03 August 2022

The causes of the collapse of productivity in the UK Economy after the Financial Crisis



In an earlier blog I showed that after the financial crisis of 2007-08 real wages fell, while the amount of output produced by each hour worked in the economy rose at a far lower pace than earlier. Figure 1 below reproduces the data from that earlier blog.


Figure 1  The collapse of productivity growth in the UK economy post the financial crisis

Source ONS data: GDP per hour is obtained from the volume measure of GDP divided by the measure for total weekly hours worked annualised. Real hourly earnings are from data for average weekly earnings divided by weekly hours worked and deflated by the consumer price index.

 In this figure labour productivity is measured as the amount of GDP (that is the output of the economy) produced per hour worked. In the chart I show the periods of recession where GDP per capita fell. These recessions started in 1974, 1980, 1990 and 2007. In 2008, when real hourly earnings were at their pre-Covid peak of £18 per hour (this is in 2019 prices) real GDP per hour was £39.2. By 2019, the year before the pandemic, real hour earnings had fallen by 7 per cent to £16.8 while GDP per hour had risen 3 per cent to £40.5 – all in 2019 prices. What happened after the financial crisis was quite different from what followed earlier recessions where there appears to be no change in the steady grow of GDP per hour.


In this blog I want to answer the question as to why that output per hour worked grew so much slower than before the financial crisis. Why was this crisis so very different from earlier recessions?


Teresa May, during the 2017 general election campaign, said there was ‘no magic money tree’ when asked by a nurse why her pay had been frozen.  Well, was that true? A ‘magic money tree’ is an informal way of describing what has been happening to now rich countries for the last 200 years in that the amount of output each produces grows faster than does the labour to produce that output. Her more accurate answer, as the above figure shows, should have been that under previous governments there had been a ‘magic money tree’ but that the conservative dominated governments after 2010 had succeeded in killing it off.  This blog is about the mechanisms by which death was affected.


If you work hard and do your job, what if each year you produced more with no additional effort on your part. It sounds a bit unlikely but that is exactly what the UK economy did up to the financial crash. Economists call it total factor productivity (TFP). It is, quite literally, a measure of how much more output you get without more inputs into production (or at least any that are measurable).

 In Figure 2 I show how much this magical effect raised output in the UK from 1970 to 2021. Between 1970 and 2007, the advent of the Financial Crisis, TFP increased by some 60 per cent. Then, as the Figure shows, it fell abruptly and by the start of Covid it had barely recovered to its level at the start of the Financial Crisis.

Figure 2 More output but not more inputs

Source ONS data: Release of Productivity Overview, UK: October to December 2021. The figure for TFP in the chart is for the Total Market Sector.

 Now, of course, there are other ways, not magical at all, as to how you can become more productive. You may have more capital to work with or your travel time to work may be reduced. All these are additional inputs increasing output. One way of summarising how these additional inputs may help increase your productivity, that is how much you can produce, is by the capital labour ratio. That is how much capital is available to each worker. How this has changed since 1970 is shown in Figure 3.


Figure 3 How much more capital did workers have?

Figure 4 Changes in capital and labour inputs

Source ONS data: Release of Productivity Overview, UK: October to December 2021. Capital services and  labour hours are for the Total Market Sector.   

While we can see from Figure 3 that the amount of capital each worker has to work with fell soon after the Financial Crisis, we need to know how this changed occurred and that we show in Figure 4.

Two changes occurred after the financial crisis which were quite different from earlier recessions. The first was that capital services grew much more slowly than before. The second was that, in dramatic contrast to earlier recessions, labour supply started to grow. The result was the decline in the capital to labour ratio.


We can now answer the question we posed in the title to this blog as to why labour productivity growth collapsed after the Financial Crisis. Labour productivity growth depends on two factors. The first is TFP (Total Factor Productivity) which is the magical factor producing more output with no more inputs. While this fell marked immediately following the crisis it then started to rise again. The second factor is the capital to labour ratio which we showed in Figure 3 fell significantly after the financial crisis. It is this fall which is the most important factor explaining the failure of labour productivity to rise by more than very modest amounts.  


Labour productivity is not the focus of political debate. Indeed, outside the columns of relatively specialised newspapers it is not mentioned at all. That is regrettable as it is the collapse of that growth in labour productivity that underlies all the UK’s present economic woes. It explains the low wages many workers face, it explains the inability of the government to fund adequately the NHS and social care, it explains why at the end of a decade of ‘austerity’ there was still no budget surplus and the advent of the pandemic ensured that to address it the result has been a public sector deficit far larger than when ‘austerity’ began. It drives real earnings. It is how all of the painful political choices can be avoided. As real incomes rise taxes will rise even if tax rates do not change, there will be more income for workers and more funds for public services.


