Lecture 1 Review I. RS Lecture 1. CLM - Assumptions. Least Squares Estimation f.o.c. Least Squares Estimation - Assumptions

Size: px
Start display at page:

Download "Lecture 1 Review I. RS Lecture 1. CLM - Assumptions. Least Squares Estimation f.o.c. Least Squares Estimation - Assumptions"

Transcription

1 RS Lecture CLM - Assuptions Lecture Review I Typicl Assuptions (A) DGP: y = Xβ + ε is correctly specified. (A) E[ε X] = (A3) Vr[ε X] = σ I T (A4) X hs full colun rnk rnk(x)=k-, where T k. Assuption (A) is clled correct specifiction. We know how the DGP. Assuption (A) is clled regression. Fro (A) we get: (i) E[ε X] = => E[y X] = f(x, θ) + E[ε X] = f(x, θ) (ii) Using the Lw of Iterted Expecttions (LIE): E[ε] = E X [E[ε X]] = E X [] = Lest Squres Estition - Assuptions Fro Assuption (A3) we get Vr[ε X] = σ I T => Vr[ε] = σ I T This ssuption iplies (i) hooscedsticity => E[ε i X] = σ for ll i. (ii) no seril/cross correltion => E[ ε i ε j X] = for i j. Fro Assuption (A4) => the k independent vribles in X re linerly independent. Then, the kxk trix X X will lso hve full rnk i.e., rnk(x X) = k. Lest Squres Estition f.o.c. Objective function: S(x i, θ) =Σ i ε i We wnt to iniize w.r.t to θ. The f.o.c. deliver the norl equtions: - Σ i [y i - f(x i, θ LS )] f (x i, θ LS ) = - (y- Xb) X = Solving for b delivers the OLS estitor: b = (X X) - X y Note: (i) b = β OLS. (Ordinry LS. Ordinry=liner) (ii) b is (liner) function of the dt (y i,x i ). (iii) X (y-xb) = X y - X X(X X) - X y = X e = => e X. OLS Estition - Properties Under the typicl ssuptions, we cn estblish properties for b. ) E[b X]= β ) Vr[b X] = E[(b-β) (b-β) X] =(X X) - X E[ε ε X] X(X X) - = σ (X X) - 3) b is BLUE (or MVLUE) => The Guss-Mrkov theore. (4) If (A5) ε X ~N(, σ I T ) => b X ~N(β, σ (X X) - ) => b k X ~N(β k, σ (X X) - kk ) (the rginls of ultivrite norl re lso norl.) Estiting σ Under (A5), E[e e X] = (T-k)σ The unbised estitor of σ is s = e e/(t-k). => there is degrees of freedo correction. Goodness of Fit of the Regression After estiting the odel, we judge the dequcy of the odel. There re two wys to do this: - Visul: plots of fitted vlues nd residuls, histogrs of residuls. - Nuericl esures: R, djusted R, AIC, BIC, etc. Nuericl esures. We cll the goodness-of-fit esures. Most populr: R. R = SSR/TSS = b X M Xb/y M y = - e e/y M y Note: R is bounded by zero nd one only if: () There is constnt ter in X --we need e M X=! (b) The line is coputed by liner lest squres. (c) R never flls when regressors re dded to the regression.

2 RS Lecture Adjusted R-squred R is odified with penlty for nuber of preters: Adjusted R R = - [(T-)/(T-k)]( - R ) = - [(T-)/(T-k)] RSS/TSS = - [RSS/(T-k)] [(T-)/TSS] => xiizing djusted R <=> iniizing [RSS/(T-k)]= s Degrees of freedo --i.e., (T-k)-- djustent ssues soething bout unbisedness. Adjusted-R includes penlty for vribles tht do not dd uch fit. Cn fll when vrible is dded to the eqution. It will rise when vrible, sy z, is dded to the regression if nd only if the t-rtio on z is lrger thn one in bsolute vlue. Other Goodness of Fit Mesures There re other goodness-of-fit esures tht lso incorporte penlties for nuber of preters (degrees of freedo). Infortion Criteri - Aeiy: [e e/(t K)] ( + k/t) - Akike Infortion Criterion (AIC) AIC = -/T(ln L k) L: Likelihood => if norlity AIC = ln(e e/t) + (/T) k (+constnts) - Byes-Schwrz Infortion Criterion (BIC) BIC = -(/T ln L [ln(t)/t] k) => if norlity AIC = ln(e e/t) + [ln(t)/t] k (+constnts) Mxiu Likelihood Estition We ssue the errors, ε, follow distribution. Then, we select the preters of the distribution to xiize the likelihood of the observed sple. Exple: The errors, ε, follow the norl distribution: (A5) ε X ~N(, σ I T ) Then, we cn write the joint pdf of y s / f ( yt ) = ( ) exp[ ( y ' ) ] t xt β πσ σ T / L = f ( y, y,..., y T β, σ ) =Π t = ( ) exp[ ( y ' ) ] exp( ) t xt β = e e T / πσ σ (πσ ) σ Tking logs, we hve the log likelihood function T T ln L = ln π ln σ e e σ Mxiu Likelihood Estition Let θ =(β,σ). Then, we wnt to T T Mxθ ln L( θ y, X ) = ln π σ ( y Xβ) ( y Xβ) σ Then, the f.o.c.: ln L = ( X y X Xβ) = ( X y X Xβ) = β σ σ ln L T = + ( y Xβ) ( y Xβ) = 4 σ σ σ Note: The f.o.c. deliver the norl equtions for β! The solution to the norl eqution, β MLE, is lso the LS estitor, b. Tht is, βˆ MLE = b = ( X X ) X y; σˆ e e = T Nice result for b: ML estitors hve very good properties! MLE Properties of ML Estitors () Efficiency. Under generl conditions, we hve tht θ Vr( θ MLE ) [ ni ( θ)] The right-hnd side is the Crer-Ro lower bound (CR-LB). If n estitor cn chieve this bound, ML will produce it. () Consistency. S n (X; θ) nd ( θˆmle ^ ^ MLE - θ) converge together to zero (i.e., expecttion). Properties of ML Estitors (4) Sufficiency. If single sufficient sttistic exists for θ, the MLE of θ ust be function of it. Tht is, θˆmle depends on the sple observtions only through the vlue of sufficient sttistic. (5) Invrince. The ML estite is invrint under functionl trnsfortions. Tht is, if θˆmle is the MLE of θ nd if g(θ) is function of θ, then g( ) is the MLE of g(θ). θˆmle (3) Theore: Asyptotic Norlity Let the likelihood function be L(X,X, X n θ). Under generl conditions, the MLE of θ is syptoticlly distributed s ˆ θ MLE N ( θ, [ ni ( θ )] )

3 RS Lecture Specifiction Errors: Oitted Vribles Oitting relevnt vribles: Suppose the correct odel is y = X β + X β + ε -i.e., with two sets of vribles. But, we copute OLS oitting X. Tht is, y = X β + ε Soe esily proved results: <= the short regression. () E[b X] = E [(X X ) - X y] = β + (X X ) - X X β β. => Unless X X =, b is bised. The bis cn be huge. () Vr[b X] Vr[b. X] => sller vrince when we oit X. (3) MSE => b y be ore precise. Specifiction Errors: Irrelevnt Vribles Irrelevnt vribles Suppose the correct odel is y = X β + ε But, we estite y = X β + X β + ε Let s copute OLS with X, X. This is clled long regression. Soe esily proved results: () Since the vribles in X re truly irrelevnt, then β =, so E[b. X] = β => No bis () Inefficiency: Bigger vrince Liner Restrictions Q: How do liner restrictions ffect the properties of the lest squres estitor? Model ( DGP): y = Xβ + ε Theory (infortion): Rβ - q = Restricted LS estitor: b* = b - (X X) - R [R(X X) - R ] - (Rb - q). Unbised? YES. E[b* X] = β. Efficiency? NO. Vr[b* X] < Vr[b X] 3. b* y be ore precise. Precision = MSE = vrince + squred bis. 4. Recll: e e = (y -Xb) (y-xb) e* e* = (y Xb*) (y-xb*) => Restrictions cnnot increse R => R R * The Generl Liner Hypothesis: H : Rβ - q = We hve J joint hypotheses. Let R be Jxk trix nd q be Jx vector. Two pproches to testing (unifying point: OLS is unbised): () Is Rb - q close to? Bsing the test on the discrepncy vector: = Rb - q. Using the Wld sttistic: W = (Vr[ X]) - Vr[ X] = R[σ (X X) - ]R. W = (Rb q) {R[σ (X X) - ]R} - (Rb q) Under the usul ssuption nd ssuing σ is known, W ~ χ J In generl, σ is unknown, we use s = e e/(t-k) W* = (Rb - q) {R[s (X X) - ]R} - (Rb - q) = (Rb q) {R[σ (X X) - ]R} - (Rb q)/(s /σ ) F = W/J / [(T-k) (s /σ )/(T-k)] = W*/J ~ F J,T-k. The Generl Liner Hypothesis: H : Rβ - q = () We know tht iposing the restrictions leds to loss of fit. R ust go down. Does it go down lot? -i.e., significntly? Recll (i) e* = y Xb* = e X(b* b) (ii) b*= b (X X) - R [R(X X) - R ] - (Rb q) => e* e* - e e = (Rb q) [R(X X) - R ] - (Rb q) Recll - W = (Rb q) {R[σ (X X) - ]R} - (Rb q) ~ χ J (if σ known) - e e/σ ~ χ T-k. Then, F = (e* e* e e)/j / [e e/(t-k)] ~ F J,T-K. Or F = { (R - R* )/J } / [( - R )/(T-k)] ~ F J,T-K. Exple: Testing H : Rβ - q = In the liner odel y = X β + ε = β + X β + X 3 β 3 + X 4 β 4 + ε We wnt to test if the slopes X 3, X 4 re equl to zero. Tht is, H : β = β = H : β or β 4 or both β 3 nd β 4 Y Y We cn use, F = (e* e* e e)/j / [e e/(t-k)] ~ F J,T-K. Define β + β X + ε = = β + β X + β X + β X + ε RSS R RSS U F (cost in df, unconstr df ) RSS = R-RSS U k U -k R RSS U T-k U 3

