FIN454 Population Forecasting

Embed Size (px)

Citation preview

  • 8/2/2019 FIN454 Population Forecasting

    1/33

    1

    Population ForecastingPopulation Forecasting

    Time Series Forecasting TechniquesTime Series Forecasting Techniques

    Wayne Foss, MBA, MAI

    Wayne Foss Appraisals, Inc.Email: [email protected]

  • 8/2/2019 FIN454 Population Forecasting

    2/33

    2

    Extrapolation TechniquesExtrapolation Techniques

    Real Estate AnalystsReal Estate Analysts -- faced with a difficult taskfaced with a difficult task

    longlong--term projections for small areas such asterm projections for small areas such as

    CountiesCounties

    Cities and/orCities and/or

    NeighborhoodsNeighborhoods

    Reliable shortReliable short--term projections for small areasterm projections for small areas

    Reliable longReliable long--term projections for regions countriesterm projections for regions countries

    Forecasting task complicated by:Forecasting task complicated by:

    Reliable, Timely and Consistent informationReliable, Timely and Consistent information

  • 8/2/2019 FIN454 Population Forecasting

    3/33

    3

    Sources of ForecastsSources of Forecasts

    Public and Private Sector ForecastsPublic and Private Sector Forecasts

    Public: California Department of FinancePublic: California Department of Finance

    Private: CACIPrivate: CACI

    Forecasts may be based on large quantities ofForecasts may be based on large quantities of

    current and historical datacurrent and historical data

  • 8/2/2019 FIN454 Population Forecasting

    4/33

    4

    Projections are ImportantProjections are Important

    Comprehensive plans for the futureComprehensive plans for the future

    Community General Plans forCommunity General Plans for

    Residential Land UsesResidential Land Uses

    Commercial Land UsesCommercial Land Uses

    Related Land UsesRelated Land Uses

    Transportation SystemsTransportation Systems

    Sewage SystemsSewage Systems SchoolsSchools

  • 8/2/2019 FIN454 Population Forecasting

    5/33

    5

    DefinitionsDefinitions

    Estimate:Estimate:

    is an indirect measure of a present or pastis an indirect measure of a present or past

    condition that can be directly measured.condition that can be directly measured.

    Projection (or Prediction):Projection (or Prediction):

    are calculations of future conditions that wouldare calculations of future conditions that would

    exist as a result of adopting a set of underlyingexist as a result of adopting a set of underlying

    assumptions.assumptions. Forecast:Forecast:

    is a judgmental statement of what the analystis a judgmental statement of what the analyst

    believes to be the most likely future.believes to be the most likely future.

  • 8/2/2019 FIN454 Population Forecasting

    6/33

    6

    Projections vs. ForecastsProjections vs. Forecasts

    The distinction between projections andThe distinction between projections and

    forecasts are important because:forecasts are important because:

    Analysts often use projections when they should beAnalysts often use projections when they should be

    using forecasts.using forecasts.

    Projections are mislabeled as forecastsProjections are mislabeled as forecasts

    Analysts prepare projections that they know will beAnalysts prepare projections that they know will be

    accepted as forecasts without evaluating theaccepted as forecasts without evaluating the

    assumptions implicit in their analytic results.assumptions implicit in their analytic results.

  • 8/2/2019 FIN454 Population Forecasting

    7/33

    7

    ProcedureProcedure

    Using Aggregate data from the past to projectUsing Aggregate data from the past to project

    the future.the future.

    Data Aggregated in two ways:Data Aggregated in two ways:

    total populations or employment without identifying thetotal populations or employment without identifying the

    subcomponents of local populations or the economysubcomponents of local populations or the economy

    I.e.: age or occupational makeupI.e.: age or occupational makeup

    deals only with aggregate trends from the past withoutdeals only with aggregate trends from the past without

    attempting to account for the underlying demographic andattempting to account for the underlying demographic andeconomic processes that caused the trends.economic processes that caused the trends.

    Less appealing than the cohortLess appealing than the cohort--componentcomponent

    techniques or economic analysis techniques thattechniques or economic analysis techniques that

    consider the underlying components of change.consider the underlying components of change.

  • 8/2/2019 FIN454 Population Forecasting

    8/33

    8

    Why Use Aggregate Data?Why Use Aggregate Data?

