Stephen King: Roosevelt's lesson... a decisive act to break the psychology of depression

What should you do if your banking system doesn't work? Some will doubtless celebrate, arguing that banks are the source of all monetary evil. Others will panic, worrying about the onset of another Great Depression. Policymakers, though, should do neither of these things. They need, instead, to find a way to make the financial system function again.

Banks, after all, are supposed to link the interests of savers and investors. They provide the glue which allows economies to allocate capital through time. Without banks – or some other form of financial intermediary – modern-day economies would implode. You've only got to go back to the Great Depression to see what happens when a banking system melts down.

Why is the system in such bad shape? Many reasons spring to mind, ranging from collapsing housing markets to pro-cyclical fair-value accounting and dodgy mortgage-backed securities, but it's the loss of trust which is the biggest single problem. Banks have a habit of lending to each other in the so-called interbank market. They'll do so, though, only if they're confident that they'll get their money back. Recently, trust has been in short supply. Banks circle each other suspiciously, unsure of the hidden dangers associated with lending to their counterparts. We're seeing a modern-day wholesale version of the bank run. It's not so much that customers are withdrawing their deposits (although, to a degree, they are). Instead, banks are simply refusing to lend to each other.

Arguably, matters have been made worse by policymakers who have adopted a piecemeal approach to bailouts. Lehman Brothers was allowed to fold but AIG was saved, leaving some stockholders penniless but others a bit better off. Washington Mutual's rescue left its creditors severely out of pocket, while, at the time of writing, Wachovia was to be sold either to Wells Fargo or, with the help of taxpayers' money, to Citibank. In the absence of systematic government policies, banks are finding themselves playing a game of financial Russian roulette.

How, then, is trust to be restored? The Japanese pretended during the 1990s that there simply wasn't a banking problem. Bad debts may have hung over the system like a dark cloud but, on the assumption that successive Japanese governments would never allow the system as a whole to implode, no one other than equity investors seemed to mind too much. (Japan eventually ended up guaranteeing deposits, but this was easier to do when banking systems elsewhere were stable: today, as opprobrium towards Ireland's deposit guarantees increases, it's not so simple.) Certainly the counterparty risk which has pushed money market rates to remarkably high levels in the US and Europe over the last 12 months was not really a feature of the Japanese financial landscape in the 1990s, even if foreign banks demanded, from time to time, a modest "Japan premium".

In the early 1990s, the Swedes partially nationalised their banking system. At the limit, nationalisation removes counterparty risk because it removes counterparties (forced mergers would do something similar). By the same token, though, it also removes competition. While there may be a case to limit competition through, for example, more effective regulation, an absence of competition altogether would surely not be in the public's interests (indeed, state involvement in Sweden's banks diminished rapidly as the 1990s progressed).

As for the US, the Troubled Assets Relief Program (TARP), eventually passed at the end of last week, is closely modelled on the Resolution Trust Corporation, the public body created at the end of the 1980s to deal with failed thrifts. On this occasion, though, the plan is to deal with bad assets, which are randomly spread throughout the banking system. Their removal, it's hoped, should allow banks to lend to each other freely again, thereby limiting cash hoarding and, hence, easing the credit crunch.

The plan has its merits, but it also has some obvious weaknesses. At what price should the bad assets be purchased? Too high a price and taxpayers will revolt. Too low a price and banks simply won't sell the bad assets, leaving counterparty risk alive and well. The plan leaves two fundamental problems outstanding, namely the still-high level of house prices relative to income and a heavy household debt burden. And, unlike the savings and loans crisis of the 1980s, capital market failures today are on a global scale: can the taxpayers of any one country, even one as big as the US, offer enough support to restore trust all over the world?

The nuclear option is to bypass the banking system altogether. In effect, this is what President Roosevelt did in 1933 and 1934. If banks are unwilling to lend to each other and, thus, unable to lend to non-banks, governments can elect to turn themselves into banks. After all, during a banking crisis, people are typically happier to hold cash and government bonds than anything else. The desire to avoid financial losses absolutely dominates, and the best avoidance scheme is to rely on the taxpayer: government bonds are attractive in these circumstances because governments in the developed world, at least, will always coerce their taxpayers to repay creditors. During banking crises, governments can, therefore, raise funds relatively cheaply.

They can do even more. They can also choose to monetise government debt, by selling bonds to the central bank rather than to the public. By doing so, governments can flood an economy with money. At a stroke, the perceived monetary shortage which leads to hoarding, bank runs and a financial climate of fear can be removed. Moreover, with excess money, other assets suddenly look more interesting, at least in nominal terms.

If, though, the banking system is in a mess, how does a government get money into the economy? The answer is simple: either tax cuts or, even better, big increases in government spending. In the first two years of the Roosevelt administration, government spending increased by 80 per cent. If ever there was a decisive act designed to break the psychology of depression, that was it.

If, though, this is the ultimate "do", what about the "don'ts"?

First, central banks should not defend their independence at all costs. This, after all, is what the Federal Reserve did during the Hoover administration at the beginning of the 1930s. The approach was a hopeless failure. During banking crises, central banks lose their power. To restore it, they need, and should ask for, fiscal help.

Second, each central bank should, ideally, speak with one voice. Better, in my view, to show strong leadership than to advertise publicly a collection of disparate views which can only sow the seeds of doubt throughout the financial system. After years of success in highlighting the nuances of the economic debate, we're now seeing the downside to the Bank of England's committee system.

Third, under no circumstances should countries resort to capital market protectionism. The Irish government's offer to underwrite deposits in Irish banks (and, hence, to protect the Irish banks' interests) is an unfortunate precedent (it would be far better if all countries were to offer deposit guarantees simultaneously, but that hasn't happened). It's reminiscent of the Smoot-Hawley tariff in 1930, designed to protect the interests of American exporters but, ultimately, a contributor to the subsequent collapse in world trade. Others were forced to launch their own "beggar-thy-neighbour" policies, contributing to a global economic collapse and, tragically, fanning the flames of fascism. We don't want to go down that route again.

Stephen King is managing director of economics at HSBC.