4 RS Lecture Functionl For: Chow Test Assuption (A) restricts f(x,β) to be liner function: f(x,β) = X β. But, within the frework of OLS estition, we cn be ore flexible: () We cn ipose non-liner functionl fors, s long s they re liner in the preters (intrinsic liner odel). () We cn use qulittive vribles (duies) to crete non-linerities (splines, chnges in regie, etc.) A Chow test (n F-test) cn be used to check for regies/ctegories or structurl breks. () Run OLS with no distinction between regies. Keep RSS R. (b) Run two seprte OLS, one for ech regie (Unrestricted regression). Keep RSS nd RSS => RSS U = RSS + RSS. (3) Run stndrd F-test (testing Restricted vs. Unrestricted odels): ( RSSR RSSU ) /( k F = ( RSS ) /( T k U U U k ) R ) ( RSSR [ RSS + RSS ]) / k = ( RSS + RSS ) /( T k) 3 Functionl For: Rsey s RESET Test To test the specifiction of the functionl for, we cn use the RESET test. Fro regression, we keep the fitted vlues, ŷ = Xb. Then, we dd ŷ to the regression specifiction. If ŷ is dded to the regression specifiction, it should pick up qudrtic nd interctive nonlinerity: y = X β + ŷ γ + ε We test H (liner functionl for): γ= H ( non liner functionl for): γ => t-test on the OLS estitor of γ. If the t-sttistic for ŷ is significnt => evidence of nonlinerity. 3 Prediction Intervls Prediction: Given x => predict y. () Estite: E[y X, x ] = β x ; () Prediction: y = β x + ε Predictor: ŷ = b x + estite of ε. (Est. ε =, but with vrince) Forecst error. We predict y with ŷ = b x. ŷ - y = b x - β x - ε = (b - β) x - ε => Vr[(ŷ -y ) x ] = E[(ŷ -y ) (ŷ -y ) x ]= x Vr[(b - β) x ]x + σ How do we estite this? Two cses: () If x is vector of constnts => For C.I. s usul. () If x hs to be estited => Coplicted (wht is the vrince of the product?). Use bootstrpping. Forecst Vrince Vrince of the forecst error is σ + x Vr[b x ]x = σ + σ [x (X X) - x ] If the odel contins constnt ter, this is K K jk Vr[ e ] = σ + + ( x j x j )( xk xk )( M ) n j= k= (where is X without x =ί). In ters squres nd cross products of devitions fro ens. Note: Lrge σ, sll n, nd lrge devitions fro the ens, decrese the precision forecsting error. Interprettion: Forecst vrince is sllest in the iddle of our experience nd increses s we ove outside it. Forecsting perfornce of odel: Tests nd esures of perfornce Evlution of odel s predictive ccurcy for individul (insple nd out-of-sple) observtions Evlution of odel s predictive ccurcy for group of (insple nd out-of-sple) observtions Chow prediction test Evlution of forecsts Sury esures of out-of-sple forecst ccurcy Men Error = Men Absolute Error (MAE) = Men Squred Error (MSE) = Root Men Squre Error (RMSE)= ( yˆ i yi ) = ei yˆ i yi = ei ( yˆ i yi ) = ei ( yˆ i yi ) = ei Theil s U-stt = U = ei T yi T

5 RS Lecture CLM: Asyptotics To get exct results for OLS, we rely on (A5) ε X ~iid N(, σ I T ) But, (A5) in ny situtions is unrelistic. Then, we study on the behvior of b (nd the test sttistics) when T i.e., lrge sples. New ssuptions: () {x i,ε i },,..., T is sequence of independent observtions. - X is stochstic, but independent of the process generting ε. - We require tht X hve finite ens nd vrinces. Siilr requireent for ε, but we lso require E[ε]=. () Well behved X: pli (X X/T) = Q (Q pd trix of finite eleents) => (not too uch dependence in X). CLM: New Assuptions Now, we hve new set of ssuptions in the CLM: (A) DGP: y = X β + ε. (A ) X stochstic, but E[X ε]= nd E[ε]=. (A3) Vr[ε X] = σ I T (A4 ) pli (X X/T) = Q (p.d. trix with finite eleents, rnk= k) We wnt to study the lrge sple properties of OLS: Q : Is b consistent? s? YES & YES Q : Wht is the distribution of b? b N(β,(σ /T)Q - ) Q 3: Wht bout the distribution of the tests? => t T =[(z T - µ)/s T ] d N(,) => W = (z T - µ) S - T (z T - µ) d χ rnk(st) => F d χ rnk(vr[]) Asyptotic Tests: Sll sple behvior? The p-vlues fro syptotic tests re pproxite for sll sples. They y be very bd. Their perfornce depends on: () Sple size, T. () Distribution of the error ters, ε. (3) The nuber of regressors, k, nd their properties (4) The reltionship between the error ters nd the regressors. A siultion/bootstrp cn help. Bootstrp tests tend to perfor better thn tests bsed on pproxite syptotic distributions. The errors coitted by both syptotic nd bootstrp tests diinish s T increses. The Delt Method It is used to obtin the syptotic distribution of non-liner function of RV (usully, n estitor). Tools: () A first-order Tylor series expnsion () Slutsky s theore. Let x n be RV, with pli x n =θ nd Vr(x n )=σ <. We use the CLT to obtin n ½ (x n - θ)/σ d N(,) Gol: g(x n )? (g(x n ) is continuous differentible function, independent of n.) Steps: () Tylor series pproxition round θ : g(x n ) = g(θ) + g (θ) (x n - θ) + higher order ters We ssue the higher order ters re o(n) --s n grows, they vnish. The Delt Method () Use Slutsky theore: pli g(x n ) = g(θ) pli g (x n ) = g (θ) Then, s n grows, g(x n ) g(θ) + g (θ) (x n - θ) => n ½ ([g(x n ) - g(θ)]) g (θ) [n ½ (x n - θ)]. => n ½ ([g(x n ) - g(θ)]/σ) g (θ) [n ½ (x n - θ)/σ]. The syptotic distribution of (g(x n ) - g(θ)) is given by tht of [n ½ (x n - θ)/σ], which is stndrd norl. Then, n ½ ([g(x n ) - g(θ)]) N(, [g (θ)] σ ). After soe work ( inversion ), we obtin: g(x n ) N(g(θ), [g (θ)] σ /n). Delt Method: Exple Let x n N(θ, σ /n) Q: g(x n )=δ/x n? (δ is constnt) First, clculte the first two oents of g(x n ): g(x n ) = δ/x n => pli g(x n )=(δ/θ) g (x n ) = -(δ/x n ) => pli g (x n )=-(δ/θ ) Recll delt ethod forul: g(x n ) Then, g(x n ) N(δ/θ, (δ /θ 4 )σ /n) N(g(θ), [g (θ)] σ /n).

