Forestation to Mitigate Climate Change: Impacts on Regional Employment Distribution--Picking Winners and Losers

Complex social-ecological systems shift constantly in response to individuals, organizations, and government decisions. Policy makers need to anticipate system-wide consequences to both targeted and non-targeted system components, to avoid costly or irreversible mistakes. However, complex joint outcomes of individual and collective decisions are difficult to predict. Scenarios can help anticipate system-wide responses to specific interventions. We illustrate their use with a policy to set aside land for forestation in an economic region, to mitigate climate change. Does this policy affect the number of jobs at locations losing development space?
We use a dynamic model with 4 nonlinearly estimated parameters to anticipate regional distribution of jobs in time, using publicly accessible data. With Northeast Ohio (NEO) and Dallas-Fort Worth (DWF) 2001-2015 data (Kaufman et al. Applied Network Science 4(1) (2019) 1-17), by 2015 the actual-predicted correlations were .98 (NEO) and .99 (DWF) for municipal job shares of the regional total. Here we consider removing land from access to development to plant trees Northeast Ohio. Forestation requires specific locations (Matthews et al. Landscape ecology, 29(2) (2014) 213-228): at the edge of existing forests trees grow faster and reduce fragmentation, enhancing ecosystem diversity. We identified NEO forests to which trees can be added, and forest-adjacent vacant commercial/industrial land, some of which would likely be redeveloped with industrial uses. NEO commercial/industrial acreage and respective number of jobs per locality are similar (r=0.93, p<0.01), so land “lost” to forestation is roughly directly proportional to jobs lost. Let AFt,x be the forested area of locality x at time t. Then the difference aft,x between the forestation area and the regional mean is 〖af〗_(t,x)=〖AF〗_(t,x)/(∑_y▒〖AF〗_(t,y) )-1/N, where N is number of localities in the region. The fraction of jobs nt,x changes by –aft,x. Thus as a locality’s forestation area goes above the regional mean, its jobs share declines. The dynamic equation for how employment redistributes in the region from t to t+1 is n_(t+1,x)-n_(t,x)=(P_(t,x)-(P_t ) ̅ ) n_(t,x)-〖af〗_(t,x), where Pt,x is the market potential at x, at time t: P_(t,x)= ∑_y▒〖q_(x,y) n_(t,y) 〗 and q is matrix of inter-localities interactions. We estimate model parameters with 2012-2013 data. In 2014 we withdraw vacant commercial/industrial land and anticipate NEO jobs location for 2014-2040. We illustrate results at 3 localities. Solon—thriving outer suburb—gets forestation above the regional average. Shaker Heights—quasi-dormitory inner-suburb—gets below-average forestation. We compare employment predictions for the forestation scenario (blue line, Figure 1) to “no action” (red line). Solon’s regional share of jobs drops while Shaker Heights’ share goes up, akin to packing urban centers and reducing urban sprawl—a positive outcome. However, jobs in 12% of other NEO small localities disappear, e.g. Cuyahoga Heights Village (Figure 2). Thus employment “winners” (e.g. Shaker Heights) and “losers” (Solon and Cuyahoga Heights Village). Our results can inform regional plans.