for Failure Times Usually explanatory variables explain/predict why some units fail quickly and some units survive

More information

Lecture 5 Models and methods for recurrent event data

Lecture 5 Models and methods for recurrent event data Lecture 5 Models and methods for recurrent event data Recurrent and multiple events are commonly encountered in longitudinal studies. In this chapter we consider ordered recurrent and multiple events.

More information

Survival Analysis Math 434 Fall 2011

Survival Analysis Math 434 Fall 2011 Survival Analysis Math 434 Fall 2011 Part IV: Chap. 8,9.2,9.3,11: Semiparametric Proportional Hazards Regression Jimin Ding Math Dept. www.math.wustl.edu/ jmding/math434/fall09/index.html Basic Model Setup

More information

ST495: Survival Analysis: Maximum likelihood

ST495: Survival Analysis: Maximum likelihood ST495: Survival Analysis: Maximum likelihood Eric B. Laber Department of Statistics, North Carolina State University February 11, 2014 Everything is deception: seeking the minimum of illusion, keeping

More information

Using Matching, Instrumental Variables and Control Functions to Estimate Economic Choice Models

Using Matching, Instrumental Variables and Control Functions to Estimate Economic Choice Models Using Matching, Instrumental Variables and Control Functions to Estimate Economic Choice Models James J. Heckman and Salvador Navarro The University of Chicago Review of Economics and Statistics 86(1)

More information

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Winter 2014 Instructor: Victor Aguirregabiria

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Winter 2014 Instructor: Victor Aguirregabiria ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Winter 2014 Instructor: Victor guirregabiria SOLUTION TO FINL EXM Monday, pril 14, 2014. From 9:00am-12:00pm (3 hours) INSTRUCTIONS:

More information

Multistate models and recurrent event models

Multistate models and recurrent event models Multistate models Multistate models and recurrent event models Patrick Breheny December 10 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/22 Introduction Multistate models In this final lecture,

More information

Statistical Inference and Methods

Statistical Inference and Methods Department of Mathematics Imperial College London d.stephens@imperial.ac.uk http://stats.ma.ic.ac.uk/ das01/ 31st January 2006 Part VI Session 6: Filtering and Time to Event Data Session 6: Filtering and

More information

ST5212: Survival Analysis

ST5212: Survival Analysis ST51: Survival Analysis 8/9: Semester II Tutorial 1. A model for lifetimes, with a bathtub-shaped hazard rate, is the exponential power distribution with survival fumction S(x) =exp{1 exp[(λx) α ]}. (a)

More information

Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals. John W. Mac McDonald & Alessandro Rosina

Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals. John W. Mac McDonald & Alessandro Rosina Mixture modelling of recurrent event times with long-term survivors: Analysis of Hutterite birth intervals John W. Mac McDonald & Alessandro Rosina Quantitative Methods in the Social Sciences Seminar -

More information

MAS3301 / MAS8311 Biostatistics Part II: Survival

MAS3301 / MAS8311 Biostatistics Part II: Survival MAS330 / MAS83 Biostatistics Part II: Survival M. Farrow School of Mathematics and Statistics Newcastle University Semester 2, 2009-0 8 Parametric models 8. Introduction In the last few sections (the KM

More information

Semiparametric Estimation of a Panel Data Proportional Hazards Model with Fixed Effects

Semiparametric Estimation of a Panel Data Proportional Hazards Model with Fixed Effects Semiparametric Estimation of a Panel Data Proportional Hazards Model with Fixed Effects Joel L. Horowitz Department of Economics Northwestern University Evanston, IL 60208 and Sokbae Lee Department of

More information

Nonparametric Identi cation and Estimation of Truncated Regression Models with Heteroskedasticity

Nonparametric Identi cation and Estimation of Truncated Regression Models with Heteroskedasticity Nonparametric Identi cation and Estimation of Truncated Regression Models with Heteroskedasticity Songnian Chen a, Xun Lu a, Xianbo Zhou b and Yahong Zhou c a Department of Economics, Hong Kong University

More information

Remarks on Structural Estimation The Search Framework

Remarks on Structural Estimation The Search Framework Remarks on Structural Estimation The Search Framework Christopher Flinn NYU and Collegio Carlo Alberto November 2009 1 The Estimation of Search Models We develop a simple model of single agent search set

