Has the AKP facilitated cronyism through public procurement reforms in Turkey?

This year will mark the three-year anniversary of the Gezi protests which swept across most of Turkey during the summer of 2013. Many things contributed to the protests, but a major a factor in galvanizing such a large segment of the population was undeniably widespread view of severe government overreach in how it used construction to remodel Turkey’s urban landscape after its own design. Several of the so-called ‘crazy projects’ have not only laid bare the degree of government control over construction projects but also that the winners of public procurement projects typically constitute a relatively new niche of business entrepreneurs with strong connections to the AKP. Moreover, it also points to the public procurement sector – and in particular the role of state discretion therein – as the main instrument for this phenomenon.

Today, there are a number of sources pointing to the degree of political connections of public procurement winners in Turkey; ranging from the data visualizations of Mülksüzleştirme Ağları) to rich descriptions of how the political connections of public procurement winners have changed over time in parallell with legal reform giving the government greater say in allocation of contracts. Examples of these include a recent book by Ayğse Buğra and Osman Savaşkan qw well as a forthcoming book by Esra Gürakar (quoted here in the Economist). There are also several informative longer news articles (for example, here, here, and here) on the topic. With the resurgence of conflict in the southeast of Turkey, there are some concern that the Turkish government would confiscate conflict-torn land for the purpose of urban transformation, a process that might very well see government-linked firms gaining most of the contracts.

This begs the question to what extent the increase in AKP-connected firms winning procurement projects is also due to AKP’s changes in the procurement law? Moreover, has this had any economic consequences?

In order to shed light on this, Esra Gurakar and myself have a new paper looking at the effect of state discretion in construction public procurement. Here’s the abstract:

“We investigate whether increased state discretion in public procurement auctions affect economic costs and facilitate favoritism using data from the Turkish construction sector between 2005-2011. After parliament passed a legal amendment to the existing procurement law in 2008, construction auction projects with an estimated cost above a specific threshold became eligible for an auction procedure giving the contracting authority greater control over participating bidders through so-called restricted auctions. Using several identification strategies including difference-in-differences, instrumental variables, and regression discontinuity design, we find that increased discretion in public procurement not only increased costs – in terms of both the winning bid and rebate value – but also increased the likelihood of the winning firm being politically connected to the ruling AKP. Moreover, our analysis further shows that among varying forms of political connections to the AKP, the increased costs are particularly driven by winning firms connected to the two most powerful elites within the AKP during this period, namely individuals in the top political leadership as well as those affiliated with the Gülen movement.”

A noteworthy component here is that the main policy shift we examine occurred back in 2008, around the time when most observers were more focused on portraying the country as a model of Muslim democracy than really trying to understand how reforms were serving to shift power from one poorly accountable center to another.

In this post, I expand on a subset of the paper focusing on the political connectedness of construction procurement winners before and after the 2008 reform. The below figure shows cumulative distribution shares of awarded contracts on the y-axis, with the x-axis showing the share of firms ordered by total awarded contracts (in log scale). The left-hand graph includes data on all types of auctions – run through either open procedure or any of the closed procedures — grouped by whether contracts were awarded before or after 2008. The blue line thus denotes the cumulative share of awarded contracts 2008 and after, the dashed line denotes the same for the period before 2008.

The way to interpret this graph is the following. In the left-hand graph, the y-axis value for the solid blue line intersecting the 0.1 point on the x-axis means that the largest 10% of contract-winning firms accounted for approximately 68 % of all contracts auctioned between 2008-2011. Looking at the 0.01 point on the axis shows the corresponding for the largest contract-winning 1% of firms (who won roughly a quarter of all contracts during the period). The solid line indicates that the values on the y- and x-axis refers to the 2008-2011 period, and the dashed lines refer to the 2005-2007 period.

The solid red line’s value at this x-value implies that among the largest 10% of firms, the AKP-connected firms within this group accounted for half of all contracts during the post-2008 period. In other words, much of the concentration in a few firms winning a disproportionately large share of all the construction procurement contracts is because a few AKP-connected firms are winning a disproportionately large share of the contracts.

The y-values corresponding to the x-axis value of 1 in the graph shows the total share of contracts won by the group. For example, AKP-connected firms won more than 60% of all auctioned construction contracts post-2008, up from around 55% pre-2008.

The middle graph shows the corresponding relationships for auctions conducted using open competitive bidding, and shows very little difference in concentration between pre-and post-2008, either for all open-auctioneered contracts or just those won by AKP-connected firms.

The right-hand graph instead shows what happened for the auctions that were exempted from open competitive bidding. A defining feature of the 2008 reform was that it expanded the scope of auctions eligible for non-open procedures in which the government could exert more control over the process. And it certainly made a difference for the propensity of AKP-connected firms winning auctions. In the pre-2008 period, these firms won just under 40% of all auctions conducted in closed bidding procedures. In the post-2008 period, this share had jumped up above 70%. The 2008 reforms thus resulted in a significant boost in AKP-connected firms likelihood of winning construction procurement contracts.

In the paper we refine the analysis further using more standard differences-in-differences as well as regression discontinuity methods.