Put simply more people in the UK have been working so output has been rising but the amount of output each produce has not been increasing – the end of the magic money tree. This shows up in Figure 5 below where, following the financial crisis, after a short term fall, output continued to grow while output per person did not. As Figure 5 also shows this had not happened after any earlier recessions.


Figure 5 How output grew and productivity did not after the financial Crisis

Sources: As for Figure 1.

While the above analysis shows how the decline in productivity has been affected, the results clearly poses two more fundamental questions. Why did labour supply start rising and investment start falling after the recession caused by the Financial Crisis, while neither had happened in earlier recessions? To that I will return in a future post.
















06 June 2022

What happened to inequality under conservative and Labour governments since 1979?


Many people, I think, would answer that question by saying that it goes up under both. But does it? The answer depends on how we think about inequality. Inequality has been generally falling in the UK since the end of the Thatcher premiership when we measure inequality, as is most commonly done, by the Gini which takes a value of 0 if all incomes are equal and a value of 1 if one person has all the income. This fall in inequality is true if we take market incomes, that is what households get from their wages and self-employed incomes, or if we take equivalised disposable income. The latter allows for cash transfers to households and the taxes they pay and makes an allowance for household size.  I show this in Figure 1 below which tracks the Gini for the period from 1979, the start of the Thatcher premiership to 2019 which is the most recent data I have. Clearly post-Thatcher the Gini for both measures of income has tracked down, albeit with a marked upward spike with the most recent conservative government. This can be seen more clearly if we extract the numbers for the Gini for the periods of Conservative and Labour governments as I do in Figure 2 below. 

Figure 1 The Gini as a Measure of Inequality

Source: ONS data: Average household incomes, taxes and benefits of all households by decile group.

Figure 2 Conservative and Labour Governments Compared

Source: As for Figure 1.

 Two points stand out from the charts. The first is that inequality of equivalised disposable incomes is far lower than that of market incomes, and this is true over the whole period. All governments, both Conservative and Labour have used tax and benefit policies to reduce the level of inequality delivered

 Two points stand out from the charts. The first is that inequality of equivalised disposable incomes is by the market. The second point is that the premiership of Margaret Thatcher increased inequality both in market incomes and in equivalised disposable incomes. While, as I have noted, both measures have trended down they have not returned to the level before Thatcher’s first term.

Now the Gini is hard to interpret. Another way of presenting the information that underlies the Gini is by what is called a Lorenz curve. This shows how much incomes households get in percentage terms. In Figure 3 I show a Lorenz curve for equivalised disposable income for 1979 and 1991, ie the beginning and end of Thatcher’s government. The way to read the chart is to begin with the 45-degree line (in green) which is perfect equality if we think in proportional terms. It says the 20 per cent of households get 20 per cent of incomes, that 30 per cent of households get 30 per cent of incomes etc. So, the further is the line plotting actual percentages from this 45-degree line the more inequality there is in the Gini sense of the term. So, we can see from Figure 3 that this Lorenz curve shifted out quite noticeably over the period of the Thatcher premiership. In Figure 4 we present the same curve for the period afterwards.

Figure 3 The Thatcher Premiership

Source: As for Figure 1.

Figure 4 Lorenz Curves Post Thatcher

Source: As for Figure 1.

 Now what is striking in Figure 4 is that, if anything, there is a move inwards in the Lorenz curve from 2019 compared with 1991. True, very slight, but no evidence here of increasing inequality in the Gini sense of the term.

 However, the Gini, while commonly used, is not how people think about inequality.  The Gini is a relative measure, it is based on the proportions of incomes going to households as we have seen in the Lorenz curves above. If the incomes of everyone doubled the Gini would not change as the proportions would not change. But the gap between the richest and the poorest would have got a lot bigger. It is that gap that most have in mind when they talk about rising inequality. Telling them about the Gini cuts no ice.

What they have in mind is the differences been the top and bottom of the income distribution. There are two ways of looking at these differences. One is to look at the ratios between the top and the bottom (see Figure 5), a second is the gap between the rich and the poor (see Figure 6).


Figure 5 Ratios of incomes between the rich and the poor

Source: As for Figure 1.

Figure 6 How much do the rich get relative to the poor

Source: As for Figure 1.