More information

Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis

Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis Statistics 262: Intermediate Biostatistics Non-parametric Survival Analysis Jonathan Taylor & Kristin Cobb Statistics 262: Intermediate Biostatistics p.1/?? Overview of today s class Kaplan-Meier Curve

More information

7.1 The Hazard and Survival Functions

7.1 The Hazard and Survival Functions Chapter 7 Survival Models Our final chapter concerns models for the analysis of data which have three main characteristics: (1) the dependent variable or response is the waiting time until the occurrence

More information

Semiparametric Estimation with Mismeasured Dependent Variables: An Application to Duration Models for Unemployment Spells

Semiparametric Estimation with Mismeasured Dependent Variables: An Application to Duration Models for Unemployment Spells Semiparametric Estimation with Mismeasured Dependent Variables: An Application to Duration Models for Unemployment Spells Jason Abrevaya University of Chicago Graduate School of Business Chicago, IL 60637,

More information

Separate Appendix to: Semi-Nonparametric Competing Risks Analysis of Recidivism

Separate Appendix to: Semi-Nonparametric Competing Risks Analysis of Recidivism Separate Appendix to: Semi-Nonparametric Competing Risks Analysis of Recidivism Herman J. Bierens a and Jose R. Carvalho b a Department of Economics,Pennsylvania State University, University Park, PA 1682

More information

Multi-state Models: An Overview

Multi-state Models: An Overview Multi-state Models: An Overview Andrew Titman Lancaster University 14 April 2016 Overview Introduction to multi-state modelling Examples of applications Continuously observed processes Intermittently observed

More information

What s New in Econometrics? Lecture 14 Quantile Methods

What s New in Econometrics? Lecture 14 Quantile Methods What s New in Econometrics? Lecture 14 Quantile Methods Jeff Wooldridge NBER Summer Institute, 2007 1. Reminders About Means, Medians, and Quantiles 2. Some Useful Asymptotic Results 3. Quantile Regression

More information

Multistate models and recurrent event models

Multistate models and recurrent event models and recurrent event models Patrick Breheny December 6 Patrick Breheny University of Iowa Survival Data Analysis (BIOS:7210) 1 / 22 Introduction In this final lecture, we will briefly look at two other

More information

A Bayesian Nonparametric Approach to Causal Inference for Semi-competing risks

A Bayesian Nonparametric Approach to Causal Inference for Semi-competing risks A Bayesian Nonparametric Approach to Causal Inference for Semi-competing risks Y. Xu, D. Scharfstein, P. Mueller, M. Daniels Johns Hopkins, Johns Hopkins, UT-Austin, UF JSM 2018, Vancouver 1 What are semi-competing

More information

β j = coefficient of x j in the model; β = ( β1, β2,

β j = coefficient of x j in the model; β = ( β1, β2, Regression Modeling of Survival Time Data Why regression models? Groups similar except for the treatment under study use the nonparametric methods discussed earlier. Groups differ in variables (covariates)

More information

MC3: Econometric Theory and Methods. Course Notes 4

MC3: Econometric Theory and Methods. Course Notes 4 University College London Department of Economics M.Sc. in Economics MC3: Econometric Theory and Methods Course Notes 4 Notes on maximum likelihood methods Andrew Chesher 25/0/2005 Course Notes 4, Andrew

More information

Syllabus. By Joan Llull. Microeconometrics. IDEA PhD Program. Fall Chapter 1: Introduction and a Brief Review of Relevant Tools

Syllabus. By Joan Llull. Microeconometrics. IDEA PhD Program. Fall Chapter 1: Introduction and a Brief Review of Relevant Tools Syllabus By Joan Llull Microeconometrics. IDEA PhD Program. Fall 2017 Chapter 1: Introduction and a Brief Review of Relevant Tools I. Overview II. Maximum Likelihood A. The Likelihood Principle B. The

More information

Stochastic Modelling Unit 1: Markov chain models

Stochastic Modelling Unit 1: Markov chain models Stochastic Modelling Unit 1: Markov chain models Russell Gerrard and Douglas Wright Cass Business School, City University, London June 2004 Contents of Unit 1 1 Stochastic Processes 2 Markov Chains 3 Poisson

More information

Estimation of discrete time (grouped duration data) proportional hazards models: pgmhaz

Estimation of discrete time (grouped duration data) proportional hazards models: pgmhaz Estimation of discrete time (grouped duration data) proportional hazards models: pgmhaz Stephen P. Jenkins ESRC Research Centre on Micro-Social Change University of Essex, Colchester

More information

Censoring mechanisms

Censoring mechanisms Censoring mechanisms Patrick Breheny September 3 Patrick Breheny Survival Data Analysis (BIOS 7210) 1/23 Fixed vs. random censoring In the previous lecture, we derived the contribution to the likelihood

More information

ST745: Survival Analysis: Cox-PH!

