There has been a long history of governments forcing banks to hold their debt in times of extreme need. This constraint, which Chari labels financial repression, is most often seen in wartime, when government spending is suddenly very high.
V.V. Chari presented a paper that asks if this is an efficient way to raise money, and if so, under what circumstances. He shows that if a government can commit to repaying its bonds, this is not an efficient way to raise government revenue. A government would be better off taxing capital directly. Forcing banks to hold government debt at below-market rates has the same distorting effects as a tax on capital, but also leaves the banks more collateral-constrained than before, thus reducing investment even further.
However, if the government cannot commit to repaying its bonds, forcing banks to hold government bonds may be a useful policy tool. Why? A government bond default has different effects depending on whether the bonds are held by individuals or by banks. Defaults on individual holdings are costless socially, but defaults on banks reduce capital investment. If the government forces the banks to hold its bonds, it places itself in a position where default is more costly. Thus, constraining banks to hold government bonds is a way governments can raise barriers and disincentives for future defaults.
The resulting pattern emerges: when there is a very high need for government spending—due to a major war, for example—the government will force banks to hold its bonds. This constraint is only needed when the outstanding supply of bonds is very high, and will typically be removed when government spending returns to normal levels.
Governments facing wide swings in spending are more likely to need to use this strategy. But if the swings are high enough, financial repression can help the government retain the credibility of its bonds.