6 RS Lecture The IV Proble Wht kes b consistent when X'ε /T p is tht pproxiting (X'ε/T ) by is resonbly ccurte in lrge sples. Now, we chllenge the ssuption tht {x i,ε i } is sequence of independent observtions. Now, we ssue pli (X ε/t) => This is the IV Proble! Q: When ight X be correlted ε? - Correlted shocks cross linked equtions - Siultneous equtions - Errors in vribles - Model hs lgged dependent vrible nd serilly correlted error ter The IV Proble We strt with our liner odel y = Xβ + ε. Now, we ssue pli(x ε/t). pli (X X/T) = Q Then, pli b = pli β + pli (X X/T) - pli (X ε ε/t) = β + Q - pli (X ε ε/t) β => b is not consistent estitor of β. New ssuption: we hve l instruentl vribles, such tht pli( X/T) but pli( ε/t) = Instruentl Vribles: Assuptions To get consistent estitor of β, we lso ssue: {x i, z i, ε i } is sequence of RVs, with: E[X X] = Q xx (pd nd finite) (LLN => pli(x X/T) =Q xx ) E[ ] = Q zz (finite) (LLN => pli( /T) =Q zz ) E[ X] = Q zx (pd nd finite) (LLN => pli( X/T) =Q zx ) E[ ε] = (LLN => pli( ε/t) = ) Following the se ide s in OLS, we get syste of equtions: W' X b IV = W' y Instruentl Vribles: Estition To get the IV estitor, we strt fro the syste of equtions: W' X b IV = W' y Cse : l = k -i.e., nuber of instruents = nuber of regressors. - hs the se diensions s X: Txk => X is kxk trix - In this cse, W is irrelevnt, sy, W=I. - Then, b IV = ( X) - y We hve two cses where estition is possible: - Cse : l = k -i.e., nuber of instruents = nuber of regressors. - Cse : l > k -i.e., nuber of instruents > nuber of regressors. IV Estitors Properties of b IV () Consistent b IV = ( X) - y = ( X) - (Xβ+ε) = ( X/T) - ( X/T) β + ( X/T) - ε/t = β + ( X/T) - ε/t p β (under ssuptions) () Asyptotic norlity T (b IV - β) = T ( X) - ε = ( X/T) - T ( ε/t) Using the Lindberg-Feller CLT T ( ε/t) d N(, σ Q zz ) Then, T (b IV - β) d N(, σ Q - zx Q zz Q - xz ) IV Estitors Properties of σˆ, under IV estition: - We define σˆ : T T ˆ = e IV ( y i x' b IV ) T = T i = i = σ where e IV = y - X b IV = y - X( X) - y = [I - X( X) - ]y = M zx y - Then, σˆ = e IV 'e IV /T = ε'm zx 'M zx ε/t = ε'ε/t ε'x ( X) - ε/t + ε' ('X) - X X( X) - ε/t => pli σˆ = pli(ε'ε/t) - pli[(ε'x/t) ( X/T) - ('ε/t)] + + pli(ε' ( X) - X X( X) - ε/t) = σ Est Asy. Vr[b IV ] = E[('X) - εε εε' ( X) - ]= ˆσ ( X) - '( X) -

7 RS Lecture IV Estitors: SLS (-Stge Lest Squres) Cse : l > k -i.e., nuber of instruents > nuber of regressors. - This is the usul cse. We cn throw l-k instruents, but throwing wy infortion is never optil. - The IV norl equtions re n l x k syste of equtions: y = Xβ+ ε Note: We cnnot pproxite ll the ε by siultenously. There will be t lest l-k non-zero residuls. (Siilr setup to regression!) IV Estitors: SLS (-Stge Lest Squres) We cn esily derive properties for b IV : biv = ( X ' P X ) X ' P y = ( X ' P P X ) Xˆ Xˆ ( ' ) X ˆ ' y ( Xˆ ' X ˆ = = ) Xˆ ' yˆ X ' P P y () b IV is consistent () b IV is syptoticlly norl. - This is estitor is lso clled GIVE (Generlized IV estitor) - Fro the IV norl equtions => W' X b IV = W' y - We define different IV estitor - Let W = ( ) - X = P X = Xˆ - Then, X'P X b IV = X'P y b X P X X P y X P P X X P P y X ˆ X ˆ = ( ' ) ' = ( ' ) ' = ( ' ) X ˆ ' yˆ IV Interprettions of b IV b b IV IV = b X ˆ X ˆ SLS = ( ' ) ( Xˆ = ' X ) X ˆ ' y X ˆ ' y This is the SLS interprettion This is the usul IV = Xˆ Asyptotic Efficiency The vrince is lrger thn tht of LS. (A lrge sple type of Guss-Mrkov result is t work.) () OLS is inconsistent. () Men squred error is uncertin: MSE[estitor β] = Vrince + squre of bis. IV y be better or worse. Depends on the dt: X nd ε. Probles with SLS X/T y not be sufficiently lrge. The covrince trix for the IV estitor is Asy. Cov(b) = σ [( X)( ) - (X )] - If X/T goes to (wek instruents), the vrince explodes. When there re ny instruents, Xˆ is too close to X; SLS becoes OLS. Populr isconception: If only one vrible in X is correlted with ε, the other coefficients re consistently estited. Flse. => The proble is sered over the other coefficients. Wht re the finite sple properties of b IV? We do not hve the condition E[ε X] =, we cnnot conclude tht b IV is unbised, or tht it hs Vr[b SLS ] equl to its syptotic covrince trix. => In fct, b SLS cn hve very bd sll-sple properties. Endogeneity Test (Husn) Exogenous Endogenous OLS Consistent, Efficient Inconsistent SLS Consistent, Inefficient Consistent Bse test on d = b SLS - b OLS - We cn use Wld sttistic: d [Vr(d)] - d Note: Under H (pli (X ε/t) = ) b OLS = b SLS = b Also, under H : Vr[b SLS ]= V SLS > Vr[b OLS ]= V OLS => Under H, one estitor is efficient, the other one is not. Q: Wht to use for Vr(d)? - Husn (978): V = Vr(d) = V SLS - V OLS H = (b SLS - b OLS ) [V SLS - V OLS ] - (b SLS - b OLS ) d χ rnk(v) Endogeneity Test: The Wu Test The Husn test is coplicted to clculte Siplifiction: The Wu test. Consider regression y = Xβ + ε, n rry of proper instruents, nd n rry of instruents W tht includes plus other vribles tht y be either clen or continted. Wu test for H : X is clen. Setup () Regress X on. Keep fitted vlues Xˆ = ( ) - X () Using W s instruents, do SLS regression of y on X, keep RSS. (3) Do SLS regression of y on X nd subset of coluns of Xˆ tht re linerly independent of X. Keep RSS. (4) Do n F-test: F = [(RSS - RSS )/]/[RSS /(T-k)].

8 RS Lecture Endogeneity Test: The Wu Test Under H : X is clen, the F sttistic hs n pproxite F,T-k distribution. Dvidson nd McKinnon (993, 39) point out tht the DWH test relly tests whether possible endogeneity of the right-hnd-side vribles not contined in the instruents kes ny difference to the coefficient estites. These types of exogeneity tests re usully known s DWH (Durbin, Wu, Husn) tests. Endogeneity Test: Augented DWH Test Dvidson nd McKinnon (993) suggest n ugented regression test (DWH test), by including the residuls of ech endogenous righthnd side vrible. Model: y = Xβ+ Uγ + ε, we suspect X is endogenous. Steps for ugented regression DWH test:. Regress x on IV () nd U: x = П + U φ + υ => sve residuls v x. Do n ugented regression: y = Xβ + Uγ + v x δ + ε 3. Do t-test of δ. If the estite of δ, sy d, is significntly different fro zero, then OLS is not consistent. Mesureent Error DGP: y* = βx* + ε - ε ~ iid D(, σ ε ) But, we do not observe or esure correctly x*. We observe x, y: x = x* + u u ~ iid D(, σ u ) -no correltion to ε,v y = y* + v v ~ iid D(, σ v ) -no correltion to ε,u Let s consider two cses: Mesureent Error CASE - Only y* is esured with error. y* = y - v = βx* + ε => y = βx* + ε + v = βx* + (ε + v) Q: Wht hppens when y is regressed on x? A: Nothing! We hve our usul OLS proble since ε nd v re independent of ech other nd x *. CLM ssuptions re not violted! CASE - Only x* is esured with error (y=y*): y = β(x- u) + ε = βx + ε - βu = βx + w E[x w] = E[(x* + u) (ε - βu)] = -βσ u => CLM ssuptions violted => OLS inconsistent! Finding n Instruent: Not Esy The IV proble requires dt on vribles () such tht () Cov(x,) -relevnce condition () Cov(,ε) = -vlid (exogeneity) condition Wek Instruents: Finnce ppliction Finnce exple: The consuption CAPM. In both liner nd nonliner versions of the odel, IVs re wek, -- see Neeley, Roy, nd Whiten (), nd Yogo (4). Then, we do first-stge regression to obtin fitted vlues of X: x = П + Uδ + V -V ~N(, σ V I) Then, using the fitted vlues we estite nd do tests on β. Finding tht eets both requireents is not esy. - The vlid condition is not tht coplicted to eet. - The relevnt condition is ore coplicted: Finding correlted with X. But, the explntory power of y not be enough to llow inference on β. In this cse, we sy is wek instruent. In the liner odel in Yogo (4): X (endogenous vrible): consuption growth (the IVs): twice lgged noinl interest rtes, infltion, consuption growth, nd log dividend-price rtio. But, log consuption is close to rndo wlk, consuption growth is difficult to predict. This leds to the IVs being wek. => Yogo (4) finds F-sttistics for H : П = in the st stge regression tht lie between.7 nd 3.53 for different countries.