ST745: Survival Analysis: Cox-PH! ST745: Survival Analysis: Cox-PH! Eric B. Laber Department of Statistics, North Carolina State University April 20, 2015 Rien n est plus dangereux qu une idee, quand on n a qu une idee. (Nothing is more

More information

Joint Modeling of Longitudinal Item Response Data and Survival

Joint Modeling of Longitudinal Item Response Data and Survival Joint Modeling of Longitudinal Item Response Data and Survival Jean-Paul Fox University of Twente Department of Research Methodology, Measurement and Data Analysis Faculty of Behavioural Sciences Enschede,

More information

Duration Models and Point Processes

Duration Models and Point Processes DISCUSSION PAPER SERIES IZA DP No. 2971 Duration Models and Point Processes Jean-Pierre Florens Denis Fougère Michel Mouchart August 27 Forschungsinstitut zur Zukunft der Arbeit Institute for the Study

More information

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data

Approximation of Survival Function by Taylor Series for General Partly Interval Censored Data Malaysian Journal of Mathematical Sciences 11(3): 33 315 (217) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Journal homepage: http://einspem.upm.edu.my/journal Approximation of Survival Function by Taylor

More information

Analysis of Gamma and Weibull Lifetime Data under a General Censoring Scheme and in the presence of Covariates

Analysis of Gamma and Weibull Lifetime Data under a General Censoring Scheme and in the presence of Covariates Communications in Statistics - Theory and Methods ISSN: 0361-0926 (Print) 1532-415X (Online) Journal homepage: http://www.tandfonline.com/loi/lsta20 Analysis of Gamma and Weibull Lifetime Data under a

More information

A nonparametric test for path dependence in discrete panel data

A nonparametric test for path dependence in discrete panel data A nonparametric test for path dependence in discrete panel data Maximilian Kasy Department of Economics, University of California - Los Angeles, 8283 Bunche Hall, Mail Stop: 147703, Los Angeles, CA 90095,

More information

Testing for Regime Switching: A Comment

Testing for Regime Switching: A Comment Testing for Regime Switching: A Comment Andrew V. Carter Department of Statistics University of California, Santa Barbara Douglas G. Steigerwald Department of Economics University of California Santa Barbara

More information

Economics 241B Estimation with Instruments

Economics 241B Estimation with Instruments Economics 241B Estimation with Instruments Measurement Error Measurement error is de ned as the error resulting from the measurement of a variable. At some level, every variable is measured with error.

More information

Identification of Models of the Labor Market

Identification of Models of the Labor Market Identification of Models of the Labor Market Eric French and Christopher Taber, Federal Reserve Bank of Chicago and Wisconsin November 6, 2009 French,Taber (FRBC and UW) Identification November 6, 2009

More information

A Guide to Modern Econometric:

A Guide to Modern Econometric: A Guide to Modern Econometric: 4th edition Marno Verbeek Rotterdam School of Management, Erasmus University, Rotterdam B 379887 )WILEY A John Wiley & Sons, Ltd., Publication Contents Preface xiii 1 Introduction

More information

Time Series Models and Inference. James L. Powell Department of Economics University of California, Berkeley

Time Series Models and Inference. James L. Powell Department of Economics University of California, Berkeley Time Series Models and Inference James L. Powell Department of Economics University of California, Berkeley Overview In contrast to the classical linear regression model, in which the components of the

More information

Nonparametric Estimation of Regression Functions In the Presence of Irrelevant Regressors

Nonparametric Estimation of Regression Functions In the Presence of Irrelevant Regressors Nonparametric Estimation of Regression Functions In the Presence of Irrelevant Regressors Peter Hall, Qi Li, Jeff Racine 1 Introduction Nonparametric techniques robust to functional form specification.

