By: Mike Denham, chairman of the TaxPayers' Alliance
Now that it is (almost certainly) an election year, we are once again hearing talk of tax cuts from the government. The opposition is also hitting hard on tax messages. This should be good news for taxpayers, but it is important to dig beneath the slogans a little.
Why should we be cutting taxes to start with? Well, it is now universally accepted (apart from worryingly prominent ‘degrowth’ campaigners) that we need to find ways to turn up the dial on economic growth. Supply side reform is obviously crucial here; that could involve deregulation or planning reform. Changing the profile of government spending might also help, with more money used for productivity-enhancing infrastructure.
But tax changes still matter. Why? Because incentives matter. To that end, there is a lot of discussion about the Office for Budget Responsibility (OBR), and how it models tax changes when making its important forecasts that drive government policy. Does the OBR focus on static modelling too much? Perhaps, although the OBR itself has attempted to communicate how it incorporates behavioural changes into its modelling.
Both the main parties will be making a big pitch on economic growth, so it is worth casting our minds back a few months to the autumn statement to analyse how this might play out ahead of the budget in March.
0.3 per cent additional growth by 2028-29
Much was made of the expanded economic output (and, conversely, slower growth) forecast by the OBR over the forecast period, not to mention the impact on investment and growth of making the ‘full expensing’ regime for corporation tax permanent. But the OBR was only expecting 0.3 percentage points of growth to come from policy announcements.
That makes some sense. Projects within smaller businesses would already be fully expensed under other allowances such as the annual investment allowance, which allows firms to expense investment in plant and machinery up to £1 million. The previous, temporary scheme, however, unless economic agents assumed that it would indeed become permanent, was only open to expenditure that fell within three years. So relatively short-timeframe projects (that were nonetheless big enough to not be covered by the annual investment allowance) would be affected (and already assumed in the OBR’s numbers). The difference would be found in large projects that would have still not been viable due to the short nature of the temporary scheme. After all, full expensing for a project where the bulk of the expenditure was expected to be incurred after the scheme ended (for example, one that took a year to get going and then five years to complete, with only the first two years falling within the scheme) might well still be unviable.
The OBR assumed the previous temporary capital allowance scheme brought forward a chunk of investment spending but didn’t alter the long-term optimal capital stock, leading to an actual fall in investment after the scheme was set to end. Here’s what they said in March 2023:
In 2024-25 in particular, this 0.3 per cent figure largely reflects the impact of the temporary increase in the generosity of capital allowances for businesses, which lets firms reduce their taxable profits by 100 per cent of the cost of their investments in plant and machinery for three years from April 2023. This incentive to accelerate investment plans boosts our business investment forecast by amounts peaking at almost 31⁄2 per cent in 2024-25 and 2025-26. But the policy’s temporary nature leaves the optimal capital stock unchanged in the long run, so in the final year of the forecast business investment is 4 per cent lower than it would otherwise be.
The OBR works with a five-year forecast horizon. That’s reasonable for many purposes. But how much of the dynamic effects of a policy like permanent full expensing could reasonably be expected to be played out within that horizon?
Suppose some marginal plant would be made viable by the measure which would enhance economic output. It's not likely to be a project where the expenditure could be substantially booked in the next three years anyway. And even then the first few operational years of such projects are often not that successful as it can take a while to both iron out teething problems and build up demand (ie, make economic use of the investment) before they become seriously profitable (and therefore adding to taxable potential GDP).
Making full expensing permanent, therefore, can only have an impact on projects which were too long to be viable under the (three year) temporary scheme but short enough to bear fruit under the (five year) forecast horizon.
The OBR also assumed that the capital stock will be higher by just 0.2 per cent at the end of the forecast period. Why? It might be because it is assuming a rising capital retirement rate:
2.23 The effect of this increase in gross investment on the stock of capital is offset by an increase in the assumed retirement rate of existing assets. The retirement rate of capital is the pace at which capital assets become obsolete, are withdrawn, and need to be replaced. This rate has risen steadily from 3.7 per cent in 2000 to 4.7 per cent in 2022, reflecting a growing share of shorter-lived intangible assets, such as computer software, in the capital stock. Having previously assumed that the retirement rate would stabilise and remain constant over our forecast period, we now assume intangible assets continue to grow as a share of the capital stock. So, the capital retirement rate continues rising to 5.0 per cent by 2028, reducing the capital stock by 0.7 per cent in 2027 compared to holding the rate flat.
Of course, if the nature of the capital stock is changing this rapidly, one might wonder if the OBR’s underlying production function (ie the rate at which additions to capital and labour are assumed to increase potential output) is understating the true dynamic effect of the new investment. Surely the whole point of IT investment is to boost productivity?
In its November 2022 Briefing paper No.8, Forecasting potential output – the supply side of the economy, the OBR noted how its output forecasts had been reasonably accurate, belied by significantly optimistic output per hour forecasts cancelled out by pessimistic total hours forecasts:
4.7 The middle and right-hand panels of Chart 4.4 show that the composition of output has been quite different from what we predicted. Total hours worked have mostly exceeded our forecasts while output per hour has mostly disappointed. From 2012 onwards, this has consistently reflected lower-than-expected unemployment. But it has also reflected other factors at different points, including higher-than-expected migration (particularly in the middle of the decade), stronger-than-expected (until recently) participation rates among older workers, and unexpected strength in average hours. Our initial overoptimism regarding productivity growth was the result of assuming that the post financial-crisis slowdown in productivity growth would be temporary and that pre-crisis rates of growth would ultimately return (albeit with no recovery of the shortfall in productivity relative to a continuation of pre-crisis trends). In the event, weaker-than-expected business investment (especially since the Brexit referendum) together with the failure of TFP growth to pick up has meant that growth in output per hour has remained sluggish for over a decade.
What might explain that forecasting error discrepancy? To what extent is that explained by underestimating the effect of personal allowance raises and welfare reforms on hours worked at the low end of the labour market while also underestimating the impact of persistently high taxes on capital and the higher end of the labour market?
It has been cheering to see the OBR shed some light on how it incorporates behavioural effects into its modelling. Nonetheless, questions remain about whether its assumptions match real-world incentive effects, particularly over longer horizons. Tax, after all, has a significant and long-term impact on economic output and growth.