The Prospects of Irish Regions Edgar Morgenroth Economic and Social Research Institute Trinity College Dublin
Introduction What has been happening to our regions and where are they going. Where would like them to go? We can say a lot about past trends but a lot less about where they are going. Of course where the regions are going will in part be determined by policy what are the right levers, what is possible, what is feasible? 2
Outline Past trends Modelling infrastructure Population and employment projections Drivers of growth 3
Population 1971 to 2016 Source: Own calculations using CSO Census Data 4
Real Gross Value Added Per Capita 1991 to 2014 Source: Own calculations using CSO Regional GVA Data 5
Real Total Income Per Capita Source: Own calculations using CSO County Income Data 6
Unemployment Rate (ILO) 1998 to 2016 Source: CSO Quarterly National Household Survey 7
Modelling Regional Development How can we interpret the past trends and how can we project likely future trends and analyse the effect of policy on these? To do that we need theory and models. To date the modelling infrastructure to provide such inputs is underdeveloped. In particular there is a lack of appropriate economic modelling tools (exception is SMILE see Hynes et al 2009). At the national level economic models are regularly used to provide projections and policy analysis much of this has been based on the ESRI HERMES model (Bergin et al, 2013). A new compact macroeconometric model (COSMO) has been developed at the ESRI (Bergin et al 2016). 8
COSMO Modelling Process NiGEM Global Model Demographic Model Fiscal Plans External Assumptions Migration Database Equations Cosmo Projected path of economy over t+5, t+25 etc. Baseline Scenario Alternative Scenarios Risk Analysis, Analyse of Policy Choices Cosmo has 170 equations including detailed modelling of the relationship between banking/finance and the real economy (e.g. Housing) Satellite models can be run in conjunction e.g. a regional satellite model. 9
Key Issues in Developing Models What type of model? I-O, CGE, DSGE, Microsimulation, Econometric... One could build bottom-up, top-down or mixed models the latter can be constructed as a regional satellite model to a national model (COSMO) which ensures consistency with national projections and can help to add additional insights on the impact of policies. What is the appropriate spatial scale region, county, ED? What is the appropriate time horizon? 5 years is too short for planning purposes. Many variables change only slowly (the population grew by 3.7% between 2011 and 2016). There are big data gaps sectoral disaggregation, lack of specific variables, short time coverage... 10
Population Models There has been some modelling work done on population e.g. CSO Regional Population Forecasts and the ESRI has an Irish County Population Projection Model (see Layte ed., 2009). The (national) regional and county level population projections are developed using the cohort component approach. County population projections have also been downscaled to the ED level (Murphy, Crawford and Morgenroth, 2014). The downscaling is achieved using an econometric model which reflects some local factors (population, density, accessibility) to attribute the population change shares to individual EDs. A similar approach was also used to project employment (jobs) in Eds using outputs from the ESRI HERMES model. 11
Cohort Component Model Cohort component models are based on the fundamental balancing equation of population growth (g i,t ): N N I I g B D ) ( I E ) ( I E ) r, t ( r, t r, t r, t r, t r, t r, t The population at a future time (t+i) is then given as: P g P r, t i r, t r, t This requires assumptions/estimates on mortality, fertility, international migration and internal migration. The influences of individual behaviour and policy is usually not modelled but simply assumed. No interaction with economic conditions (e.g. Income or unemployment), the housing market, development capacity etc. 12
Annual Average Population Growth 2011-2021 Source: Own calculations using CSO Census Data ESRI 2014 13
Population Growth 2011 to 2051 Source: Own calculations using CSO Census Data 14
Job Density 2011 and 2051 Jobs per km2 51-200 1-25 201-500 26-50 501 and above Source: Own calculations using CSO Census Data 15
Issues and Model Improvements The models are done on a business as usual basis. They predict continued shift of population and activity east and continued sprawl is this what we want? These projections have some weaknesses. Internal migration patterns have been quite unstable current assumptions are subjective and likely to be wrong. A model of the underlying drivers of internal migration patterns is needed house price differentials, employment opportunities, wages... They don t properly model the regional economies and their dependencies. This is difficult but can be done! 16
Internal Migration 17
What Determines Regional Development Prospects? There is an extensive literature on the determinants of growth the key lesson is that capital accumulation in the widest sense matters: - Population/Labour Force; - Private capital; - Infrastructure; - Human Capital; - Knowledge capital (R&D, innovation). - Social capital (community, rule of law, institutions, leadership...) The literature also shows that spatial factors matter: - Density; - Agglomeration; - Amenities. How markets function also matters: - Competition (or lack thereof) 18
Real Investment Per Worker (Industry) Source: Own calculations using CSO Census of Industrial Production 19
Real Per Capita Public Capital Expenditure by Region 3500 3000 2500 2000 1500 1000 500 0 Border Midlands West Dublin Mid-East Mid-West South-East South-West Source: Own calculations based on Morgenroth (2010). 20
Real Per Capita Public Capital Expenditure on Roads by Region 1,400 1,200 1,000 800 600 400 200 - Border Midlands West Dublin Mid-East Mid-West South-East South-West Source: Own calculations based on Morgenroth (2010). 21
Human Capital: Years of Schooling Source: Own calculations using CSO Census Data 22
1996q1 1996q4 1997q3 1998q2 1999q1 1999q4 2000q3 2001q2 2002q1 2002q4 2003q3 2004q2 2005q1 2005q4 2006q3 2007q2 2008q1 2008q4 2009q3 2010q2 2011q1 2011q4 2012q3 2013q2 2014q1 2014q4 House Price Dispersion Across Counties Standard Deviation of House Prices 100,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 - Source: Own calculations using data from the ESRI/PTSB house price index and Daft.ie. 23
Implications The regional development patterns don t just follow the national business cycle some regions are faring systematically better. The existing projections suggests a continued shift of the population and activity east and continued sprawl. Scale has something to do with this but so do the underlying drivers of growth. However, it is difficult to give an authoritative assessment of the likely economic prospects of the regions at this point given the lack of appropriate modelling tools. Ongoing work aims at building a regional satellite model to COSMO in order to enhance the modelling capacity and to provide consistent projections. 24