9 RS Lecture Wek Instruents: Sury Even if the instruent is good i.e., it eets the relevnt condition--, tters cn be de fr worse with IV s opposed to OLS ( the cure cn be worse... ). Wek correltion between IV nd endogenous regressor cn pose severe finite-sple bis. Even sll Cov(,e) will cuse inconsistency, nd this will be excerbted when Cov(X,) is sll. Lrge T will not help. A&K nd Consuption CAPM tests hve very lrge sples! Wek Instruents: Detection nd Reedies Sypto: The relevnce condition, pli( X/T ) not zero, is close to being violted. Detection of wek IV: Stndrd F test in the st stge regression of x k on. Stiger nd Stock (997) suggest tht F < is sign of probles. Low prtil-r X,. Lrge Vr[b IV ] s well s potentilly severe finite-sple bis. Reedy: Not uch ost of the discussion is bout the condition, not wht to do bout it. Use LIML? Requires norlity ssuption. Probbly, not too restrictive. (Text, ) Wek Instruents: Detection nd Reedies Sypto: The vlid condition, pli( ε/t ) zero, is close to being violted. Detection of instruent exogeneity: Endogenous IV s: Inconsistency of b IV tht kes it no better (nd probbly worse) thn b OLS Durbin-Wu-Husn test: Endogeneity of the proble regressor(s) Reedy: Avoid endogeneous wek instruents. (Also void wek IV!) Generl proble: It is not esy to find good instruents in theory nd in prctice. Find nturl experients. M-Estition An extreu estitor is one obtined s the optiizer of criterion function, q(z,b). Exples: OLS: b = rg x (-e e/t) MLE: b MLE = rg x ln L =,,T ln f (y i,x i,b) GMM: b GMM = rg x - g(y i,x i,b) W g(y i,x i,b) There re two clsses of extreu estitors: - M-estitors: The objective function is sple verge or su. - Miniu distnce estitors: The objective function is esure of distnce. "M" stnds for xiu or iniu estitors --Huber (967). M-Estition The objective function is sple verge or su. For exple, we wnt to iniize popultion (first) oent: in b E[q(z,β)] Using the LLN, we ove fro the popultion first oent to the sple verge: i q(z i,b)/t p E[q(z,β)] We wnt to obtin: b = rgin i q(z i,b) (or divided by T) In generl, we solve the f.o.c. (or zero-score condition): ero-score: i q(z i,b)/ b = M-Estition If s(z,b) = q(z,b)/ b exists (lost everywhere), we solve i s(z i,b M )/T = (*) If, in ddition, E X [s(z,b)] = / b E X [q(z,b)] -i.e., differentition nd integrtion re exchngeble-, then E X [ q(z,β)/ β ] =. Under these ssuptions, the M-estitor is sid to be of ψ-type (ψ= s(z,b)=score). Often, b M is tken to be the solution of (*) without checking whether it is indeed iniu). To check the s.o.c., we define the (pd) Hessin: H = i q(z i,b)/ b b Otherwise, the M-estitor is of ρ-type. (ρ= q(z,β)).

10 RS Lecture M-Estition: LS & ML Lest Squres DGP: y = f(x,β) + ε, z =[y,x] q(z;β) = S(β) = ε ε =,,T (y i - f(x i ;β)) Now, we ove fro popultion to sple oents q(z;b) = S(b) = e e =,,T (y i - f(x i ;b)) b NLLS = rgin S(b) Mxiu Likelihood Let f (x i,β) be the pdf of the dt. L(x,β) = Π,,T f (x i ;β) log L(x,β) =,,T ln f (x i ;β) Now, we ove fro popultion to sple oents q(z,b) = -log L(x,b) b MLE = rgin log L(x;b) M-Estitors: Properties Under generl ssuptions, M-estitors re: - b p M b - b M N(b,Vr[b ]) - Vr[b M ] =(/T) H - V H - - If the odel is correctly specified: -H = V. Then, Vr[b] = V H nd V re evluted t b : - H = i [ q (z i,b)/ b b ] - V = i [ q(z i,b)/ b][ q(z i,b)/ b ] Nonliner Lest Squres: Exple Exple: Min β S(β) ={½ Σ i [y i - f(xβ)] } Fro the f.o.c., we cnnot solve for β explicitly. But, using soe steps, we cn still iniize RSS to obtin estites of β. Nonliner regression lgorith:. Strt by guessing plusible vlues for β, sy β.. Clculte RSS for β => get RSS(β ) 3. Mke sll chnges to β, => get β. 4. Clculte RSS for β => get RSS(β ) 5. If RSS(β ) < RSS(β ) => β becoes your new strting point. 6. Repet steps 3-5 until you RSS(β j ) cnnot be lowered. => get β j. => β j is the (nonliner) lest squres estites. 4 NLLS: Lineriztion We strt with nonliner odel: y i = f(x i,β) + ε i We expnd the regression round soe point, β : f(x i,β) f(x i,β ) + Σ k [ f(x i,β )/ β k ]( β k - β k ) = f(x i,β ) + Σ k x i ( β k - β k ) = [f(x i,β ) - Σ k x i β k ] + Σ k x i β k = f + Σ k x i β k = f + x i β where f i = f(x i,β ) - x i β (f i does not depend on unknowns) Now, f(x i,β) is (pproxitely) liner in the preters. Tht is, y i = f i + x i β + ε i (ε i = ε i + lineriztion error i) => y i = y i f i = x i β + ε i NLLS: Lineriztion We linerized f(x i,β) to get: y = f + X β + ε (ε = ε + lineriztion error) => y = y - f = X β + ε Now, we cn do OLS: b NLLS = (X X ) - X y Note: X re clled pseudo-regressors. NLLS: Lineriztion Copute the syptotic covrince trix for the NLLS estitor s usul: Est. Vr[b NLLS X ] = s NLLS (X X ) - s NLLS = [y - f(x i, b NLLS )] [y - f(x i, b NLLS )]/(T-k). Since the results re syptotic, we do not need degrees of freedo correction. However, df correction is usully included. In generl, we get different b NLLS for different β. An lgorith cn be used to get the best b NLLS. We will resort to nuericl optiiztion to find the b NLLS.

11 RS Lecture Guss-Newton Algorith b NLLS depends on β. Tht is, b NLLS (β ) = (X X ) - X y We use Guss-Newton lgorith to find the b NLLS. Recll GN: β k+ = β k + (J T J) - J T ε -- J: Jcobin = δf(xi;β)/δβ. Given b NLLS t step, b(j), we find the b NLLS for step j+ by: b(j+) = b(j) + [X (j) X (j)] - X (j) e (j) Coluns of X (j) re the derivtives: f(x i,b(j))/ b(j) e (j) = y - f[x,b(j)] The updte vector is the slopes in the regression of the residuls on X. The updte is zero when they re orthogonl. (Just like OLS) Box-Cox Trnsfortion A siple trnsfortion tht llows non-linerities in the CLM. y = f(x i,β) + ε = Σ k x k (λ) β k + ε x k (λ) = (x kλ -)/λ li λ (x kλ -)/λ = ln x k For given λ, OLS cn be used. An itertive process cn be used to estite λ. OLS s.e. hve to be corrected. Not very efficient ethod. NLLS or MLE will work fine. We cn hve ore generl Box-Cox trnsfortion odel: y (λ) = Σ k x k (λ) β k + ε Testing non-liner restrictions Testing liner restrictions s before. Non-liner restrictions chnge he usul tests. We wnt to test: H : R(β) = where R(β) is non-liner function, with rnk[ R(β)/ β=g(β)]=j. Let = R(b NLLS ). Then, W= (Vr[ X]) - = R(b NLLS ) (Vr[R(b NLLS ) X]) - R(b NLLS ) But, we do not know the distribution of R(b NLLS ). We know the distribution of b NLLS. Then, we linerize R(b NLLS ) round β: R(b NLLS ) R(β) + G(b NLLS ) (b NLLS - β) Testing non-liner restrictions Linerize R(b NLLS ) round β (=b ) R(b NLLS ) R(β) + G(b NLLS ) (b NLLS - β) Recll T (b M - b ) d N(, Vr[b ]) where Vr[b ] = H(β) - V (β) H(β) - => T [R(b NLLS ) - R(β)] N(, G(β) Vr[b ] G(β) d ) => Vr[R(b NLLS )] = (/T) G(β) Vr[b ] G(β) Then, W = T R(b NLLS ) {G(b NLLS ) Vr[b NLLS ] G(b NLLS ) } - R(b NLLS ) => W d χ J

ECONOMETRIC THEORY. MODULE IV Lecture - 16 Predictions in Linear Regression Model

ECONOMETRIC THEORY. MODULE IV Lecture - 16 Predictions in Linear Regression Model ECONOMETRIC THEORY MODULE IV Lecture - 16 Predictions in Liner Regression Model Dr. Shlbh Deprtent of Mthetics nd Sttistics Indin Institute of Technology Knpur Prediction of vlues of study vrible An iportnt

More information

Lecture 6: Singular Integrals, Open Quadrature rules, and Gauss Quadrature

Lecture 6: Singular Integrals, Open Quadrature rules, and Gauss Quadrature Lecture notes on Vritionl nd Approximte Methods in Applied Mthemtics - A Peirce UBC Lecture 6: Singulr Integrls, Open Qudrture rules, nd Guss Qudrture (Compiled 6 August 7) In this lecture we discuss the

More information

OXFORD H i g h e r E d u c a t i o n Oxford University Press, All rights reserved.

