David Prescott

Bootstrapping Confidence Intervals: An Application to Forecasting the Supply of Pork

The article demonstrates how the distribution-free method of bootstrapping can be applied to the construction of confidence intervals for forecasts generated by a dynamic econometric model. Because the exogenous variables must be forecast, the forecasts of the dependent variable are functions of stochastic forecast-period exogenous variables. A dynamic model of pork supply is used to illustrate the procedure.

Travel to Canada: the role of Canada's immigrant populations.

The effect of Canadas immigrant populations on the annual flow of visitors to Canada is investigated. A simple utility-maximizing model of the travel decision motivates the role of immigrant populations in the aggregate demand equation for visits to Canada. The model implies testable hypotheses: price and income elasticities differ by purpose of trip. Using time-series cross-section data on 22 OECD countries an empirical demand model is estimated. Demand is measured by both the number of visitors and person-nights and separate equations are estimated for four subcategories of ‘purpose of trip’. Immigrant populations are found to have a strong influence on the annual flow of foreign visitors. It is estimated that the present value of the stream of spending by foreign visitors attributable to an additional immigrant is approximately

Public/private sector wage differentials in Canada-evidence from the 1991 and 1982 surveys of consumer finance

4550 in 1996 dollars. In accordance with the models predictions, price and income elasticity estimates are greater for vacationers than for those visiting family and friends.

The effects of tax increases on negotiated wage increases in the Canadian private sector

This paper extends the work of Gunderson and Shapiro and Stelcner that considered wage differentials between public and private sector workers in Canada using the 1971 and 1981 Censuses, respectively. Here the Survey of Consumer Finances from 1991 and 1982 is used, which allows for better distinction between public and private sector workers. The estimated public sector premium is decomposed into an endowments component and a residual term. Estimates are corrected for sample selection bias with respect to the choice between full-time and part-time work in the two sectors. Tests for bias caused by self-selection into the private and public sectors revealed no such bias. As a benchmark, the overall wage premium calculated for 1981 compares very well with that of Shapiro and Stelcner. During the 1980s no significant increases in the public sector wage premium for males are found but the female premium increased significantly. This may be due to changes in pay equity legislation in the public sector relative to the private sector.

Hypothesis testing in regression models with AR(1) errors and a lagged dependent variable: Bootstrapping and artificial regression compared

The major objective of this study is to estimate the short-run inflationary ‘supply-side’ effects of increases in sales tax rates and income tax rates on Canadian negotiated wage rates. Empirical results are based on an econometric analysis of 867 COLA wage contracts and 1662 non-COLA wage contracts signed during the 1979 to 1992 period in the Canadian private sector. No statistical evidence is found to show that the substantial increases in the average rate of income tax in Canada during this period led to larger negotiated wage increases. On the other hand, it is found that union workers were able to negotiate wage increases to offset part of the increase in consumer prices attributable to increases in sales tax rates. Given a 1% increase in the CPI directly attributable to sales tax increases, non-COLA contracts provided a 0.25% wage increase and COLA contracts provided a 0.75% wage increase. The 6.4% increase in the Canadian Consumer Price Index directly attributable to sales tax increases during the ...

Labour Supply Estimates for Low-Income Female Heads of Household Using Mincome Data

Abstract Serially correlated errors in dynamic models render the standard conditional estimator of the covariance matrix inconsistent. A Monte Carlo experiment confirms that the downward bias in the conventional variance estimator also exists in small samples. The results favour a consistent estimator based on an artificial regression (suggested by Davidson and Mackinnon) over bootstrapping the distribution of parameter estimates.