    Easier to obtain and analyzeEasier to obtain and analyze

    Conserves time and costsConserves time and costs

    Disaggregated population or employment dataDisaggregated population or employment data

    often is unavailable for small areasoften is unavailable for small areas

  • 8/2/2019 FIN454 Population Forecasting

    9/33

    9

    Extrapolation: A Two Stage ProcessExtrapolation: A Two Stage Process

    Curve FittingCurve Fitting --

    Analyzes past data to identify overall trends ofAnalyzes past data to identify overall trends of

    growth or declinegrowth or decline

    Curve ExtrapolationCurve Extrapolation --

    Extends the identified trend to project the futureExtends the identified trend to project the future

  • 8/2/2019 FIN454 Population Forecasting

    10/33

    10

    Assumptions and ConventionsAssumptions and Conventions

    Graphic conventions Assume:Graphic conventions Assume:

    Independent variable:Independent variable: x axisx axis

    Dependent variable:Dependent variable: y axisy axis

    This suggests that population change (y axis) isThis suggests that population change (y axis) is

    dependent on (caused by) the passage of time!dependent on (caused by) the passage of time!

    Is this true or false?Is this true or false?

  • 8/2/2019 FIN454 Population Forecasting

    11/33

    11

    Assumptions and ConventionsAssumptions and Conventions

    Population change reflects the change inPopulation change reflects the change in

    aggregate of three factors:aggregate of three factors:

    birthsbirths

    deathsdeaths

    migrationmigration

    These factors are time related and are causedThese factors are time related and are caused

    by other time related factors:by other time related factors: health levelshealth levels

    economic conditionseconomic conditions

    Time is a proxy that reflects the net effect of aTime is a proxy that reflects the net effect of a

    large number of unmeasured events.large number of unmeasured events.

  • 8/2/2019 FIN454 Population Forecasting

    12/33

  • 8/2/2019 FIN454 Population Forecasting

    13/33

    13

    Alternative Extrapolation CurvesAlternative Extrapolation Curves

    LinearLinear

    GeometricGeometric

    ParabolicParabolic

    Modified ExponentialModified Exponential

    GompertzGompertz

    LogisticLogistic

  • 8/2/2019 FIN454 Population Forecasting

    14/33

    14

    Linear CurveLinear Curve

    Formula:Formula: Yc = a + bxYc = a + bx

    a = constant or intercepta = constant or intercept

    b = slopeb = slope

    Substituting values of x yields YcSubstituting values of x yields Yc

    Conventions of the formula:Conventions of the formula:

    curve increases without limit if the b value > 0curve increases without limit if the b value > 0

    curve is flat if the b value = 0curve is flat if the b value = 0

    curve decreases without limit if the b value < 0curve decreases without limit if the b value < 0

  • 8/2/2019 FIN454 Population Forecasting

    15/33

    15

    Linear CurveLinear Curve

  • 8/2/2019 FIN454 Population Forecasting

    16/33

    16

    Geometric CurveGeometric Curve

    Formula:Formula: Yc = abYc = abxx

    a = constant (intercept)a = constant (intercept)

    b = 1 plus growth rate (slope)b = 1 plus growth rate (slope)

    Difference between linear and geometricDifference between linear and geometric

    curves:curves: Linear:Linear: constant incremental growthconstant incremental growth

    Geometric:Geometric: constant growth rateconstant growth rate

    Conventions of the formula:Conventions of the formula:

    if b value > 1 curve increases without limitif b value > 1 curve increases without limit

    b value = 1, then the curve is equal to ab value = 1, then the curve is equal to a

    if b value < 1 curve approaches 0 as x increasesif b value < 1 curve approaches 0 as x increases

  • 8/2/2019 FIN454 Population Forecasting

    17/33

    17

    Geometric CurveGeometric Curve

  • 8/2/2019 FIN454 Population Forecasting

    18/33

    18

    Parabolic CurveParabolic Curve

    Formula:Formula: Yc = a + bx + cxYc = a + bx + cx22

    a = constant (intercept)a = constant (intercept)

    b = equal to the slopeb = equal to the slope c = when positive: curve is concave upwardc = when positive: curve is concave upward

    when = 0, curve is linearwhen = 0, curve is linear

    when negative, curve is concave downwardwhen negative, curve is concave downward

    growth increments increase or decrease as the x variablegrowth increments increase or decrease as the x variable

    increasesincreases

    Caution should be exercised when using forCaution should be exercised when using for

    long range projections.long range projections.

