Juha M. Alho

Overview

Juha M. Alho is a Finnish statistician and demographer (University of Joensuu, later University of Helsinki) who has been one of the leading methodologists for stochastic and probabilistic population forecasting since the late 1980s. His central contribution is the naïve/base-line error approach: using the historical accuracy of simple naïve forecasts as a conservative upper bound on the ex ante error of more sophisticated official forecasts, thereby providing a model-free standard against which to evaluate official demographic projections. He has applied this framework to US Census Bureau fertility forecasts (with Bruce Spencer), US Social Security Administration mortality forecasts (Alho and Spencer 1990), and IIASA/UN world population forecasts.

Key Contributions / Features

• Alho and Spencer (1985) — "Uncertain Population Forecasting," JASA 80(390): 306–314; introduces approximately linear propagation-of-error model $V(t+1)=R(t)V(t)$; three error sources; mixed estimation of data and expert priors; establishes USBC high-low intervals too narrow for 67% coverage; fertility as dominant uncertainty source. See Alho and Spencer 1985 — Uncertain Population Forecasting.
• Alho (1990) — "Stochastic Methods in Population Forecasting," Int. J. Forecast. 6: 521–530; foundational volatility-based error assessment framework; naïve forecast as conservative upper bound; establishes $\sigma_e \approx 0.08$ for log TFR in industrialized countries.
• Alho and Spencer (1990) — "Error Models for Official Mortality Forecasts," JASA 85: 609–616; parallel error-bounding methodology applied to SSA mortality forecasts; establishes that SSA mortality intervals systematically understate uncertainty.
• Alho (1997) — "Scenarios, Uncertainty, and Conditional Forecasts of the World Population," JRSS-A 160(1): 71–85; shows IIASA 1994 high-low for world population in 2030 is ~85% PI; shows UN 1993 forecast implies implausible demographic policy success (b = 0.39–0.55); introduces conditional forecast framework for interpreting scenario differences. See Stochastic Fertility Forecasting.
• Alho and Spencer (1991, 1996) — "A population forecast as a database" and "The practical specification of the expected error of population forecasts"; extensions of the volatility-based approach to full cohort-component propagation of error.

Related

• Stochastic Fertility Forecasting
• SSA Mortality Forecasting
• Lee-Carter Model
• Demographic Transition
• Bruce D. Spencer

Sources

• Alho and Spencer 1985 — Uncertain Population Forecasting
• Alho 1997 — Scenarios Uncertainty and Conditional Forecasts of the World Population