1. Development and Validation of Firm Demography Model Case Study Analysis in Phoenix and Tucson Megaregion
16th TRB National Transportation Planning Applications Conference
Arun Kuppam, Cambridge Systematics Srinath Ravulaparthy, Citilabs Brent Selby, Cambridge Systematics, Inc Kyunghwi Jeon, Maricopa Association of Governments Sreevatsa Nippani, Maricopa Association of Governments Vladimir Livshits, Maricopa Association of Governments
May 18, 2017

2. Firm Synthesis Population Evolution Model Firm Evolution Model
Emigration / Immigration
Aging / Mortality / Fertility
Household formation / dissolution / child leaving
Education / Occupation / Wages
Firm Evolution Model
Firm conception
Firm migration
Firm death
Firm growth – Change in employment
Technology – Productivity

3. Evolution of Firm Population
Accurately describes evolution of firm population in space and time
[Ft+1 = Ft + Bt – Dt + It – Ot]
Firm population at time t
Firm population at time t+1
Within migration
In-migration (I)
Formation or births
(B)
Out-migration
(O)
Dissolution or deaths
(D) Ft
Ft+1 Bt Dt It Ot

4. Drivers of Firm Evolution
Firm internal
Employment size
Sales
Age
Standalone/headquarters
Market Area
Population
Average household income
Student enrollment by school and university
Agglomeration economies
Localization – number of employees within same industry
Urbanization – number of employees from other types
Regional indicators and transport access
Miles of roadway by facility type
Major hubs in the region

5. Evolution of Firm Population
National Establishment Time Series (NETS) Database 1991 to 2012
F = S + B - D + I - O

6. Firm Evolution 1991 to 2012
Formation and Dissolution Rates

7. Model Components Firm Birth Firm Death In-migrate Out-migrate
Negative binomial distribution is fitted to the data, for sampling new born
Attributes of employment size and type are sampled at random from population
Firm Death
Binary choice model – firm is the unit of analysis determining the probability of death.
One model specified by controlling for industry type
Within-migration
Binary choice – firm is the unit that determines probability of relocation.
In-migrate
Determine a fixed rate of in-migration from data
Out-migrate
Determine a fixed rate of out-migration from data
Growth / Decline
Log-linear specification of employment size
Control for previous year employment to determine future
Location choice
MNL specification with universal utility function
Control for industry type in model specification
Applied to – birth, within-migration and in-migration firms

8. Model Validation Population Aggregates County level totals
NAICS 2-digit Industry types
Data sources InfoGroup and Maricopa County Employer Database (ED)
Simulation of 20% validation sample from NETS
Firm evolution simulation from 2012 to 2015
Zonal employment validation
Spatial distribution with simulated and observed
Scatter plots by zone by sector types

9. Firm Synthesizer Model vs. Observed

10. Conclusions
Firmography approach is better than assuming constant shares for forecast years (as it accounts for formation, relocation, dissolution)
Simulation & validation results also confirm predictability and reliability of the Firmography framework
Micro-level and zonal predictions closely resemble observed trend