    Assumes growth or decline has no limitsAssumes growth or decline has no limits

  • 8/2/2019 FIN454 Population Forecasting

    19/33

    19

    Parabolic CurveParabolic Curve

  • 8/2/2019 FIN454 Population Forecasting

    20/33

    20

    Modified Exponential CurveModified Exponential Curve

    Formula:Formula: Yc = c + abYc = c + abxx

    c = Upper limitc = Upper limit

    b = ratio of successive growthb = ratio of successive growth a = constanta = constant

    This curve recognizes that growth willThis curve recognizes that growth will

    approach a limitapproach a limit Most municipal areas have defined areasMost municipal areas have defined areas

    i.e.: boundaries of cities or countiesi.e.: boundaries of cities or counties

  • 8/2/2019 FIN454 Population Forecasting

    21/33

    21

    Modified Exponential CurveModified Exponential Curve

  • 8/2/2019 FIN454 Population Forecasting

    22/33

    22

    Gompertz CurveGompertz Curve

    Formula:Formula: Log Yc = log c + log a(bLog Yc = log c + log a(bxx)) c = Upper limitc = Upper limit

    b = ratio of successive growthb = ratio of successive growth

    a = constanta = constant

    Very similar to the Modified Exponential CurveVery similar to the Modified Exponential Curve

    Curve describes:Curve describes:

    initially quite slow growthinitially quite slow growth

    increases for a period, thenincreases for a period, then growth tapers offgrowth tapers off

    very similar to neighborhood and/or city growth patternsvery similar to neighborhood and/or city growth patterns

    over the long termover the long term

  • 8/2/2019 FIN454 Population Forecasting

    23/33

    23

    Gompertz CurveGompertz Curve

  • 8/2/2019 FIN454 Population Forecasting

    24/33

    24

    Logistic CurveLogistic Curve

    Formula:Formula: Yc = 1 / YcYc = 1 / Yc--11 where Ycwhere Yc--11 = c + ab= c + abXX

    c = Upper limitc = Upper limit

    b = ratio of successive growthb = ratio of successive growth

    a = constanta = constant

    Identical to the Modified Exponential andIdentical to the Modified Exponential and

    Gompertz curves, except:Gompertz curves, except: observed values of the modified exponential curve and theobserved values of the modified exponential curve and the

    logarithms of observed values of the Gompertz curve are replacedlogarithms of observed values of the Gompertz curve are replacedby the reciprocals of the observed values.by the reciprocals of the observed values.

    Result: the ratio of successive growth increments of theResult: the ratio of successive growth increments of the

    reciprocals of the Yc values are equal to a constantreciprocals of the Yc values are equal to a constant

    Appeal: Same as the Gompertz CurveAppeal: Same as the Gompertz Curve

  • 8/2/2019 FIN454 Population Forecasting

    25/33

    25

    Logistic CurveLogistic Curve

  • 8/2/2019 FIN454 Population Forecasting

    26/33

    26

    Selecting Appropriate ExtrapolationSelecting Appropriate Extrapolation

    ProjectionsProjections

    First: Plot the DataFirst: Plot the Data

    What does the trend look like?What does the trend look like?

    Does it take the shape of any of the six curvesDoes it take the shape of any of the six curves

    Curve AssumptionsCurve Assumptions

    Linear: if growth incrementsLinear: if growth increments -- or the firstor the first

    differences for the observation data aredifferences for the observation data areapproximately equalapproximately equal --

    Geometric: growth increments are equal to aGeometric: growth increments are equal to a

    constantconstant

  • 8/2/2019 FIN454 Population Forecasting

    27/33

    27

    Selecting Appropriate ExtrapolationSelecting Appropriate Extrapolation

    Projections, contProjections, cont

    Curve AssumptionsCurve Assumptions

    Parabolic: Characterized by constant 2ndParabolic: Characterized by constant 2nd

    differences (differences between the first differencedifferences (differences between the first differenceand the dependent variable) if the 2nd differencesand the dependent variable) if the 2nd differences

    are approximately equalare approximately equal

    Modified Exponential: characterized by firstModified Exponential: characterized by first

    differences that decline or increase by a constantdifferences that decline or increase by a constantpercentage; ratios of successive first differences arepercentage; ratios of successive first differences are

    approximately equalapproximately equal

  • 8/2/2019 FIN454 Population Forecasting

    28/33

    28

    Selecting Appropriate ExtrapolationSelecting Appropriate Extrapolation

    Projections, contProjections, cont

    Curve AssumptionsCurve Assumptions

    Gompertz: Characterized by first differences in theGompertz: Characterized by first differences in the

    logarithms of the dependent variable that declinelogarithms of the dependent variable that declineby a constant percentageby a constant percentage