More information

Lecture 7 Time-dependent Covariates in Cox Regression

Lecture 7 Time-dependent Covariates in Cox Regression Lecture 7 Time-dependent Covariates in Cox Regression So far, we ve been considering the following Cox PH model: λ(t Z) = λ 0 (t) exp(β Z) = λ 0 (t) exp( β j Z j ) where β j is the parameter for the the

More information

Problem Set 3: Bootstrap, Quantile Regression and MCMC Methods. MIT , Fall Due: Wednesday, 07 November 2007, 5:00 PM

Problem Set 3: Bootstrap, Quantile Regression and MCMC Methods. MIT , Fall Due: Wednesday, 07 November 2007, 5:00 PM Problem Set 3: Bootstrap, Quantile Regression and MCMC Methods MIT 14.385, Fall 2007 Due: Wednesday, 07 November 2007, 5:00 PM 1 Applied Problems Instructions: The page indications given below give you

More information

STAT Sample Problem: General Asymptotic Results

STAT Sample Problem: General Asymptotic Results STAT331 1-Sample Problem: General Asymptotic Results In this unit we will consider the 1-sample problem and prove the consistency and asymptotic normality of the Nelson-Aalen estimator of the cumulative

More information

Identification of the timing-of-events model with multiple competing exit risks from single-spell data

Identification of the timing-of-events model with multiple competing exit risks from single-spell data COHERE - Centre of Health Economics Research Identification of the timing-of-events model with multiple competing exit risks from single-spell data By: Bettina Drepper, Department of Econometrics and OR,

More information

An Overview of Methods for Applying Semi-Markov Processes in Biostatistics.

An Overview of Methods for Applying Semi-Markov Processes in Biostatistics. An Overview of Methods for Applying Semi-Markov Processes in Biostatistics. Charles J. Mode Department of Mathematics and Computer Science Drexel University Philadelphia, PA 19104 Overview of Topics. I.

More information

PoissonprocessandderivationofBellmanequations

PoissonprocessandderivationofBellmanequations APPENDIX B PoissonprocessandderivationofBellmanequations 1 Poisson process Let us first define the exponential distribution Definition B1 A continuous random variable X is said to have an exponential distribution

More information

Structural Econometrics: Dynamic Discrete Choice. Jean-Marc Robin

Structural Econometrics: Dynamic Discrete Choice. Jean-Marc Robin Structural Econometrics: Dynamic Discrete Choice Jean-Marc Robin 1. Dynamic discrete choice models 2. Application: college and career choice Plan 1 Dynamic discrete choice models See for example the presentation

More information

Microeconometrics: Clustering. Ethan Kaplan

Microeconometrics: Clustering. Ethan Kaplan Microeconometrics: Clustering Ethan Kaplan Gauss Markov ssumptions OLS is minimum variance unbiased (MVUE) if Linear Model: Y i = X i + i E ( i jx i ) = V ( i jx i ) = 2 < cov i ; j = Normally distributed

More information

Chapter 4 Fall Notations: t 1 < t 2 < < t D, D unique death times. d j = # deaths at t j = n. Y j = # at risk /alive at t j = n

Chapter 4 Fall Notations: t 1 < t 2 < < t D, D unique death times. d j = # deaths at t j = n. Y j = # at risk /alive at t j = n Bios 323: Applied Survival Analysis Qingxia (Cindy) Chen Chapter 4 Fall 2012 4.2 Estimators of the survival and cumulative hazard functions for RC data Suppose X is a continuous random failure time with

More information

Analysis of competing risks data and simulation of data following predened subdistribution hazards

Analysis of competing risks data and simulation of data following predened subdistribution hazards Analysis of competing risks data and simulation of data following predened subdistribution hazards Bernhard Haller Institut für Medizinische Statistik und Epidemiologie Technische Universität München 27.05.2013

More information

Estimation for Modified Data

Estimation for Modified Data Definition. Estimation for Modified Data 1. Empirical distribution for complete individual data (section 11.) An observation X is truncated from below ( left truncated) at d if when it is at or below d

More information

Notes largely based on Statistical Methods for Reliability Data by W.Q. Meeker and L. A. Escobar, Wiley, 1998 and on their class notes.