OXFORD H i g h e r E d u c a t i o n Oxford University Press, All rights reserved. Renshw: Mths for Econoics nswers to dditionl exercises Exercise.. Given: nd B 5 Find: () + B + B 7 8 (b) (c) (d) (e) B B B + B T B (where 8 B 6 B 6 8 B + B T denotes the trnspose of ) T 8 B 5 (f) (g) B

More information

P 3 (x) = f(0) + f (0)x + f (0) 2. x 2 + f (0) . In the problem set, you are asked to show, in general, the n th order term is a n = f (n) (0)

P 3 (x) = f(0) + f (0)x + f (0) 2. x 2 + f (0) . In the problem set, you are asked to show, in general, the n th order term is a n = f (n) (0) 1 Tylor polynomils In Section 3.5, we discussed how to pproximte function f(x) round point in terms of its first derivtive f (x) evluted t, tht is using the liner pproximtion f() + f ()(x ). We clled this

More information

Improper Integrals, and Differential Equations

Improper Integrals, and Differential Equations Improper Integrls, nd Differentil Equtions October 22, 204 5.3 Improper Integrls Previously, we discussed how integrls correspond to res. More specificlly, we sid tht for function f(x), the region creted

More information

CS667 Lecture 6: Monte Carlo Integration 02/10/05

CS667 Lecture 6: Monte Carlo Integration 02/10/05 CS667 Lecture 6: Monte Crlo Integrtion 02/10/05 Venkt Krishnrj Lecturer: Steve Mrschner 1 Ide The min ide of Monte Crlo Integrtion is tht we cn estimte the vlue of n integrl by looking t lrge number of

More information

Non-Linear & Logistic Regression

Non-Linear & Logistic Regression Non-Liner & Logistic Regression If the sttistics re boring, then you've got the wrong numbers. Edwrd R. Tufte (Sttistics Professor, Yle University) Regression Anlyses When do we use these? PART 1: find

More information

Exam 2, Mathematics 4701, Section ETY6 6:05 pm 7:40 pm, March 31, 2016, IH-1105 Instructor: Attila Máté 1

Exam 2, Mathematics 4701, Section ETY6 6:05 pm 7:40 pm, March 31, 2016, IH-1105 Instructor: Attila Máté 1 Exm, Mthemtics 471, Section ETY6 6:5 pm 7:4 pm, Mrch 1, 16, IH-115 Instructor: Attil Máté 1 17 copies 1. ) Stte the usul sufficient condition for the fixed-point itertion to converge when solving the eqution

More information

PHYS 601 HW3 Solution

PHYS 601 HW3 Solution 3.1 Norl force using Lgrnge ultiplier Using the center of the hoop s origin, we will describe the position of the prticle with conventionl polr coordintes. The Lgrngin is therefore L = 1 2 ṙ2 + 1 2 r2

More information

CMDA 4604: Intermediate Topics in Mathematical Modeling Lecture 19: Interpolation and Quadrature

CMDA 4604: Intermediate Topics in Mathematical Modeling Lecture 19: Interpolation and Quadrature CMDA 4604: Intermedite Topics in Mthemticl Modeling Lecture 19: Interpoltion nd Qudrture In this lecture we mke brief diversion into the res of interpoltion nd qudrture. Given function f C[, b], we sy

More information

NUMERICAL INTEGRATION. The inverse process to differentiation in calculus is integration. Mathematically, integration is represented by.

NUMERICAL INTEGRATION. The inverse process to differentiation in calculus is integration. Mathematically, integration is represented by. NUMERICAL INTEGRATION 1 Introduction The inverse process to differentition in clculus is integrtion. Mthemticlly, integrtion is represented by f(x) dx which stnds for the integrl of the function f(x) with

More information

SUMMER KNOWHOW STUDY AND LEARNING CENTRE

SUMMER KNOWHOW STUDY AND LEARNING CENTRE SUMMER KNOWHOW STUDY AND LEARNING CENTRE Indices & Logrithms 2 Contents Indices.2 Frctionl Indices.4 Logrithms 6 Exponentil equtions. Simplifying Surds 13 Opertions on Surds..16 Scientific Nottion..18

More information

Review of Calculus, cont d

Review of Calculus, cont d Jim Lmbers MAT 460 Fll Semester 2009-10 Lecture 3 Notes These notes correspond to Section 1.1 in the text. Review of Clculus, cont d Riemnn Sums nd the Definite Integrl There re mny cses in which some

More information

The Regulated and Riemann Integrals

The Regulated and Riemann Integrals Chpter 1 The Regulted nd Riemnn Integrls 1.1 Introduction We will consider severl different pproches to defining the definite integrl f(x) dx of function f(x). These definitions will ll ssign the sme vlue

More information

Continuous Random Variables

Continuous Random Variables STAT/MATH 395 A - PROBABILITY II UW Winter Qurter 217 Néhémy Lim Continuous Rndom Vribles Nottion. The indictor function of set S is rel-vlued function defined by : { 1 if x S 1 S (x) if x S Suppose tht

More information

PHY 5246: Theoretical Dynamics, Fall Assignment # 5, Solutions. θ = l mr 2 = l

PHY 5246: Theoretical Dynamics, Fall Assignment # 5, Solutions. θ = l mr 2 = l PHY 546: Theoreticl Dynics, Fll 15 Assignent # 5, Solutions 1 Grded Probles Proble 1 (1.) Using the eqution of the orbit or force lw d ( 1 dθ r)+ 1 r = r F(r), (1) l with r(θ) = ke αθ one finds fro which

More information

UNIQUENESS THEOREMS FOR ORDINARY DIFFERENTIAL EQUATIONS WITH HÖLDER CONTINUITY

UNIQUENESS THEOREMS FOR ORDINARY DIFFERENTIAL EQUATIONS WITH HÖLDER CONTINUITY UNIQUENESS THEOREMS FOR ORDINARY DIFFERENTIAL EQUATIONS WITH HÖLDER CONTINUITY YIFEI PAN, MEI WANG, AND YU YAN ABSTRACT We estblish soe uniqueness results ner 0 for ordinry differentil equtions of the

More information

Construction of Gauss Quadrature Rules

Construction of Gauss Quadrature Rules Jim Lmbers MAT 772 Fll Semester 2010-11 Lecture 15 Notes These notes correspond to Sections 10.2 nd 10.3 in the text. Construction of Guss Qudrture Rules Previously, we lerned tht Newton-Cotes qudrture

More information

Review of basic calculus

Review of basic calculus Review of bsic clculus This brief review reclls some of the most importnt concepts, definitions, nd theorems from bsic clculus. It is not intended to tech bsic clculus from scrtch. If ny of the items below

More information

f(a+h) f(a) x a h 0. This is the rate at which

f(a+h) f(a) x a h 0. This is the rate at which M408S Concept Inventory smple nswers These questions re open-ended, nd re intended to cover the min topics tht we lerned in M408S. These re not crnk-out-n-nswer problems! (There re plenty of those in the

More information

Chapters 4 & 5 Integrals & Applications

Chapters 4 & 5 Integrals & Applications Contents Chpters 4 & 5 Integrls & Applictions Motivtion to Chpters 4 & 5 2 Chpter 4 3 Ares nd Distnces 3. VIDEO - Ares Under Functions............................................ 3.2 VIDEO - Applictions

More information

Data Assimilation. Alan O Neill Data Assimilation Research Centre University of Reading

Data Assimilation. Alan O Neill Data Assimilation Research Centre University of Reading Dt Assimiltion Aln O Neill Dt Assimiltion Reserch Centre University of Reding Contents Motivtion Univrite sclr dt ssimiltion Multivrite vector dt ssimiltion Optiml Interpoltion BLUE 3d-Vritionl Method

More information

Discrete Mathematics and Probability Theory Spring 2013 Anant Sahai Lecture 17

Discrete Mathematics and Probability Theory Spring 2013 Anant Sahai Lecture 17 EECS 70 Discrete Mthemtics nd Proility Theory Spring 2013 Annt Shi Lecture 17 I.I.D. Rndom Vriles Estimting the is of coin Question: We wnt to estimte the proportion p of Democrts in the US popultion,

More information

Properties of Integrals, Indefinite Integrals. Goals: Definition of the Definite Integral Integral Calculations using Antiderivatives

Properties of Integrals, Indefinite Integrals. Goals: Definition of the Definite Integral Integral Calculations using Antiderivatives Block #6: Properties of Integrls, Indefinite Integrls Gols: Definition of the Definite Integrl Integrl Clcultions using Antiderivtives Properties of Integrls The Indefinite Integrl 1 Riemnn Sums - 1 Riemnn

More information

UNIFORM CONVERGENCE. Contents 1. Uniform Convergence 1 2. Properties of uniform convergence 3

UNIFORM CONVERGENCE. Contents 1. Uniform Convergence 1 2. Properties of uniform convergence 3 UNIFORM CONVERGENCE Contents 1. Uniform Convergence 1 2. Properties of uniform convergence 3 Suppose f n : Ω R or f n : Ω C is sequence of rel or complex functions, nd f n f s n in some sense. Furthermore,

More information

THE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS.

