Statistics show that enforcing the follow-on doesn't increase your chance of winning
In the past 25 years or so, teams have been reluctant to enforce follow-ons. A good landmark to separate eras is the 2001 Kolkata Test between India and Australia, in which VVS Laxman's 281 helped India win after following on, the last instance of this kind. That game caused teams to reconsider the value of making a team bat again, especially on pitches that may break up and become difficult for batting on day five, as the Kolkata one did. Also, since then, Test cricket has changed significantly. Scoring rates are higher, which means teams can bat in the third innings and still have time to bowl the opposition out in the fourth, and the schedule is packed, making preservation of bowlers a priority.

Since 2001, teams have enforced the follow-on just 56.60% of the time, as compared to 91% of the time in the 20 years before that. When the lead is less than 300, teams enforce the follow-on just 37.23% of the time. So Kohli's decision actually aligns with what the majority of captains would have done. Also, captains have realised that not enforcing the follow-on doesn't hamper your chances of winning. In fact, since 2001, teams that have declined enforcing the follow-on have ended up winning 91.12% of the time, while teams that have enforced the follow-on have won 84.44% of the time.

The Sydney Test starts just three days after this one ends
India's three frontline quicks have bowled almost 300 overs between them this series, and with the Sydney Test beginning on January 3, Kohli would not have wanted to risk any of them becoming fatigued. India have used a four-bowler strategy this series - unlike in South Africa and England, where they played five bowlers - so workload management is even more important. One might argue that it was worth going for a victory here to take a 2-1 lead even if it affected fitness levels for Sydney, but had India somehow not been able to complete the job despite enforcing the follow-on, they would have been left at 1-1 and with a tired bowling attack going into the decider.

Why lose the advantage of bowling last on a deteriorating track?
The pitch at the MCG is becoming harder to bat on. The bounce is becoming uneven, with several balls keeping low, and some balls turning out of the footmarks. All indications are that it will get even tougher for batsmen on days four and five as it breaks up further, so why not make Australia bat then?

Jadeja and the rough
The more Australia's right-arm quicks run in, the looser the rough outside the left-hander's off stump will become. And that will be something Ravindra Jadeja can exploit while bowling to the four left-handers in Australia's top five.

There's time to get 10 wickets even if it rains
There is some rain forecast on days four and five in Melbourne, but it's expected to be mostly brief showers that shouldn't cause entire sessions to be lost. There's plenty of time still left in the Test, so Kohli would be confident of getting a result even if there are some stoppages.