Notes largely based on Statistical Methods for Reliability Data by W.Q. Meeker and L. A. Escobar, Wiley, 1998 and on their class notes. Unit 2: Models, Censoring, and Likelihood for Failure-Time Data Notes largely based on Statistical Methods for Reliability Data by W.Q. Meeker and L. A. Escobar, Wiley, 1998 and on their class notes. Ramón

More information

ST745: Survival Analysis: Nonparametric methods

ST745: Survival Analysis: Nonparametric methods ST745: Survival Analysis: Nonparametric methods Eric B. Laber Department of Statistics, North Carolina State University February 5, 2015 The KM estimator is used ubiquitously in medical studies to estimate

More information

Frailty Modeling for clustered survival data: a simulation study

Frailty Modeling for clustered survival data: a simulation study Frailty Modeling for clustered survival data: a simulation study IAA Oslo 2015 Souad ROMDHANE LaREMFiQ - IHEC University of Sousse (Tunisia) souad_romdhane@yahoo.fr Lotfi BELKACEM LaREMFiQ - IHEC University

More information

Economics 620, Lecture 18: Nonlinear Models

Economics 620, Lecture 18: Nonlinear Models Economics 620, Lecture 18: Nonlinear Models Nicholas M. Kiefer Cornell University Professor N. M. Kiefer (Cornell University) Lecture 18: Nonlinear Models 1 / 18 The basic point is that smooth nonlinear

More information

Unobserved Heterogeneity

Unobserved Heterogeneity Unobserved Heterogeneity Germán Rodríguez grodri@princeton.edu Spring, 21. Revised Spring 25 This unit considers survival models with a random effect representing unobserved heterogeneity of frailty, a

More information

Lagged Duration Dependence in Mixed Proportional Hazard Models

Lagged Duration Dependence in Mixed Proportional Hazard Models FACULTEIT ECONOMIE EN BEDRIJFSKUNDE TWEEKERKENSTRAAT 2 B-9000 GENT Tel. : 32 - (0)9 264.34.61 Fax. : 32 - (0)9 264.35.92 WORKING PAPER Lagged Duration Dependence in Mixed Proportional Hazard Models Matteo

More information

New Developments in Econometrics Lecture 16: Quantile Estimation

New Developments in Econometrics Lecture 16: Quantile Estimation New Developments in Econometrics Lecture 16: Quantile Estimation Jeff Wooldridge Cemmap Lectures, UCL, June 2009 1. Review of Means, Medians, and Quantiles 2. Some Useful Asymptotic Results 3. Quantile

More information

A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods. Jeff Wooldridge IRP Lectures, UW Madison, August 2008

A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods. Jeff Wooldridge IRP Lectures, UW Madison, August 2008 A Course in Applied Econometrics Lecture 14: Control Functions and Related Methods Jeff Wooldridge IRP Lectures, UW Madison, August 2008 1. Linear-in-Parameters Models: IV versus Control Functions 2. Correlated

More information

FULL LIKELIHOOD INFERENCES IN THE COX MODEL

FULL LIKELIHOOD INFERENCES IN THE COX MODEL October 20, 2007 FULL LIKELIHOOD INFERENCES IN THE COX MODEL BY JIAN-JIAN REN 1 AND MAI ZHOU 2 University of Central Florida and University of Kentucky Abstract We use the empirical likelihood approach

More information

Typical Survival Data Arising From a Clinical Trial. Censoring. The Survivor Function. Mathematical Definitions Introduction

Typical Survival Data Arising From a Clinical Trial. Censoring. The Survivor Function. Mathematical Definitions Introduction Outline CHL 5225H Advanced Statistical Methods for Clinical Trials: Survival Analysis Prof. Kevin E. Thorpe Defining Survival Data Mathematical Definitions Non-parametric Estimates of Survival Comparing

More information

Longitudinal + Reliability = Joint Modeling

Longitudinal + Reliability = Joint Modeling Longitudinal + Reliability = Joint Modeling Carles Serrat Institute of Statistics and Mathematics Applied to Building CYTED-HAROSA International Workshop November 21-22, 2013 Barcelona Mainly from Rizopoulos,

More information

Chapter 2 - Survival Models

Chapter 2 - Survival Models 2-1 Chapter 2 - Survival Models Section 2.2 - Future Lifetime Random Variable and the Survival Function Let T x = ( Future lifelength beyond age x of an individual who has survived to age x [measured in

More information

Logistic regression model for survival time analysis using time-varying coefficients

Logistic regression model for survival time analysis using time-varying coefficients Logistic regression model for survival time analysis using time-varying coefficients Accepted in American Journal of Mathematical and Management Sciences, 2016 Kenichi SATOH ksatoh@hiroshima-u.ac.jp Research

More information

DAGStat Event History Analysis.