THE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS. THE EXISTENCE-UNIQUENESS THEOREM FOR FIRST-ORDER DIFFERENTIAL EQUATIONS RADON ROSBOROUGH https://intuitiveexplntionscom/picrd-lindelof-theorem/ This document is proof of the existence-uniqueness theorem

More information

g i fφdx dx = x i i=1 is a Hilbert space. We shall, henceforth, abuse notation and write g i f(x) = f

g i fφdx dx = x i i=1 is a Hilbert space. We shall, henceforth, abuse notation and write g i f(x) = f 1. Appliction of functionl nlysis to PEs 1.1. Introduction. In this section we give little introduction to prtil differentil equtions. In prticulr we consider the problem u(x) = f(x) x, u(x) = x (1) where

More information

r = cos θ + 1. dt ) dt. (1)

r = cos θ + 1. dt ) dt. (1) MTHE 7 Proble Set 5 Solutions (A Crdioid). Let C be the closed curve in R whose polr coordintes (r, θ) stisfy () Sketch the curve C. r = cos θ +. (b) Find pretriztion t (r(t), θ(t)), t [, b], of C in polr

More information

Goals: Determine how to calculate the area described by a function. Define the definite integral. Explore the relationship between the definite

Goals: Determine how to calculate the area described by a function. Define the definite integral. Explore the relationship between the definite Unit #8 : The Integrl Gols: Determine how to clculte the re described by function. Define the definite integrl. Eplore the reltionship between the definite integrl nd re. Eplore wys to estimte the definite

More information

Lecture 1. Functional series. Pointwise and uniform convergence.

Lecture 1. Functional series. Pointwise and uniform convergence. 1 Introduction. Lecture 1. Functionl series. Pointwise nd uniform convergence. In this course we study mongst other things Fourier series. The Fourier series for periodic function f(x) with period 2π is

More information

Math 1B, lecture 4: Error bounds for numerical methods

Math 1B, lecture 4: Error bounds for numerical methods Mth B, lecture 4: Error bounds for numericl methods Nthn Pflueger 4 September 0 Introduction The five numericl methods descried in the previous lecture ll operte by the sme principle: they pproximte the

More information

Chapter 3 Solving Nonlinear Equations

Chapter 3 Solving Nonlinear Equations Chpter 3 Solving Nonliner Equtions 3.1 Introduction The nonliner function of unknown vrible x is in the form of where n could be non-integer. Root is the numericl vlue of x tht stisfies f ( x) 0. Grphiclly,

More information

Lecture 3 Gaussian Probability Distribution

Lecture 3 Gaussian Probability Distribution Introduction Lecture 3 Gussin Probbility Distribution Gussin probbility distribution is perhps the most used distribution in ll of science. lso clled bell shped curve or norml distribution Unlike the binomil

More information

Recitation 3: More Applications of the Derivative

Recitation 3: More Applications of the Derivative Mth 1c TA: Pdric Brtlett Recittion 3: More Applictions of the Derivtive Week 3 Cltech 2012 1 Rndom Question Question 1 A grph consists of the following: A set V of vertices. A set E of edges where ech

More information

f(x) dx, If one of these two conditions is not met, we call the integral improper. Our usual definition for the value for the definite integral

f(x) dx, If one of these two conditions is not met, we call the integral improper. Our usual definition for the value for the definite integral Improper Integrls Every time tht we hve evluted definite integrl such s f(x) dx, we hve mde two implicit ssumptions bout the integrl:. The intervl [, b] is finite, nd. f(x) is continuous on [, b]. If one

More information

The Atwood Machine OBJECTIVE INTRODUCTION APPARATUS THEORY

The Atwood Machine OBJECTIVE INTRODUCTION APPARATUS THEORY The Atwood Mchine OBJECTIVE To derive the ening of Newton's second lw of otion s it pplies to the Atwood chine. To explin how ss iblnce cn led to the ccelertion of the syste. To deterine the ccelertion

More information

AP Calculus Multiple Choice: BC Edition Solutions

AP Calculus Multiple Choice: BC Edition Solutions AP Clculus Multiple Choice: BC Edition Solutions J. Slon Mrch 8, 04 ) 0 dx ( x) is A) B) C) D) E) Divergent This function inside the integrl hs verticl symptotes t x =, nd the integrl bounds contin this

More information

Best Approximation. Chapter The General Case

Best Approximation. Chapter The General Case Chpter 4 Best Approximtion 4.1 The Generl Cse In the previous chpter, we hve seen how n interpolting polynomil cn be used s n pproximtion to given function. We now wnt to find the best pproximtion to given

More information

1B40 Practical Skills

1B40 Practical Skills B40 Prcticl Skills Comining uncertinties from severl quntities error propgtion We usully encounter situtions where the result of n experiment is given in terms of two (or more) quntities. We then need

More information

Lecture 21: Order statistics

Lecture 21: Order statistics Lecture : Order sttistics Suppose we hve N mesurements of sclr, x i =, N Tke ll mesurements nd sort them into scending order x x x 3 x N Define the mesured running integrl S N (x) = 0 for x < x = i/n for

More information

Exponents and Powers

Exponents and Powers EXPONENTS AND POWERS 9 Exponents nd Powers CHAPTER. Introduction Do you know? Mss of erth is 5,970,000,000,000, 000, 000, 000, 000 kg. We hve lredy lernt in erlier clss how to write such lrge nubers ore

More information

Lecture 14: Quadrature

Lecture 14: Quadrature Lecture 14: Qudrture This lecture is concerned with the evlution of integrls fx)dx 1) over finite intervl [, b] The integrnd fx) is ssumed to be rel-vlues nd smooth The pproximtion of n integrl by numericl

More information

Numerical Analysis: Trapezoidal and Simpson s Rule

Numerical Analysis: Trapezoidal and Simpson s Rule nd Simpson s Mthemticl question we re interested in numericlly nswering How to we evlute I = f (x) dx? Clculus tells us tht if F(x) is the ntiderivtive of function f (x) on the intervl [, b], then I =

More information

The Wave Equation I. MA 436 Kurt Bryan

The Wave Equation I. MA 436 Kurt Bryan 1 Introduction The Wve Eqution I MA 436 Kurt Bryn Consider string stretching long the x xis, of indeterminte (or even infinite!) length. We wnt to derive n eqution which models the motion of the string

More information

Second degree generalized gauss-seidel iteration method for solving linear system of equations. ABSTRACT

Second degree generalized gauss-seidel iteration method for solving linear system of equations. ABSTRACT Ethiop. J. Sci. & Technol. 7( 5-, 0 5 Second degree generlized guss-seidel itertion ethod for solving liner syste of equtions Tesfye Keede Bhir Dr University, College of Science, Deprtent of Mthetics tk_ke@yhoo.co

More information

Math 360: A primitive integral and elementary functions

Math 360: A primitive integral and elementary functions Mth 360: A primitive integrl nd elementry functions D. DeTurck University of Pennsylvni October 16, 2017 D. DeTurck Mth 360 001 2017C: Integrl/functions 1 / 32 Setup for the integrl prtitions Definition:

More information

AQA Further Pure 1. Complex Numbers. Section 1: Introduction to Complex Numbers. The number system

AQA Further Pure 1. Complex Numbers. Section 1: Introduction to Complex Numbers. The number system Complex Numbers Section 1: Introduction to Complex Numbers Notes nd Exmples These notes contin subsections on The number system Adding nd subtrcting complex numbers Multiplying complex numbers Complex

More information

Quantum Physics II (8.05) Fall 2013 Assignment 2

Quantum Physics II (8.05) Fall 2013 Assignment 2 Quntum Physics II (8.05) Fll 2013 Assignment 2 Msschusetts Institute of Technology Physics Deprtment Due Fridy September 20, 2013 September 13, 2013 3:00 pm Suggested Reding Continued from lst week: 1.

