A lot of analysts, including me, think that Turkey's recent crackdown on PKK, combined with increased rhetorical pressure on the Kurdish-oriented Peoples' Democratic Party (HDP), is designed with domestic politics in mind. The presumption is that, the 7 June elections having produced a hung parliament, President Erdogan will call for early elections to be held in November. In this state of crisis, Turkish voters will rally to support his Justice and Development Party (AKP), giving him the parliamentary (super)majority he wants to increase his presidential power.
But will it actually work?
Erdogan is counting on a phenomenon called the "rally 'round the flag" effect. During times of crisis (or after a big military or diplomatic victory), citizen support for the ruler/ruling party increases. In theory, this translates into electoral support: more people will vote for the ruling party. Knowing this, rulers have an incentive to create or escalate diplomatic or military crises when they are unpopular or before an election. This is popularly known as the "wag the dog" effect, though the proper term would be a "diversionary conflict." In short, this is what we think Erdogan is trying to do.
But what do we actually know about rally effects? Here are some stylized claims that I believe accurately represent the state of research on the topic:
Rally effects are real and can be very large: I mention this mostly to note that there is nothing particularly "Turkish" about this. Rally effects have been observed in a variety of contexts, so it is reasonable to expect to see it in this case. These effects can be large; effects of over 10 percentage points have been observed. But they are also inconsistent, and the rally effects can be small or non-existent.
It is not as clear if rally effects translate into election results: This claim must be mitigated by the simple fact of evidence availability and timing. Most studies of the rally effect examine either leader approval or vote intention. Because elections are infrequent and not necessarily timed conveniently, it is harder to determine whether rally effects actually increase votes for the incumbent in subsequent elections. That said, history provides some guidance. While the Republican Party of George W. Bush made gains in the 2002 midterm elections, his victory in 2004 showed no particular rally effect. And the ongoing Iraq War and the 'surge' did not help his party's electoral fortunes in 2006 or 2008. Another study notes that the Falkland Islands War, often seen as a classic case of a rally effect, may have had a more limited effect on the electoral fortunes of Margaret Thatcher's Conservative Party (gated).
Rally effects can be short in duration: The general consensus is that rally effects "decay" quickly. Voters tend to have short memories, and today's new conflict quickly becomes tomorrow's endless war. The evidence suggests that the surge in leader approval tends to fade back to pre-crisis levels within 6-12 months. This effect is also reliant upon some degree of success: George W. Bush maintained a high approval rating throughout 2002 after the 11 September attacks as the invasion of Afghanistan went well initially. But Jimmy Carter's approval rating decayed throughout 1980 during the Iran Hostage Crisis as he was unable to resolve it.
Rally effects are mitigated by trust: One is more likely to rally around a country's leaders if one actually trusts them. That is the finding of a recent study (gated). If one does not trust the leader, then there may be no rally effect for that individual. It can even decrease trust if citizens blame the leader for the conflict. This caveat seems particularly relevant in the case of Turkey. Given the high degree of distrust towards Erdogan among many opposition supporters (especially CHP voters), the potential magnitude of any rally effect in Turkey is limited. Instead, one imagines that many (though certainly not all) CHP voters are more likely to accept the claim that Erdogan has manipulated this crisis for his own benefit.
Rally effects are mitigated by the domestic debate: One of the purported reasons for a rally effect is that opposition leaders and media self-censor their criticism of the leaders during a time of crisis. In effect, this suggests a form of cue-taking by voters; opposition elites rally in support of the country's leaders, so they do as well. This could be one reason why rally effects don't necessarily translate into election outcomes: while opposition leaders may be reluctant to criticize a leader during a crisis, they are more likely to do so during an election campaign (even if there is a crisis). This point seems relevant in Turkey insofar as CHP and HDP leaders are not muting their criticism of Erdogan. In anything, they are increasing their criticism of Erdogan and accusing him of leading Turkey into a misguided conflict. Again, this would lead to the prediction that those parties' voters will not rally to AKP.
Rally effects are mitigated by credibility: Evidence from the United States suggests that the rally effect is larger when there is UN Security Council support for the military action (see this gated study and this gated study too). Presumably, Security Council support provides credible third-party evidence that the use of military force is necessary. What does this mean for Turkey? On one hand, PKK is recognized as a terrorist organization by the US and the EU, and both have supported Turkey's right to defend itself against PKK publicly. So that would tend to increase the rally effect. However, criticism of Turkey's motivations or actions by other regional governments may have a mitigating effect. And, if the US or EU were to change their tunes and increase criticism of Turkey's conduct, then that could further mitigate the rally effect.
The Upshot: There is likely to be a modest rally effect producing an increase in support for Erdogan and AKP in the coming months. That effect is likely to last through possible November elections, particularly if the military can claim successes in defeating/killing PKK militants. However, that effect is likely only to occur among those who trust Erdogan, which means mostly supporters of MHP, a few of the more nationalist CHP supporters, and supporters of the smaller right-wing parties (e.g., Saadet). Howver, most CHP and HDP voters seem unlikely to swing to AKP as a result, muting the possible effect. Instead, the rhetorical and legal attacks against HDP seem more likely to yield results--but that would occur by demoralizing and dividing that party's support rather than by increasing AKP support.