A subsidy program worth tens of billions of dollars, which came into effect in Turkey in 2012, lifted revenue, jobs and productivity at firms in the country’s poorest regions. However, long-run modelling shows it closed the wage gap with wealthier regions by only 1%, according to research presented at a UNSW Business School economics workshop.

The finding carries a warning for governments (including Australia’s) that are turning back to place-based industrial policy through programs such as Future Made in Australia. Chad Syverson, George C. Tiao Distinguished Service Professor of Economics at the University of Chicago Booth School of Business, told the 4th Firm and Industry Dynamics Workshop, jointly hosted by UNSW Sydney and the University of Hong Kong, that subsidised firms performed far better than the aggregate wage data suggested.

“A five-percentage-point increase in the investment tax credits corresponds to 3.2% increase in firms’ total factor productivity. It increased revenues and employment even more than that, quite a bit more than that,” said Prof. Syverson, whose research spans several topics, with a particular focus on the interactions of firm structure, market structure, and productivity.

Subsidy program lifted firm revenue and jobs

Turkey introduced the program in 2012, dividing the country into six regions ranked from wealthiest to poorest and offering escalating subsidies to firms investing in physical capital in poorer areas. The package combined tax exemptions on capital purchases, interest rate support on loans, refundable corporate tax credits worth up to 50% of investment costs, and relief on social security contributions.

Prof. Syverson said the combined effect of these measures made capital purchases inexpensive for a firm in region six (the poorest of the six regions) of Turkey. A firm there could receive up to half its investment cost back directly through the tax credit, on top of a subsidised loan and a tax exemption on the purchase price of the equipment itself.

Prof. Syverson and his co-authors used firm balance sheet data, production network records, and linked employer-employee data covering most of the Turkish economy to estimate the causal effect of the payments. A five-percentage-point rise in the investment tax credit rate, roughly the gap between the average paid in the poorest region and the average paid in the wealthiest, lifted firm-level revenue by 16%, employment by 8%, and total factor productivity by 3.2%.

Learn more: What decades of productivity research mean for business leaders

The researchers could not confirm whether the productivity gain reflected efficiency improvements from newer machinery or simply pricing power. Prof. Syverson said the absence of business-stealing effects across the supply chain pointed toward the former. “Our guess is it’s probably some sort of vintage capital effect: you’ve got old machinery, you buy this newfangled new machinery with the subsidy, that new lathe is now more efficient than the old lathe, and your TFP (total factor productivity) goes up as a result,” he said.

Spillovers to suppliers and customers along the production chain added a further one per cent to revenue and employment for firms connected to subsidised businesses. Those spillovers did not extend to productivity, a result Prof. Syverson said simplified the task of feeding the estimates into the aggregate model used later in the study.

The program targeted manufacturing, mining, and warehousing rather than services, and eligibility depended on a firm operating in the right region and industry and making an investment in physical capital. Take-up varied across industries and did not always rise in step with the generosity of the subsidy, with some sectors recording higher uptake in the second-poorest region than in the poorest.

"How successful were they at closing wage gaps between regions, to give the answer away? Not very successful"

CHAD SYVERSON

To isolate the causal effect of the payments from firms simply choosing to apply when they already expected strong returns, the researchers instrumented actual subsidies received against each firm’s statutory eligibility, and checked that subsidised and non-subsidised firms in the same region showed no difference in growth before the program began. The data covered only the formal economy, so the team adjusted the aggregate results using separate estimates of informal activity by industry and region.

Wage gap between regions barely moved

Despite those firm-level gains, the researchers estimated the program’s effect on the wage gap between the poorest and wealthiest regions was small: close to one percentage point over the life of the program, rising only slightly further by 2040, decades after payments stopped.

“How successful were they at closing wage gaps between regions, to give the answer away? Not very successful,” Prof. Syverson told the workshop, before setting out why he and his co-authors built a computational trade model, spanning industries and regions, to trace how the subsidy dollars actually moved through the economy once households, firms, and capital owners responded.

