Addressing the evidence deficit: how experimentation and microsimulation can inform the basic income debate

Posted in: Basic income

Dr Luke Martinelli is Research Associate on the IPR's universal basic income project. This post draws in part on material presented in his recent IPR working paper The Fiscal and Distributional Implications of Alternative Universal Basic Income Schemes in the UK.

New and forthcoming IPR working papers – as well as experimental data from policy trials currently and imminently taking place across the world – address some of the core empirical issues around the feasibility of universal basic income (UBI), and how it can be designed most effectively. However, no amount of evidence can provide an escape from difficult political choices in the face of unavoidable conflicts between policy goals – or eliminate the need for advocates to address longstanding normative objections to UBI.

Despite numerous and well-publicised desirable qualities which appeal across the political spectrum, the case in favour of universal basic income (UBI) is far from decisive. There is much we still don’t know about UBI that must be addressed in order for the policy to move beyond superficial, ‘cheap’ support and to gain serious traction in the political sphere.

As Malcolm Torry argued in a previous post, as policymakers have begun to pay increasing attention to grassroots enthusiasm for UBI and acknowledge its theoretical strengths (as well as the weaknesses of existing social security provisions), there has been a shift from concerns about whether UBI is desirable per se, to questions of political feasibility and the best way to design and implement the policy. According to Torry, this progression is “evidence for the increasingly serious nature of the current debate”.

What policymakers need to know

Objections to UBI are numerous, but it is perhaps possible to identify three core barriers to feasibility as follows:

  • it would dis-incentivise work and encourage idleness;
  • it would be too costly; and
  • it would be inadequate to meet the diverse and complex needs of the poor.

As a result of these attributes – which are also, of course. inherently undesirable – sceptical observers have claimed that UBI would never generate the political support required for implementation. If labour supply was expected to contract significantly, the tax base would collapse and the policy would be seen as unsustainable. If the cost was seen as too high, voters would not consent to the requisite tax rises. And if disadvantaged people were to become poorer as a result, the policy would be seen as unacceptably unjust.

In response, advocates have claimed on the contrary that UBI would remove poverty and unemployment traps, would only require minor tax increases, and could easily accommodate provisions for those with higher needs (for example due to disability or housing costs).

Of course, the extent to which these attributes pertain – the labour market effects, fiscal costs, and distributional implications – depends wholly upon the specifics of the UBI scheme in question. As I discussed in a previous post, while basic income’s core attributes are a matter of definition, many are variable: most crucially, the level of payments, and the extent to which the UBI is conceived as replacement for or complement to existing benefits.

UBI is often discussed as a monolithic policy, which obscures clear understanding about likely effects; opponents debate at cross purposes, discussing completely different schemes and using them to support their favoured stance. Common ground – and an end to the impasse on these core issues – can only be achieved as a result of greater clarity about the diversity of proposals that exist, and the specific effects of varying core design features.

Policymakers are ultimately concerned about the practicalities of implementation: they need to know which schemes are feasible in the short- to medium-term, in the prevailing socio-political climate. They simply cannot afford to entertain utopian visions of a future in which no-one is compelled to work, or in which people are happy to accept tax rates an order of magnitude greater than those prevailing today. If they are to support UBI, they need to know whether the aforementioned barriers to feasibility can be bypassed – and how.

Fortunately, empirical evidence can help to assess the contradictory claims of advocates and opponents more effectively. There are two main forms of evidence: ex-ante (‘before the event’) models/simulations and ex-post (‘after the fact’) impact evaluations.

Ex-post evidence

Trials and policy experiments are important for several reasons. Most obviously, they give us important information about the effects of implementing basic income that cannot be gleaned from theory (for example and probably most crucially, on disputed labour market behavioural effects, but also on other non-financial outcomes such as health and wellbeing). Trials are also invaluable in uncovering unexpected outcomes and implementation issues, and in fine-tuning the detail of the policy in advance of scaling-up.

Perhaps equally importantly, they serve to foment conversation and political debate which simply would not occur in the absence of the trial. As Jurgen De Wispelaere observed in an IPR seminar, experiments

have important political demonstration effects… advancing the policy agenda by raising awareness amongst key stakeholders/general public, keeping open a window of opportunity, building a broad political coalition “en route”, and overcoming objections by demonstrating impacts

Basic income advocates have long drawn on trials that took place in the US and Canada between 1968 and 1980 (Widerquist, 2005). Observers note that despite some contraction in labour supply, these were far from the employment exodus predicted by UBI’s harshest critics; Forget (2011) has documented the health and wellbeing benefits of the Canadian trials.

There are also two prominent historical examples of universal payment programmes (not trials) in diverse country contexts: the Alaskan Permanent Fund Dividend and Iran’s reform of consumption subsidies (Widerquist and Howard, 2012; De Wispelaere, 2016). These examples demonstrate the administrative and political feasibility of UBI-type schemes. Experiments with basic income have recently taken place in Namibia (Haarmann et al., 2009) and India (Davala et al., 2015) with strong positive findings. Standing (2008) argues that the lessons from the implementation of various forms of (non-UBI) cash transfer in developing countries also provide compelling evidence in favour of UBI, in that negative labour market responses have been minimal and it is likely that the poor use income transfers to invest in productive activities – not for frivolous consumption, as is often portrayed.