 It is the difference shown in Figure 6 that people partly have in mind when they talk of increasing inequality. Between 1991 the gap in household incomes between the top decile and that of the bottom decile, increased from £57,863 to £88,785 (both numbers in constant 2019 prices), an increase of over 50 per cent. What they also have in mind is the levels of income for the poor and the rich.

In my last blog ‘The assault on the poor by the current conservative government’ I documented what had happened to the levels of income in the poorest decile, see Figure 7 below. What these 

Figure 7 Incomes of the poorest

Source: As for Figure 1.

policies were that had such a catastrophic effect on the incomes of the poor is well documented in an article first published in the Guardian on the 1st June by John Harris https://www.theguardian.com/politics/2022/jun/01/the-decade-that-broke-britain-the-disastrous-decisions-that-left-millions-in-a-cost-of-living-crisis.

In Figure 7 I show the incomes of the top decile. In the period since 2015 when the incomes of the bottom decile fell by some £3,000 (in 2019 prices) that of the top decile increased by some £10,000 (in 2019 prices).

Figure 7 Incomes of the richest

Source: As for Figure 1.

 An overview

So, what is the answer to the question posed by this blog: What happened to inequality under Conservative and Labour governments? If we focus on the Gini for equivalised disposable income this fell under Labour and rose markedly under the Thatcher and current Conservative government. If we focus on how the top decile have fared relative to the bottom, on this measure inequality remained unchanged for the labour government but nearly doubled for both Conservative governments.

 If rather than focus on gaps, we focus on the level of incomes of the bottom decile this increased by 34 per cent under the Labour government, while under the current Conservative government their incomes fell by a similar percentage – back indeed to nearly their level in 1979. Those, often on the left, who think Labour policies under Blair were Thatcherite, could not be more wrong.   

 A note on the data

 As indicated under the Figures the source of this data is from the ONS which provides data for the incomes across the deciles of the income distribution for each year, the most recent data being for 2019. The ONS data is drawn from household surveys. Many economists think that using such a source is inadequate for capturing the full extent of inequality in an economy. This is because the very rich are so few they will not be captured by surveys. This has led to the use of tax data, where the super-rich will appear, and to combine such tax data with surveys to give a more accurate account of inequality. Thus, the data used in this blog may understate inequality. In a future blog I will review some work which has combined tax with survey data. I will also examine how far the incomes of the super-rich exceed those of the top deciles shown in Figure 7 above. Such work does not affect the picture shown above for the levels of incomes for the vast majority of households.


26 May 2022

The Current Conservative Party’s Assault on the Poor



As is well know the period since the election of the coalition/ conservative government in 2010 has seen no rise in average real wages measured either on a weekly or hourly basis as I show in Figure 1. In 2019 real hourly earnings were 6 per cent below their level in 2010, while real weekly earnings were 3 per cent lower. The difference is due to people working more even though their hourly earnings were not rising.

Figure 1 Hourly and Weekly Earnings 1979-2019

Sources: ONS Data. Employee earnings in the UK and LFS: Total actual weekly hours worked.

So, on average over the period from 2010 to 2019 real wages were not rising but what of the poorest, what was happening to their incomes? Neither wages nor average incomes tell us about the incomes of the poorest. The best measure is their disposable income, that is their market income, what they earn from wages, self-employment or non-labour sources, plus the befits they receive less their taxes. In the Figure below I show a measure of disposable income that makes an allowance for household size -, incomes from a two-person household cannot be directly compared with a one-person household or ones with children. This is termed equivalised disposable incomes. For the bottom decile (that is the poorest 10 per cent of the population) equivalised disposable incomes declined by 48 per cent between 2015, the election of the conservative government, and the last full year before the pandemic. Figure 2 below shows that this unprecedented fall took this measure of income back to below the level of 1991 and only marginally above that of 1979. 

Figure 2 Equivalised disposable incomes for the bottom 10%

Source: ONS data: Average household incomes, taxes and benefits of ALL households by decile group

What explains this dramatic fall? First, it is not due to falls in market incomes. As is shown in the Figure below it increased by 17 per cent from 2015 to 2019, helped in particular by the real rise in the minimum wage.

Figure 3 Market Incomes

Source: As for Figure 2.

The answer is in Figure 4 below. On the left we see that the cash benefits to supplement earnings were slashed by nearly 25 per cent, while of the right taxes and national insurance contributions increased more than two and a half times. 

Figure 4 Benefits and Taxes of the Bottom 10%

Source: As for Figure 2.