Employment across the subsidised regions rose faster than wages, a pattern the researchers linked to a flat labour supply curve within Turkey. “If you shift out labour demand with the subsidy, what happens is you get more employment, but you don’t get more wages, higher wages,” Prof. Syverson said. Additional workers entering the subsidised regions were absorbed without raising pay, meaning the program shifted the geography of jobs more than the geography of pay.

Migration and trade blunted the policy

Three channels ate into the wage gains intended for poorer regions. The first was labour mobility. People had been leaving Turkey’s poorest regions for wealthier cities, including Istanbul, Ankara, Izmir, and Antalya, at more than 1% a year before the subsidies began, and the payments mainly slowed that outflow rather than reversing it.

“You shouldn’t think of the net flows as being one to six,” Prof. Syverson said, referring to the richest and poorest regions. “It’s just there was a lot of six-to-one already before the program, and the program slowed down the out-migration from six to wealthier regions.”

The second channel ran through supply chains. Firms in poorer regions buy from, and sell to, businesses in wealthier ones, so a share of every subsidy dollar flowed straight back to the country’s richer cities.

Subscribe for the latest research, analysis and insights from UNSW Business School

“If you subsidise activity in, say, a poor area, and you increase activity there, some of the customers and some of the suppliers of those plants in those poor areas are in rich areas, and because of the spillover effects, they’re going to get some of the benefits of the subsidy, so effectively you throw 10 million lira at the poor region, but some of that 10 million lira goes to producers in wealthier regions, and so the actual effect is reduced in terms of what happens to the dispersion in activity,” Prof. Syverson said.

The third channel was capital. Landlords and investors captured part of the benefit through higher returns on capital in subsidised regions, an effect the researchers linked to Ricardian rent, in which the owner of a fixed asset ends up absorbing much of the value of the subsidy attached to it.

When the researchers removed all three channels from their model, the implied reduction in wage inequality grew to roughly three and a half times the actual result, with capital flows accounting for about a quarter of that gap and migration and trade splitting the rest.

Lesson for Future Made in Australia

Workshop attendees pressed Prof. Syverson on the relevance of the findings to current policy, including the United States CHIPS Act and the Future Made in Australia program, both of which direct public money toward building domestic industrial capacity in specific locations or sectors.

Prof. Syverson would not say whether the Turkish program was worthwhile, noting the analysis excluded the cost side of the ledger. “What we have been careful not to say is that this program was a good idea, and not because we believe it’s not. We deliberately didn’t try to compute it, because we really don’t have the other side of the scale. All this money has to come from somewhere,” he said.

Every dollar of public spending carries a cost beyond its face value, whether raised through taxation or inflation, and Prof. Syverson said the research had not modelled potential capital misallocation the subsidies might have caused, nor any offsetting benefit from agglomeration effects the model did not capture.

Governments must be clear on their objective

Asked what advice he would offer governments weighing similar programs, Prof. Syverson said the answer depended on what a government was trying to achieve, rather than on the mechanics of the policy itself. “If you want something to happen somewhere, you can make that happen,” he explained.

“The issue is that all of this doesn’t come for free. There are all sorts of spillovers. You might just pull activity that would have happened somewhere else anyway from there to here, and you have to ask yourself, are you willing to give up that there to have this here?” said Prof. Syverson, who noted that redistribution goals could justify a program even without an efficiency case – provided governments were explicit about the trade-off and the value they placed on shifting activity between regions or groups.

"The issue is that all of this doesn’t come for free. There are all sorts of spillovers"

CHAD SYVERSON

He extended that reasoning to national security arguments behind semiconductor subsidies in the United States. In that case, he said, a government accepts some economic inefficiency in exchange for a form of insurance against a strategic risk, rather than expecting a direct return on the money spent.

Five recommendations for industry professionals

1. Policy teams should define whether the objective is employment, productivity, wage convergence, national security or redistribution.
2. Business leaders should test whether subsidies change project viability or only change the timing of investment.
3. Investors should model returns without tax credits, loan support and social contribution relief.
4. Firms should map which suppliers, customers, landlords, and investors will capture a share of the subsidy.
5. Decision-makers should set review points and withdrawal conditions before funding begins.