Since 1 January this year, 2,000 Finnish individuals have been receiving an unconditional payment of €560 per month as part of a two-year government pilot. A number of city authorities in the Netherlands have been granted permission – under central government legislation allowing policy experimentation – to conduct trials eliminating the imposition of behavioural (labour market) conditions for benefit claimants. Both trials look sure to provide valuable information on how recipients respond to unconditional payments.

There is now a spate of proposals for further pilot schemes to be implemented in the coming years. Some of the interest has come from devolved regional and city-level authorities (Ontario in Canada; Fife and Glasgow in Scotland); some has come from the development assistance community (as in the case of the Give Directly pilot in Kenya); some has come from national administrations (India); and some has come from the private (tech) sector (Y-Combinator in California).

Limits to experimental evidence

However, there are limits to the insights that can be gained from the aforementioned experiments, including the ongoing Finnish and Dutch case studies.

In many cases the experiments fall far short of the evidential requirements of randomised control trials – the so-called ‘gold standard’ of policy evaluation. This is certainly the case in the example of the Namibian Basic Income Grant scheme, which is heavily criticised by Osterkamp (2013) as employing biased outcome indicators and lacking a control group – but methodological problems abound more broadly. For one thing, trials are necessarily of limited duration, and may not easily pick up longer-term effects of policy. In addition, the behavioural response to a trial of limited duration may be very different to the response to a policy that provides income security for a lifetime.

Further, policy outcomes depend heavily on the specific contexts in which they are implemented, limiting the applicability (or ‘external validity’) of trials to other countries, time periods, or groups of recipients. Experiments carried out in developing countries provide limited insight into the potential for basic income to be inserted into comprehensive welfare states such as the UK’s. The US and Canadian experiments are now several decades old. In the case of the experiments in Finland and the Netherlands, researchers are limited in applying the UBI ‘treatment’ on existing benefit claimants. It is not clear how the observed effects would vary to those for other groups in the context of a truly universal payment.

Most of the trials only fairly approximately resemble UBI ‘proper’, or the types of UBI upon which policy interest is most focused today in high-income countries (which aim at least to partially replace existing welfare state policies). The Alaskan and Iranian programmes differ in several crucial respects from the proposals of most basic income advocates, namely in the low and fluctuating value of payments and in their funding mechanisms; being paid from natural resource revenues, they are arguably significantly less likely to face political opposition (De Wispelaere, 2016). The US and Canadian experiments differ in that they involved a tax rebate mechanism based on reported income rather than an upfront payment (these are more usually termed negative income tax schemes rather than UBI). In the Dutch experiments, the implementing authorities were restricted by central government for political reasons, resulting in watered-down research designs. Specifically, claimants in the treatment group can only keep 50% of additional earnings, up to a maximum of €199 (ensuring their total combined income remains less than someone would earn working full-time at minimum wage), completely contradicting the principles of UBI. Even in the Finnish experiment – the first trial that could reasonably be described as a ‘real’ UBI within a high-income, mature welfare state – researchers were unable, for practical and administrative reasons, to implement tax changes alongside the implementation of the UBI (KELA, 2016). Considering that changes to the tax system are almost always a core element of any realistic basic income proposal, this is a significant weakness.

Another crucial limitation is that it seems likely that the effects of upscaling a policy to the national level would result in markedly different effects – with different implementing authorities, and significant macroeconomic effects not captured in the trial. As Widequist (2005) observes, being trials of limited scale, the US and Canadian experiments did not give rise to demand response “and therefore could not estimate the market response” of the policy (p. 50). Moving from an experiment run by a small, dedicated team of researchers to a nation-wide policy administered by a sprawling and perhaps badly-funded bureaucracy is likely to give rise to unforeseen implementation problems. In other words, the effects of a trial may be very different from those of the same policy rolled out across the board.

Finally, even if we are able to observe – reliably – the impacts of a policy, find that the effects are positive, and generalise the findings to another context, experiments such as these do not offer any way of weighing up beneficial impacts (relating to improved income security) against UBI’s fiscal costs (and against the costs and benefits of alternative policies). It is hardly surprising that giving people money would have a number of positive impacts; the question is whether UBI is the best use of the funds.

Ex-ante evidence

Fortunately, these are questions to which ex-ante microsimulation methods can readily be applied.  Microsimulation is a common approach to evaluating the effects of tax and benefit reforms with respect to fiscal implications, distributional effects, and (less commonly) impacts on static work incentives. Advances in computing power combined with the availability of large, representative income surveys makes it possible to compare outcomes of the prevailing ‘base’ policy environment with other hypothetical policy systems. This means that we have much greater capacity to assess and compare large numbers of different permutations of UBI.