The poorest were the subject of an assault on their incomes without historical precedent as their cash benefits were cut and their taxes increase, both government decisions. For those inclined to believe this is what Conservative governments always do it needs to be noted that the Thatcher government did nothing similar (see figure 4 above). Why was the period after 2015 so different?

The Guardian reported in July 2015 that  ‘George Osborne has unveiled an unexpected boost for low-paid workers, introducing a compulsory national “living wage” in a budget that also contained an assault on many of the measures introduced by Labour to tackle poverty.

In a final flourish in the first Conservative-only budget in 19 years, the chancellor portrayed the Tories as the party of working people by declaring that the living wage will be introduced from next April at a rate of £7.20 an hour for people over the age of 25.’

This rhetoric continues to this day. What the data shows is that the benefits to the poorest introduced by Labour are essential to prevent falls in disposable comes for the poorest. George Osborne may well have believed his argument that wages could replace benefits, he was imply wrong. 





25 May 2021

Falling hourly earnings and increasing employment


The fundamental puzzle of the UK economy post the financial crisis


One of the many ironies of the Brexit debate in the UK was the fact that over a period when the problem of immigrants taking British jobs was the focus of public concern the number of people at work in the UK had risen to previous unknown heights. In this note I document how dramatic have been the changes of the last decade and how different was that job growth compared with the recessions prior to that of 2002-08. I show that the British economy has been good at increasing earnings or increasing the number of jobs, it has not proved good at doing both.




Just how different has been the pattern of job growth in the UK since the financial crisis, of 2007 to 2008, as compared with the previous recession in 1991 is shown in the Figure 1 below.


Figure 1 Employment in the UK: 1970 - 2020

Not only was the decline in jobs following the financial crisis modest relative to the fall after the 1991 recession but the recovery was far more rapid and the numbers of people in work grew rapidly after 2011 until the advent of the pandemic in 2020. The decline in employment with the lockdowns in 2020 is very modest as the ONS counts furloughed workers as employed on the grounds that they still have jobs and are temporarily absent. That does means that for 2020 that the numbers classified as employed are not the same as the numbers working, a point to which I return below.


In fact, the left-hand chart understates the magnitude of the achievement of the UK economy in creating jobs. Employment will grow because the population of those of working age increases. Indeed, over much of this period immigration increased. So, we want to know how many jobs there were relative to the population of working age. This is called the employment rate. It tells you what percentage of those in the population who could work found jobs.


In the three decades from 1970 to 2000 the employment rate fluctuated between 73 and 66 per cent. It reached its nadir of 66 per cent in the early 1980s after the first of the Thatcher recessions in 1981. However, this recession simply continued a downwards trend in the employment rate which can be dated from the mid-1970s. The second Thatcher recession of 1991 saw a sharp fall in the employment rate reducing it by some 4 percent points. The 1990s saw some recovery but by 2000 it was still below the level of the mid-1970s. The proportion of those in work since the 1970 had simply fluctuated between 73 and 66 per cent, after 2010 it took off.


Contrast the pattern before 2000 with the pattern afterwards. Compared with earlier recessions from 1970 to 2000 the financial crisis had a very modest effect on the employment rate, a decline of only 2 percentage points, while within two years of the crisis the employment rate started to increase and rose continually for the next five years to reach 76 per cent by 2019/2020, far exceeding the highest rates achieved in the three decades from 1970 to 2000. 


Hours worked


I noted above that for 2020 employment levels did not reflect hours worked. If we want to understand what determines the labour supply to the economy we need to consider not simply the numbers employed but how many hours they work. Indeed, it is hours worked that is the relevant measure of how much labour is supplied to the economy. In Figure 2 below I show how average and total weekly hours have changed from 1970 to 2020. We can now see the effects of the Covid pandemic on labour supply with the collapse in total hours worked in the right-hand chart of the Figure. Between its nadir in 1983 of 795.3 million hours it had reached 1 billion hours in 2019, the year before the pandemic decimated labour input into the economy. As the left-hand chart shows from 1983 to the financial crisis of 2008 average hours fell so over that period the number of jobs was rising faster than the labour supply.


Figure 2 Labour supply: Average and Total Hours Worked

What we need to explain


We can now clearly state what needs to be explained. Why after the most severe recession in post war UK history in 2008 did labour supply grow in a way it had never done before?


Figure 3 GDP and Earnings

There were two other major differences between the post Financial crisis and previous recessions which are shown in the left-hand chart of Figure 3 above. The first was that the growth rate of productivity, measured as GDP per hour worked, fell in a sustained way. The behaviour of real wages was even more dramatic. For the five years after the start of the financial crisis in 2007 the growth rate of real hourly earnings was negative. As the charts show this fall was unprecedented. The contrast with the first of the Thatcher recession is striking. In the early 1980s neither the growth of productivity nor real earnings were ever negative.