More information

Definition of Continuity: The function f(x) is continuous at x = a if f(a) exists and lim

Definition of Continuity: The function f(x) is continuous at x = a if f(a) exists and lim Mth 9 Course Summry/Study Guide Fll, 2005 [1] Limits Definition of Limit: We sy tht L is the limit of f(x) s x pproches if f(x) gets closer nd closer to L s x gets closer nd closer to. We write lim f(x)

More information

Predict Global Earth Temperature using Linier Regression

Predict Global Earth Temperature using Linier Regression Predict Globl Erth Temperture using Linier Regression Edwin Swndi Sijbt (23516012) Progrm Studi Mgister Informtik Sekolh Teknik Elektro dn Informtik ITB Jl. Gnesh 10 Bndung 40132, Indonesi 23516012@std.stei.itb.c.id

More information

New Expansion and Infinite Series

New Expansion and Infinite Series Interntionl Mthemticl Forum, Vol. 9, 204, no. 22, 06-073 HIKARI Ltd, www.m-hikri.com http://dx.doi.org/0.2988/imf.204.4502 New Expnsion nd Infinite Series Diyun Zhng College of Computer Nnjing University

More information

A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H. Thomas Shores Department of Mathematics University of Nebraska Spring 2007

A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H. Thomas Shores Department of Mathematics University of Nebraska Spring 2007 A REVIEW OF CALCULUS CONCEPTS FOR JDEP 384H Thoms Shores Deprtment of Mthemtics University of Nebrsk Spring 2007 Contents Rtes of Chnge nd Derivtives 1 Dierentils 4 Are nd Integrls 5 Multivrite Clculus

More information

We partition C into n small arcs by forming a partition of [a, b] by picking s i as follows: a = s 0 < s 1 < < s n = b.

We partition C into n small arcs by forming a partition of [a, b] by picking s i as follows: a = s 0 < s 1 < < s n = b. Mth 255 - Vector lculus II Notes 4.2 Pth nd Line Integrls We begin with discussion of pth integrls (the book clls them sclr line integrls). We will do this for function of two vribles, but these ides cn

More information

Monte Carlo method in solving numerical integration and differential equation

Monte Carlo method in solving numerical integration and differential equation Monte Crlo method in solving numericl integrtion nd differentil eqution Ye Jin Chemistry Deprtment Duke University yj66@duke.edu Abstrct: Monte Crlo method is commonly used in rel physics problem. The

More information

Math Lecture 23

Math Lecture 23 Mth 8 - Lecture 3 Dyln Zwick Fll 3 In our lst lecture we delt with solutions to the system: x = Ax where A is n n n mtrix with n distinct eigenvlues. As promised, tody we will del with the question of

More information

= a. P( X µ X a) σ2 X a 2. = σ2 X

= a. P( X µ X a) σ2 X a 2. = σ2 X Co902 problem sheet, solutions. () (Mrkov inequlity) LetX be continuous, non-negtive rndom vrible (RV) ndpositive constnt. Show tht: P(X ) E[X] Solution: Let p(x) represent the pdf of RVX. Then: E[X] =

More information

Numerical integration

Numerical integration 2 Numericl integrtion This is pge i Printer: Opque this 2. Introduction Numericl integrtion is problem tht is prt of mny problems in the economics nd econometrics literture. The orgniztion of this chpter

More information

Integrals - Motivation

Integrals - Motivation Integrls - Motivtion When we looked t function s rte of chnge If f(x) is liner, the nswer is esy slope If f(x) is non-liner, we hd to work hrd limits derivtive A relted question is the re under f(x) (but

More information

3.4 Numerical integration

3.4 Numerical integration 3.4. Numericl integrtion 63 3.4 Numericl integrtion In mny economic pplictions it is necessry to compute the definite integrl of relvlued function f with respect to "weight" function w over n intervl [,

More information

Solution for Assignment 1 : Intro to Probability and Statistics, PAC learning

Solution for Assignment 1 : Intro to Probability and Statistics, PAC learning Solution for Assignment 1 : Intro to Probbility nd Sttistics, PAC lerning 10-701/15-781: Mchine Lerning (Fll 004) Due: Sept. 30th 004, Thursdy, Strt of clss Question 1. Bsic Probbility ( 18 pts) 1.1 (

More information

ODE: Existence and Uniqueness of a Solution

ODE: Existence and Uniqueness of a Solution Mth 22 Fll 213 Jerry Kzdn ODE: Existence nd Uniqueness of Solution The Fundmentl Theorem of Clculus tells us how to solve the ordinry differentil eqution (ODE) du = f(t) dt with initil condition u() =

More information

20 MATHEMATICS POLYNOMIALS

20 MATHEMATICS POLYNOMIALS 0 MATHEMATICS POLYNOMIALS.1 Introduction In Clss IX, you hve studied polynomils in one vrible nd their degrees. Recll tht if p(x) is polynomil in x, the highest power of x in p(x) is clled the degree of

More information

PHYS 705: Classical Mechanics. Small Oscillations: Example A Linear Triatomic Molecule

PHYS 705: Classical Mechanics. Small Oscillations: Example A Linear Triatomic Molecule PHYS 75: Clssicl echnics Sll Oscilltions: Exple A Liner Tritoic olecule A Liner Tritoic olecule x b b x x3 x Experientlly, one ight be interested in the rdition resulted fro the intrinsic oscilltion odes

More information

and that at t = 0 the object is at position 5. Find the position of the object at t = 2.

and that at t = 0 the object is at position 5. Find the position of the object at t = 2. 7.2 The Fundmentl Theorem of Clculus 49 re mny, mny problems tht pper much different on the surfce but tht turn out to be the sme s these problems, in the sense tht when we try to pproimte solutions we

More information

Mapping the delta function and other Radon measures

Mapping the delta function and other Radon measures Mpping the delt function nd other Rdon mesures Notes for Mth583A, Fll 2008 November 25, 2008 Rdon mesures Consider continuous function f on the rel line with sclr vlues. It is sid to hve bounded support

More information

1 Probability Density Functions

1 Probability Density Functions Lis Yn CS 9 Continuous Distributions Lecture Notes #9 July 6, 28 Bsed on chpter by Chris Piech So fr, ll rndom vribles we hve seen hve been discrete. In ll the cses we hve seen in CS 9, this ment tht our

More information

Best Approximation in the 2-norm

Best Approximation in the 2-norm Jim Lmbers MAT 77 Fll Semester 1-11 Lecture 1 Notes These notes correspond to Sections 9. nd 9.3 in the text. Best Approximtion in the -norm Suppose tht we wish to obtin function f n (x) tht is liner combintion

More information

W. We shall do so one by one, starting with I 1, and we shall do it greedily, trying

W. We shall do so one by one, starting with I 1, and we shall do it greedily, trying Vitli covers 1 Definition. A Vitli cover of set E R is set V of closed intervls with positive length so tht, for every δ > 0 nd every x E, there is some I V with λ(i ) < δ nd x I. 2 Lemm (Vitli covering)

More information

Discrete Mathematics and Probability Theory Summer 2014 James Cook Note 17

Discrete Mathematics and Probability Theory Summer 2014 James Cook Note 17 CS 70 Discrete Mthemtics nd Proility Theory Summer 2014 Jmes Cook Note 17 I.I.D. Rndom Vriles Estimting the is of coin Question: We wnt to estimte the proportion p of Democrts in the US popultion, y tking

More information

Math 8 Winter 2015 Applications of Integration

Math 8 Winter 2015 Applications of Integration Mth 8 Winter 205 Applictions of Integrtion Here re few importnt pplictions of integrtion. The pplictions you my see on n exm in this course include only the Net Chnge Theorem (which is relly just the Fundmentl

More information

Improper Integrals. Type I Improper Integrals How do we evaluate an integral such as

Improper Integrals. Type I Improper Integrals How do we evaluate an integral such as Improper Integrls Two different types of integrls cn qulify s improper. The first type of improper integrl (which we will refer to s Type I) involves evluting n integrl over n infinite region. In the grph

More information

Advanced Calculus: MATH 410 Notes on Integrals and Integrability Professor David Levermore 17 October 2004

Advanced Calculus: MATH 410 Notes on Integrals and Integrability Professor David Levermore 17 October 2004 Advnced Clculus: MATH 410 Notes on Integrls nd Integrbility Professor Dvid Levermore 17 October 2004 1. Definite Integrls In this section we revisit the definite integrl tht you were introduced to when

More information

Chapter 3 Polynomials

Chapter 3 Polynomials Dr M DRAIEF As described in the introduction of Chpter 1, pplictions of solving liner equtions rise in number of different settings In prticulr, we will in this chpter focus on the problem of modelling

More information

Physics 116C Solution of inhomogeneous ordinary differential equations using Green s functions

Physics 116C Solution of inhomogeneous ordinary differential equations using Green s functions Physics 6C Solution of inhomogeneous ordinry differentil equtions using Green s functions Peter Young November 5, 29 Homogeneous Equtions We hve studied, especilly in long HW problem, second order liner

More information

1. Gauss-Jacobi quadrature and Legendre polynomials. p(t)w(t)dt, p {p(x 0 ),...p(x n )} p(t)w(t)dt = w k p(x k ),

1. Gauss-Jacobi quadrature and Legendre polynomials. p(t)w(t)dt, p {p(x 0 ),...p(x n )} p(t)w(t)dt = w k p(x k ), 1. Guss-Jcobi qudrture nd Legendre polynomils Simpson s rule for evluting n integrl f(t)dt gives the correct nswer with error of bout O(n 4 ) (with constnt tht depends on f, in prticulr, it depends on

More information

5.7 Improper Integrals

5.7 Improper Integrals 458 pplictions of definite integrls 5.7 Improper Integrls In Section 5.4, we computed the work required to lift pylod of mss m from the surfce of moon of mss nd rdius R to height H bove the surfce of the

More information

Chapter 5 : Continuous Random Variables

Chapter 5 : Continuous Random Variables STAT/MATH 395 A - PROBABILITY II UW Winter Qurter 216 Néhémy Lim Chpter 5 : Continuous Rndom Vribles Nottions. N {, 1, 2,...}, set of nturl numbers (i.e. ll nonnegtive integers); N {1, 2,...}, set of ll

More information

ARITHMETIC OPERATIONS. The real numbers have the following properties: a b c ab ac

ARITHMETIC OPERATIONS. The real numbers have the following properties: a b c ab ac REVIEW OF ALGEBRA Here we review the bsic rules nd procedures of lgebr tht you need to know in order to be successful in clculus. ARITHMETIC OPERATIONS The rel numbers hve the following properties: b b

More information

1 Part II: Numerical Integration

1 Part II: Numerical Integration Mth 4 Lb 1 Prt II: Numericl Integrtion This section includes severl techniques for getting pproimte numericl vlues for definite integrls without using ntiderivtives. Mthemticll, ect nswers re preferble

More information

CAAM 453 NUMERICAL ANALYSIS I Examination There are four questions, plus a bonus. Do not look at them until you begin the exam.