Because it models the effects of policy reforms over a representative sample, microsimulation enables researchers to draw an accurate picture of overall impacts on the income distribution at the national level. However, a major shortcoming of this type of analysis is that it assumes no behavioural change (e.g. labour market response). This seems unrealistic in the context of such a wide-ranging reform as the implementation of a universal basic income, especially one paid at a generous level. For these reasons, microsimulation evidence should be complemented with ex-post analysis of observed behavioural responses.

A number of microsimulation studies have already modelled the effects of specific UBI schemes in the UK (e.g. Torry, 2016; Reed and Lansley, 2016). These studies have been instrumental in identifying ‘feasible’ ways of implementing basic income so as to minimise household losses and keep costs within the boundaries of ‘politically acceptable’ tax increases. The downside is that such schemes require the retention of the existing structure of social security alongside the UBI.

With our IPR Working Papers, we add to this burgeoning literature in important ways, with the specific intention of objectively informing policy audiences about the difficult decisions involved in designing UBI schemes. In particular, in The Fiscal and Distributional Implications of Alternative UBI Schemes in the UK, I systematically compare UBI schemes with a range of payment levels and compensatory tax/benefit changes. Unlike previous studies, I start from the presumption that at least part of the appetite for basic income arises from its promise to sweep away the mainstay of complex, intrusive and stigmatising means-tested benefits. In another (forthcoming) paper, Exploring the Distributional and Work Incentive Effects of Plausible UBI Schemes, I look at the distribution of winners and losers in more detail and introduce important indicators of static work incentives.

Combined, these microsimulation studies provide a great deal of important information required by policymakers in assessing competing UBI proposals, particularly bearing in mind the need to restrict net costs and the motivation to reduce poverty and unemployment traps that arise in means-tested systems. ‘Transitional’ forms of UBI – for example, one replacing the personal income tax allowance with a payment of equivalent value, and others covering subsets of the population – are suggestive of possible pathways to more generous and comprehensive forms of UBI. In consideration of the likely political imperatives, we model a number of (broadly) revenue-neutral schemes as well.

The inescapable conclusion of my research is that there are no easy answers to the questions facing UBI advocates; no ‘optimal’ basic income scheme. Rather, policymakers are faced with a series of trade-offs between competing goals of a) meeting need, b) controlling cost and c) retaining administrative simplicity and enhancing work incentives (through the elimination of means-testing). The analysis thus draws our attention to the difficulty involved in designing basic income schemes that satisfactorily compensate existing beneficiaries of the system while retaining the principle of universalism.

Complementary forms of evidence

This blog post has summarised the potential of two forms of evidence to inform debate and bring the current impasse around the feasibility of UBI to a close. I hope to have shown that ex-ante and ex-post studies are complementary; ex-ante simulations can say much about the fiscal and distributional effects of basic income, but nothing about behavioural responses or implementation challenges – and ex-post evaluations can provide insights into these outcomes, but have a number of shortcomings that limit their applicability to wider contexts and their utility in assessing different policy design features against alternatives.

While much public attention has been devoted to the upcoming trials, therefore – and while such trials certainly have their place – they cannot give us the full picture on UBI, particularly in relation to the fiscal feasibility of schemes. This is the value of the microsimulation approach I've presented in the IPR’s work; the evidence generated, I hope, will tell policymakers the other half of the story.

The full working paper The Fiscal and Distributional Implications of Alternative Universal Basic Income Schemes in the UK can be read and downloaded here.

 

References

Davala, S., Jhabvala, R., Standing, G. and Mehta, S. K. (2015). Basic income: A transformative policy for India. London: Bloomsbury Publishing.

De Wispelaere, J. (2016). “Basic Income in Our Time: Improving Political Prospects Through Policy Learning?” Journal of Social Policy, 45(4): 617-634.

Forget, E. (2011). “The Town with No Poverty: The Health Effects of a Canadian Guaranteed Annual Income Field Experiment.” Canadian Public Policy / Analyse de Politiques, 37(3): 283-305.

Haarmann, C., Haarmann, D., Jauch, H., Shindondola-Mote, H., Nattrass, N., van Niekerk, I. and Samson, M. (2009). Making the difference! – The basic income grant in Namibia. Assessment Report. Windhoek: BIG Coalition.

KELA (2016). From Idea to Experiment: Report on universal basic income experiment in Finland. KELA Working Paper 106/2016. Helsinki: KELA.

Osterkamp, R. “The Basic Income Grant Pilot Project in Namibia: A Critical Assessment.” Basic Income Studies, 8(1): 71-91.

Standing, G. (2008). “How Cash Transfers Promote the Case for Basic Income.” Basic Income Studies, 3(1).

Widerquist, K. (2005). “A failure to communicate: what (if anything) can we learn from the negative income tax experiments?” The Journal of Socio-Economics, 34(2005): 49-81.

Widerquist, K. and Howard, M. (eds.) (2012). Alaska’s Permanent Fund Dividend: Examining Its Suitability as a Model. New York: Palgrave.

Posted in: Basic income

Responses

  • (we won't publish this)

Write a response