DAGStat Event History Analysis. DAGStat 2016 Event History Analysis Robin.Henderson@ncl.ac.uk 1 / 75 Schedule 9.00 Introduction 10.30 Break 11.00 Regression Models, Frailty and Multivariate Survival 12.30 Lunch 13.30 Time-Variation and

More information

Lecture 3 Stationary Processes and the Ergodic LLN (Reference Section 2.2, Hayashi)

Lecture 3 Stationary Processes and the Ergodic LLN (Reference Section 2.2, Hayashi) Lecture 3 Stationary Processes and the Ergodic LLN (Reference Section 2.2, Hayashi) Our immediate goal is to formulate an LLN and a CLT which can be applied to establish sufficient conditions for the consistency

More information

1 Outline. 1. Motivation. 2. SUR model. 3. Simultaneous equations. 4. Estimation

1 Outline. 1. Motivation. 2. SUR model. 3. Simultaneous equations. 4. Estimation 1 Outline. 1. Motivation 2. SUR model 3. Simultaneous equations 4. Estimation 2 Motivation. In this chapter, we will study simultaneous systems of econometric equations. Systems of simultaneous equations

More information

4 Testing Hypotheses. 4.1 Tests in the regression setting. 4.2 Non-parametric testing of survival between groups

4 Testing Hypotheses. 4.1 Tests in the regression setting. 4.2 Non-parametric testing of survival between groups 4 Testing Hypotheses The next lectures will look at tests, some in an actuarial setting, and in the last subsection we will also consider tests applied to graduation 4 Tests in the regression setting )

More information

A Bivariate Weibull Regression Model

A Bivariate Weibull Regression Model c Heldermann Verlag Economic Quality Control ISSN 0940-5151 Vol 20 (2005), No. 1, 1 A Bivariate Weibull Regression Model David D. Hanagal Abstract: In this paper, we propose a new bivariate Weibull regression

More information

Lecture 4 - Survival Models

Lecture 4 - Survival Models Lecture 4 - Survival Models Survival Models Definition and Hazards Kaplan Meier Proportional Hazards Model Estimation of Survival in R GLM Extensions: Survival Models Survival Models are a common and incredibly

More information

The relationship between treatment parameters within a latent variable framework

The relationship between treatment parameters within a latent variable framework Economics Letters 66 (2000) 33 39 www.elsevier.com/ locate/ econbase The relationship between treatment parameters within a latent variable framework James J. Heckman *,1, Edward J. Vytlacil 2 Department

More information

Longitudinal and Multilevel Methods for Multinomial Logit David K. Guilkey

Longitudinal and Multilevel Methods for Multinomial Logit David K. Guilkey Longitudinal and Multilevel Methods for Multinomial Logit David K. Guilkey Focus of this talk: Unordered categorical dependent variables Models will be logit based Empirical example uses data from the

More information

IDENTIFIABILITY OF THE MULTIVARIATE NORMAL BY THE MAXIMUM AND THE MINIMUM

IDENTIFIABILITY OF THE MULTIVARIATE NORMAL BY THE MAXIMUM AND THE MINIMUM Surveys in Mathematics and its Applications ISSN 842-6298 (electronic), 843-7265 (print) Volume 5 (200), 3 320 IDENTIFIABILITY OF THE MULTIVARIATE NORMAL BY THE MAXIMUM AND THE MINIMUM Arunava Mukherjea

More information

Discrete Time Duration Models with Group level Heterogeneity

Discrete Time Duration Models with Group level Heterogeneity This work is distributed as a Discussion Paper by the STANFORD INSTITUTE FOR ECONOMIC POLICY RESEARCH SIEPR Discussion Paper No. 05-08 Discrete Time Duration Models with Group level Heterogeneity By Anders

More information

GOV 2001/ 1002/ E-2001 Section 10 1 Duration II and Matching

GOV 2001/ 1002/ E-2001 Section 10 1 Duration II and Matching GOV 2001/ 1002/ E-2001 Section 10 1 Duration II and Matching Mayya Komisarchik Harvard University April 13, 2016 1 Heartfelt thanks to all of the Gov 2001 TFs of yesteryear; this section draws heavily

More information