    Logistic: characterized by first differences in theLogistic: characterized by first differences in the

    reciprocals of the observation value that decline byreciprocals of the observation value that decline by

    a constant percentagea constant percentage

    Observation data rarely correspond to anyObservation data rarely correspond to any

    assumption underlying the extrapolationassumption underlying the extrapolation

    curvescurves

  • 8/2/2019 FIN454 Population Forecasting

    29/33

    29

    Selecting Appropriate ExtrapolationSelecting Appropriate Extrapolation

    Projections, contProjections, cont

    Test Results using measures of dispersionTest Results using measures of dispersion

    CRV (Coefficient of relative variation)CRV (Coefficient of relative variation)

    ME (Mean Error)ME (Mean Error)

    MAPE (Mean Absolute Percentage Error)MAPE (Mean Absolute Percentage Error)

    In General: Curve with the lowest CRV,MEIn General: Curve with the lowest CRV,ME

    and MAPE should be considered the best fitand MAPE should be considered the best fit

    for the observation datafor the observation data JudgementJudgementis requiredis required

    Select the Curve that produces resultsSelect the Curve that produces results

    consistent with the most likely futureconsistent with the most likely future

  • 8/2/2019 FIN454 Population Forecasting

    30/33

    30

    Selecting Appropriate ExtrapolationSelecting Appropriate Extrapolation

    Projections, contProjections, cont

    Year Linear Linear Geometric Parabolic Modified

    Odd Even Exponential

    1960 98,640 101,114 98,956 94,683 101,017

    1965 109,050 109,669 108,263 111,029 109,979

    1970 119,460 118,223 118,444 123,417 118,6721975 129,870 126,777 129,583 131,849 127,105

    1980 140,280 135,331 141,770 136,323 135,285

    1985 150,690 143,886 155,102 136,840 143,220

    1990 161,100 152,440 169,688 133,400 150,917

    1995 171,510 160,994 185,647 126,003 158,384

    2000 181,920 169,549 203,106 114,649 165,626

    Alternate Estimates and Projections

    Curve CRV ME MAPE Upper Limit

    Linear (Odd) 0.01 0.00 2.82% none

    Linear (Even) 0.01 -1,030.95 1.61% none

    Geometric 0.01 56.87 3.17% none

    Parabolic 351.27 0.00 1.41% none

    Modified Exponential 73.29 -46.53 3.55% 400,000

    Input and Output Evaluation Statistics

  • 8/2/2019 FIN454 Population Forecasting

    31/33

    31

    Housing Unit MethodHousing Unit Method

    Formulas:Formulas:

    1) HH1) HHgg = ((BP*N)= ((BP*N)--D+HUD+HUaa)*OCC)*OCC

    2) POP2) POPgg = HH= HHgg * PHH* PHH

    3) POP3) POPff= POP= POPcc + POP+ POPgg Where:Where: HHHHgg Growth In Number of HouseholdsGrowth In Number of Households

    BPBP Average Number of Bldg. Permits issued per year sinceAverage Number of Bldg. Permits issued per year since

    most recent censusmost recent census

    NN Forecast period in YearsForecast period in Years

    HUHUaa No. of Housing Units in Annexed AreaNo. of Housing Units in Annexed Area

    OCCOCC Occupancy RateOccupancy Rate

    POPPOPgg Population GrowthPopulation Growth

    PHHPHH Persons per HouseholdPersons per Household

    POPPOPcc Population at last censusPopulation at last census

    POPPOPff

    Population ForecastPopulation Forecast

  • 8/2/2019 FIN454 Population Forecasting

    32/33

    32

    Housing Unit Method ExampleHousing Unit Method Example

    Forecast Growth in Number of Housing UnitsForecast Growth in Number of Housing Units

    1)1) HHHHgg = ((BP*N)= ((BP*N)--D+HUD+HUaa)*OCC)*OCC

    HHHHgg = ((193*5)= ((193*5)--0+0)*95.1%0+0)*95.1%

    HHHHgg = 918= 918

    Forecast Growth in PopulationForecast Growth in Population 2) POP2) POPgg = HH= HHgg * PHH* PHH

    POPPOPgg = 918 * 2.74= 918 * 2.74

    POPPOPgg = 2,515= 2,515

    Forecast Total PopulationForecast Total Population

    3)3) POPPOPff= POP= POPcc + POP+ POPgg POPPOPff= 126,003 + 2,515= 126,003 + 2,515

    POPPOPff= 128,518= 128,518

  • 8/2/2019 FIN454 Population Forecasting

    33/33

    33

    So ThatsSo Thats

    Population ForecastingPopulation Forecasting

    Wayne Foss, MBA, MAI, Fullerton, CA USA

    Email: [email protected]