Another part of the paper focuses on the distribution of the gains of the 2008 amendment among firms within the AKP movement. The data available allow us to code, not just whether a firm has any connection to the AKP, but also whether this connection refers to a direct connection to the AKP (for example, being an AKP official, being part of the AKP cabinet, or having family ties to any of the previous two) versus an indirect connection through membership of any of the business associations affiliated with the AKP. These include organizations such as the MÜSİAD (a long-time supporter of the Millî Görüş movement) as well as the main business associations belonging to the Gülen movement (TUSKON). In reality, though, 80% of firms with such indirect connections are members of Gülen-linked associations.

The below graph is similar to the previous one but instead of showing the cumulative shares of awarded contracts for any AKP connection, the graph instead splits these into two groups: direct AKP connections and indirect (effectively: Gülen) connections.


The left-hand graph reveals that increases in the share of awarded contracts occurred for both firms with direct links to the AKP as well as those connected to the Gülen movement. As before, this is almost entirely driven by changes within auction types that forego open competitive bidding (the right-hand graph), with much smaller changes among open auctions (middle). The lıon’s share of the closed auctions are nonetheless captured by those with direct links to the AKP.

This represents one of the more pervasive results in the paper: the increased state discretion following the 2008 amendment resulted in a larger share of procurement auction contracts going to both those connected directly to the AKP as well as those connected to the AKP through the Gülen movement.

Undeniably, this analysis is done using data from 2005-2011, several years before the corruption cases of 2013 and the public fallout between Erdogan and the Gülen movement. Since then, current as well as former contributors to the Gülenist media have taken to Western media outlets to voice their concerns over the AKP government, often staking the movement’s motive for confronting its former ally Erdogan in the media and the courts as driven by concern over corruption:

“People think that it’s about a power struggle between the AKP government and the Gulen movement, but in fact it all comes down to that massive investigation.”
– Sevgi Akarçeşme interviewed in Deutsche Welle

Our fault was to report on government actions that are undermining the foundations of a democratic Turkey.Anyone who strays is harassed or fired. But as members of the free press, or whatever is left of it in Turkey, we are simply doing our jobs.
– Ekrem Dumanli, Editor-in-Chief of Zaman newspaper in The Washington Post

It’s a sad commentary on contemporary Turkey that people have to reach for conspiracy theories to explain why public officials are doing their job to prosecute corruption. It doesn’t seem to have crossed people’s minds that disunity within the A.K.P. coalition has given wiggle room to those who actually believe in the rule of law and want to enforce it.
– Andrew Finkel op-ed in The New York Times

In many ways this marks a significant shift in posture from earlier periods, especially during the crackdown on the Doğan media group for reporting on government corruption back in 2008. As I’ve noted earlier, after the government levied a hefty tax fine on the Doğan Group and subsequent voicing of concern by an international press organization  (WAN-IFRA)  a few months later, several pro-government newspapers objected to the critique that the government was muzzling the press. Among these were the Gülenist press, as was the case here and here.

In 2010 and 2011, the government initiated a number of large highway projects to be auctioned out under a new Highways Directorate, which had seen its powers boosted from another one of the AKP’s amendments to the procurement law. Many of these were enormous, dwarfing in size many past projects. This time, however, as we show in the paper, the Gülen movement found itself on the outside, winning very few of the contracts, which instead went to other firms with more direct connections to the AKP.

A year later, a spectacular raid on the Public Procurement Authority (PPA) resulted in several suspects detained on charges of tender rigging. Among these were the PPA’s Vice President Ali Kaya, “two KİK officials in charge of preparing reports on companies before tenders begin“, and a number of “well-known businessmen”. At the time, Cihan reported:

Upon finding out that Fermak A.Ş., which is run by Ferit Rızvanoğlu, the deputy chairman of the Kasımpaşaspor football club, allegedly bribed several KİK officials to manipulate the Ankara-Konya highway construction tender in favor of the company in 2009, the prosecutor’s office also found out that Fermak A.Ş. allegedly bribed the officials in order to win the tender of another public project, the İstanbul Pendik-Tuzla highway construction project, the daily also reported. Although a warrant was issued for Rızvanoğlu’s arrest, the police have been unable to find him. The police suspect Rızvanoğlu may have fled abroad after getting word of the warrant issued against him. Custom has been informed to take measures to prevent Rızvanoğlu’s escape abroad, the police say.

HDNER reported further that the “name of Fermak could also be seen at a report by the Prime Ministry’s Higher Inspection Committee regarding the Turkish State Railways (TCDD).”

Fermak (by most accounts a firm connected to TUSKON and the Gülen movement) and Ali Kaya – as well as the other detained in the sting – had most likely been dragged into the civil war between the Erdoğan’s side of the AKP and the Gülen movement.

The results from our paper are consistent with a story of how cronyism allowed those connected to the AKP, both directly and through the Gülen movement, to reap significant benefits from increased state discretion in the allocation of procurement contracts – practices that many today lament as corruption. For the Gülen movement, these benefits dried up by around 2011, a year before a major corruption investigation involving Gülen-linked firms, and two years before a larger corruption probe into Erdoğan’s inner circle.

The conflict between Erdoğan and the Gülen could very well be about corruption as the mainstream narrative suggests. But if this is the case, why is it then that members of the Gülenist media didn’t raise their voices earlier, but instead kept defending the government against earlier corruption allegations? An alternative interpretation of events, and one supported by the results in our paper, still include a role for corruption –  the difference being that the conflict is about the division of the spoils brought by it, rather than the sole moral concerns over the practice itself.