There seems to be a fair chance the Office for Budget Responsibility鈥檚 view of the housing market may be badly awry.

Why would that be that bad?

1. Because it potentially supports a complacent attitude among policy makers towards problems within the housing market.

2. Because it may well leave the chancellor a few billion pounds light on stamp duty in a few years time.

But before galumphing into why the Office for Budget Responsibility (OBR) may be wrong it鈥檚 worth making a couple of points.

Forecasting is in many ways a mugs鈥� game. The chances are that you鈥檒l be wrong, even if your reasoning is broadly correct. Furthermore, if you do end up predicting the right result there鈥檚 a high chance you got there despite making duff assumptions.

That doesn鈥檛 make forecasting pointless - far from it. You need something to guide the way in murky times and someone has to do it.

So when I took a sharp intake of breath as I looked at the latest forecast from the OBR in the light of trends in the market, I did so with some sympathy rather than incredulity and indignation.

Also, I鈥檓 not saying I鈥檓 right and it鈥檚 wrong. Frankly no one 鈥渒nows鈥� the future. And it may be that someone scrutinising this blog may find my assumptions are skewiff or that I鈥檝e messed up my numbers.

Anyway, to the main point.

I have a horrible sense that the good folk at OBR remain over-optimistic about the performance of the housing market, despite growing evidence that we may be in for a more protracted slough than their figures recognise.

And my crude calculations suggest it is not unreasonable to expect a hole of 拢2 billion or more a year in the residential stamp duty line.

While I鈥檓 a fan of the OBR, it has been over optimistic before on stamp duty (see ) and found the need to adjust at a later date.

Anyway, to unpick the problems it鈥檚 worth starting with the OBR鈥檚 position. It is summed up in the first graph which shows transactions returning toward what is assumed to be the trend.

The supporting text states: 鈥淥ur medium-term projection of property transactions are conditioned on a return toward the long-run average rate of turnover in the housing market 鈥� which implies owner-occupiers move once every 19 years. The depressed current level of transactions is consistent with owner-occupiers moving only once every 30 years. We are forecasting a relatively slow return to the long-run average, which is not quite complete by the end of the forecast period. By this time, and despite strong growth rates, the level of property transactions is still around 20 per cent below the pre-crisis peak.鈥�

That seems fine. Taking a revert-to-trend approach is common in forecasting. It鈥檚 a reasonable approach if your assumptions on trend are correct and there鈥檚 no obvious reason to think there鈥檚 been a significant shift in the market.

On the back of its assumptions over residential property transactions OBR then assumes/calculates that residential stamp duty will return to and head above the pre-boom peak of 拢6.7 billion for the financial year 2007/08 to 拢8.5 billion in the financial year 2016/17, according to table 2.6 of the .

That鈥檚 where alarm bells ring. It sounds very optimistic (see the second graph). How will the Treasury raise 拢1.8 billion more on 20% fewer transactions?

I am not the only one to raise an eyebrow at these projections. For instance, head of UK residential research at Knight Frank points to the optimism inherent in the OBR forecast in her post-Budget blog.

Let鈥檚 look at the numbers by comparing the situation in 2007/08 with that in 2016/17, when OBR forecasts there will be about 27% more revenue in residential stamp duty.

The determinants of the stamp duty take are, broadly speaking, the number of transactions, their value, the levy on those transactions and the ability of the tax authorities to collect the tax and close loopholes.

So starting with 20% fewer transactions isn鈥檛 helpful to the OBR case, but what about the value of those transactions?

Looking first at house prices, OBR鈥檚 own forecast suggests they will be roughly at the same level as at the peak of late 2007 in nominal terms. So broadly equal across the two periods we are considering, but it could be that the proportion of high-value homes within the mix of those sold will be higher in 2017 than in 2007.

Now let鈥檚 consider the changes to the stamp duty regime between 2007/08 and the latest Budget rates. I have tried to be 鈥済enerous鈥� on the impact of these changes when drawing on the estimates within the Treasury and OBR documents. For reference, the figures are based on post-behavioural estimates.

Comparing the regime in 2007/08, and what is currently proposed for 2017, the main differences (there are other minor differences) I see are:

The introduction of a 5% rate on homes priced above 拢1 million introduced in Labour鈥檚 last Budget. Let鈥檚 estimate it might be worth an extra 拢300 million by 2017 based on the 拢230 million uplift estimated for 2012/13 in the Budget figures.

The 2011 Budget brought a tax avoidance pledge. This might raise 拢50 million a year more.

But that Budget also brought relief on bulk purchases. This might mean about 拢150 million a year less.

The latest Budget brought us 7% on 拢2 million-plus homes, which the documentation suggests should net an extra 拢300 million annually by 2017. And we had the loophole-closing measures on company schemes to avoid stamp duty on house sales. This move should net 拢75 million.

Add and subtract accordingly and I get near as damn it 拢500 million.

Now, assuming there鈥檚 not miscalculation on my part or other significant factors I鈥檝e missed, the scale of those changes doesn鈥檛 provide comfort in the OBR forecast, given that transactions are down.

The mix of homes sold would have to significantly shift to hit the 拢8.5 billion estimate for residential stamp duty.

In the OBR鈥檚 favour there is a shift in the mix of homes in the market underway, as sales activity moves both southward and towards richer buyers and sellers.

However, this shift comes with an obvious constraint on the numbers of transactions and throws into doubt the assumption that 50% more homes will be traded in 2017 than currently.

So it all looks to me a bit shaky. But this is where I have to go a bit further and question strongly the assumption of trend in transactions made by OBR.

Look at the final two graphs. This is an update by me of data provided for the report 2008 by Richard Donnell of .

Call me Mr Glum, but I see a downward trend in transactions from the late 1980s when I look at the first of those two graphs.

More worryingly, looking at the response to the previous heavy housing recession in the 1990s (see the final graph), it is tempting to suggest we may have to consider hysteresis, whereby after the shock of a housing crash the system is not returning to its historic position. This would throw into doubt the revert-to-mean assumptions implicit in the OBR forecast.

This may seem fanciful and, while I could postulate in pub-chat fashion an argument to support it, I鈥檓 not sure what might actually lie behind this effect. But, for me, it鈥檚 equally as plausible 鈥� if not more so 鈥� as the OBR forecast of a rising trend in transactions.

So unless I have missed something, from where I sit the OBR forecast seems optimistic with the spread of risk very heavily on the downside.

Now I would not blame the OBR for being wrong, if that were t prove the case. But perhaps I would have been more comforted if there were a bigger health warning on that particular bit of the forecast.

Because it does seem to me quite possible, if not probable, that George Osborne, or his successor, will have to find at least 拢2 billion if not more to fill the hole left by lower than forecast residential stamp duty receipts.