We’ve come a long way. We’ve studied most of the important concepts a professional in this field needs, got hands-on with a tax-benefit model, and carried out some ambitious policy design.
The model you’ve been using is static and very conventional, and we’ve discussed its problems at length. Nonetheless I hope the policy simulations you’ve been doing show that static microsimulation is a powerful and useful tool.
Earlier, I expressed scepticism about whether a model that addressed all the limitations of this present one was feasible, and whether, if such an all-singing, all-dancing model was somehow built, it would actually be very useful. But interesting extensions in various directions have been made, and in this final section I’ll discuss a couple of them.
Last year, I and my colleague Howard Reed of Landman Economics1 were commissioned to produce long run poverty projections for Scotland, as part of the anti-child poverty policy2 we mentioned right at the beginning. We’ve seen that poverty simulation is classic tax-benefit model territory. But how could we project forward over 15 or more years? It turns out that we’ve already encountered the tricks that you need: reweighting and uprating. We saw above how, because of differential non-response, we needed to re-weight our FRS dataset - give more emphasis to some households than others in the output - so that the final results matched know facts about the overall populations, such as the numbers of people of different ages and genders. We can extend this idea to produce weights such that our data matches not current levels of these things, but projected levels. Likewise, for incomes, we can uprate the recorded levels in the data to match projections of thee things from macroeconomic forecasters. Such projections exist: the ONS and Public Records Scotland produce long-run forecasts for populations and household composition3 and the Scottish Fiscal Commission produces income forecasts4. Note that we’re not forecasting populations and incomes ourselves; rather, we’re making projections of poverty that are consistent with official forecasts.
An example of a successful model that captures some of the interactions that our model leaves out is DIMMSIM5, developed by the ADRS6 company for the South African Government. DIMMSIM is a micro- to- macro- model of the South African Economy. This merges MEMSA7, a conventional macroeconomic forecasting model of South Africa, with SATTSIM, a tax-benefit model of the sort we now know well. The model starts by doing the kind of tax-benefit calculations you’re now used to, but instead of just stopping there, the results for tax revenues, net incomes and the like are used as inputs to the macroeconomic side of the model, replacing the equations for these things that a conventional macro model would have. The macro model then uses the micro outputs to produce new estimates for employment and incomes, which are fed back into the micro part. This process continues until the model ‘converges’ - produces a set of micro outputs and macro outputs which are consistent with each other, and with the other constraints on the macro side (for trade deficits, for example). This procedure produces results that capture the feedbacks between, for instance, tax increases and reduced economic activity that the model you’ve been using model ignores.
Fisc, Scottish. “Scotland’s Economic and Fiscal Forecasts May 2019.” Scottish Fiscal Commission, May 2019. http://www.fiscalcommission.scot/publications/scotlands-economic-and-fiscal-forecasts/scotlands-economic-and-fiscal-forecasts-may-2019/.
Reed, Howard, and Graham Stark. “Tackling Child Poverty Delivery Plan: Forecasting Child Poverty in Scotland - Gov.scot,” March 2018. https://www.gov.scot/publications/tackling-child-poverty-delivery-plan-forecasting-child-poverty-scotland/pages/8/.
Scotland Web, National Records of. “National Records of Scotland - Population Projections.” Document. National Records of Scotland, 2019. /statistics-and-data/statistics/statistics-by-theme/population/population-projections/population-projections-scotland.