CAAM 453 NUMERICAL ANALYSIS I Examination There are four questions, plus a bonus. Do not look at them until you begin the exam. Exmintion 1 Posted 23 October 2002. Due no lter thn 5pm on Mondy, 28 October 2002. Instructions: 1. Time limit: 3 uninterrupted hours. 2. There re four questions, plus bonus. Do not look t them until you

More information

Discrete Least-squares Approximations

Discrete Least-squares Approximations Discrete Lest-squres Approximtions Given set of dt points (x, y ), (x, y ),, (x m, y m ), norml nd useful prctice in mny pplictions in sttistics, engineering nd other pplied sciences is to construct curve

More information

Definite integral. Mathematics FRDIS MENDELU. Simona Fišnarová (Mendel University) Definite integral MENDELU 1 / 30

Definite integral. Mathematics FRDIS MENDELU. Simona Fišnarová (Mendel University) Definite integral MENDELU 1 / 30 Definite integrl Mthemtics FRDIS MENDELU Simon Fišnrová (Mendel University) Definite integrl MENDELU / Motivtion - re under curve Suppose, for simplicity, tht y = f(x) is nonnegtive nd continuous function

More information

Partial Derivatives. Limits. For a single variable function f (x), the limit lim

Partial Derivatives. Limits. For a single variable function f (x), the limit lim Limits Prtil Derivtives For single vrible function f (x), the limit lim x f (x) exists only if the right-hnd side limit equls to the left-hnd side limit, i.e., lim f (x) = lim f (x). x x + For two vribles

More information

221B Lecture Notes WKB Method

221B Lecture Notes WKB Method Clssicl Limit B Lecture Notes WKB Method Hmilton Jcobi Eqution We strt from the Schrödinger eqution for single prticle in potentil i h t ψ x, t = [ ] h m + V x ψ x, t. We cn rewrite this eqution by using

More information

Y i = β 0 + β 1 X i + β 2 Z i + ε i

Y i = β 0 + β 1 X i + β 2 Z i + ε i Homework Exercise #7 Economics 4340 Fll 009 Dr. W. Ken Frr Nme Instructions: Answer the questions given below on lined sheets of pper (or they my be typed) nd provide ll the necessry printouts of informtion

More information

Lecture 20: Numerical Integration III

Lecture 20: Numerical Integration III cs4: introduction to numericl nlysis /8/0 Lecture 0: Numericl Integrtion III Instructor: Professor Amos Ron Scribes: Mrk Cowlishw, Yunpeng Li, Nthnel Fillmore For the lst few lectures we hve discussed

More information

Math& 152 Section Integration by Parts

Math& 152 Section Integration by Parts Mth& 5 Section 7. - Integrtion by Prts Integrtion by prts is rule tht trnsforms the integrl of the product of two functions into other (idelly simpler) integrls. Recll from Clculus I tht given two differentible

More information

Definite integral. Mathematics FRDIS MENDELU

Definite integral. Mathematics FRDIS MENDELU Definite integrl Mthemtics FRDIS MENDELU Simon Fišnrová Brno 1 Motivtion - re under curve Suppose, for simplicity, tht y = f(x) is nonnegtive nd continuous function defined on [, b]. Wht is the re of the

More information

Abstract inner product spaces

Abstract inner product spaces WEEK 4 Abstrct inner product spces Definition An inner product spce is vector spce V over the rel field R equipped with rule for multiplying vectors, such tht the product of two vectors is sclr, nd the

More information

Overview of Calculus I

Overview of Calculus I Overview of Clculus I Prof. Jim Swift Northern Arizon University There re three key concepts in clculus: The limit, the derivtive, nd the integrl. You need to understnd the definitions of these three things,

More information

Bernoulli Numbers Jeff Morton

Bernoulli Numbers Jeff Morton Bernoulli Numbers Jeff Morton. We re interested in the opertor e t k d k t k, which is to sy k tk. Applying this to some function f E to get e t f d k k tk d k f f + d k k tk dk f, we note tht since f

More information

8.3 THE TRIGONOMETRIC FUNCTIONS. skipped 8.4 THE ALGEBRAIC COMPLETENESS OF THE COMPLEX FIELD. skipped 8.5 FOURIER SERIES

8.3 THE TRIGONOMETRIC FUNCTIONS. skipped 8.4 THE ALGEBRAIC COMPLETENESS OF THE COMPLEX FIELD. skipped 8.5 FOURIER SERIES 8.5 FOURIER SERIES 0 8.3 THE TRIGONOMETRIC FUNCTIONS skipped 8.4 THE ALGEBRAIC COMPLETENESS OF THE COMPLEX FIELD skipped 8.5 FOURIER SERIES 8.9 Orthogonl Functions, Orthonorl: Let { n }, n, 2, 3,...,besequence

More information

Section 17.2 Line Integrals

Section 17.2 Line Integrals Section 7. Line Integrls Integrting Vector Fields nd Functions long urve In this section we consider the problem of integrting functions, both sclr nd vector (vector fields) long curve in the plne. We

More information

Theoretical foundations of Gaussian quadrature

Theoretical foundations of Gaussian quadrature Theoreticl foundtions of Gussin qudrture 1 Inner product vector spce Definition 1. A vector spce (or liner spce) is set V = {u, v, w,...} in which the following two opertions re defined: (A) Addition of

More information

UNIT 1 FUNCTIONS AND THEIR INVERSES Lesson 1.4: Logarithmic Functions as Inverses Instruction

UNIT 1 FUNCTIONS AND THEIR INVERSES Lesson 1.4: Logarithmic Functions as Inverses Instruction Lesson : Logrithmic Functions s Inverses Prerequisite Skills This lesson requires the use of the following skills: determining the dependent nd independent vribles in n exponentil function bsed on dt from

More information

Math 113 Exam 2 Practice

Math 113 Exam 2 Practice Mth Em Prctice Februry, 8 Em will cover sections 6.5, 7.-7.5 nd 7.8. This sheet hs three sections. The first section will remind you bout techniques nd formuls tht you should know. The second gives number

More information

Orthogonal Polynomials and Least-Squares Approximations to Functions

Orthogonal Polynomials and Least-Squares Approximations to Functions Chpter Orthogonl Polynomils nd Lest-Squres Approximtions to Functions **4/5/3 ET. Discrete Lest-Squres Approximtions Given set of dt points (x,y ), (x,y ),..., (x m,y m ), norml nd useful prctice in mny

More information

MIXED MODELS (Sections ) I) In the unrestricted model, interactions are treated as in the random effects model:

MIXED MODELS (Sections ) I) In the unrestricted model, interactions are treated as in the random effects model: 1 2 MIXED MODELS (Sections 17.7 17.8) Exmple: Suppose tht in the fiber breking strength exmple, the four mchines used were the only ones of interest, but the interest ws over wide rnge of opertors, nd

More information

Polynomial Approximations for the Natural Logarithm and Arctangent Functions. Math 230

Polynomial Approximations for the Natural Logarithm and Arctangent Functions. Math 230 Polynomil Approimtions for the Nturl Logrithm nd Arctngent Functions Mth 23 You recll from first semester clculus how one cn use the derivtive to find n eqution for the tngent line to function t given

More information

. Double-angle formulas. Your answer should involve trig functions of θ, and not of 2θ. sin 2 (θ) =

. Double-angle formulas. Your answer should involve trig functions of θ, and not of 2θ. sin 2 (θ) = Review of some needed Trig. Identities for Integrtion. Your nswers should be n ngle in RADIANS. rccos( 1 ) = π rccos( - 1 ) = 2π 2 3 2 3 rcsin( 1 ) = π rcsin( - 1 ) = -π 2 6 2 6 Cn you do similr problems?

More information