The cumulative effects of growth rates are hard to see so in the right-hand chart of Figure 3 I show the implication of these growth rates for the level of productivity and for real hourly earnings. The results of these very different patterns of growth rates implied that the level of productivity grew very little while real hourly earnings fell such that by 2019, on the eve of the Covid pandemic, they were back to their level of 2005, two years before the financial crisis.


So, what we need to explain is very puzzling? Falls in labour productivity would normally lead to a fall in the demand for labour and a fall in real earnings. While the demand for labour may well have shifted down clearly the supply effect was more than sufficient to offset this as the result was an increase in hours worked. Economists normally assume that to induce more employment you need higher earnings not lower ones. So, what explains this link between falling hourly earnings and more working hours?


There are broadly three answers to that question. One is that more workers are joining the labour force in lower wage areas so that it is not true the more hours are forthcoming with lower wages but that the proportion of low wage jobs is increasing. The second is that wages are indeed falling across all occupations and this is the result of falling productivity in those jobs. The third is that sectoral occupational change is occurring and workers are shifting to lower wage sectors.


Which of these explain the pattern we observe will be the subject of another blog posting.

12 March 2021

Why is the labour party so ashamed of its last government?


In a YouGov poll conducted of labour party members in early 2020 Tony Blair was by far the most unpopular of those who have led the labour party since 1964. Only 37 per cent of labour party members had a favourable view of him as compared with 71 per cent for Jeremy Corbyn and 70 per cent for Ed Miliband.

At the time of the most recent labour leadership election none of the candidates was willing to praise the record of his government when compared with the Tories. A similar silence emanated under Jeremy Corbyn at the last election. Surely the first time in democratic politics in this country where a party has tried to get elected on the basis that it was very bad last time but would do much better this time. It did not work, which might well not be thought a surprise.


Now the issue that probably dominates in the minds of labour party members is the Iraq war. A deeply divisive and, even at the time, a very unpopular policy. But the critique is wider than that. The argument is often heard that New Labour was simply a continuation of the neo-liberal free market individualist policies instigated by the Thatcher premiership. Indeed, it appears many members of the labour party draw no distinction between the outcomes for the two periods in the area of economic policy. There are those who argue for a more positive view of new labour, for example:


but even those willing to defend the Blair and Brown governments seem unaware that its economic policy effectively reversed the pattern of income growth under the Thatcher government.


To demonstrate this I present the data for rises in average household incomes across the distribution for the periods of the Thatcher and New Labour governments, the latter including Gordon Brown’s premiership and thus the years of the financial crisis. In Figure 1 I show the total increase based on the trend growth of income in the period. While this captures what happens on average over the period it may mislead as to the differences between the beginning and end points of the premierships. So, in Figure 2 I compare the average incomes for the top and bottom decile at the beginning and end points of the two premierships. In Figure 2 I also show how the median household fared and the size of the gap in incomes between the top and bottom deciles.


It would be little exaggeration to say that the New Labour years were the exact opposite of those of the Thatcher premiership. The grey line shows increases in market incomes, while the red lines show increases in equivalised disposable income, both are total percentage changes over the period of the respective premierships. For both you can see how those increases differed across the distribution from the bottom decile (that is the bottom 10 per cent) and the top decile (the top ten per cent).


Figure 1  Thatcher and New Labour Compared

Market incomes are the wages of employees and the self-employed that the market delivers. Disposable incomes are those available to households after transfers and income taxes and national insurance. Clearly how well off is a household depends on how many adults and children there are in the households. What equivalised means is that an adjustment has been made for such differences. Thus, equivalised disposable incomes are our best means of comparing how households across the income distribution have fared.


As Figure 1 shows for the Thatcher years increases in disposable household incomes for the bottom 30 per cent of the households were derisory, averaging about 6 per cent - that is a total increase over eleven years. From the fourth to the top decile there is a steady march up with the higher the household income the higher the percentage growth rates.


The contrast with New Labour is stark. The bottom 30 per cent of the household had increases that compare favourably with those of the top 30 per cent for both market and equivalised disposable incomes. Indeed, while the bottom 30 per cent of households under Thatcher saw falls in their market incomes under New Labour they had massively larger increases than those at the top of the distribution.


Percentage changes are a bit abstract so what do these percentages means for actual incomes across the distribution. Further what do they mean for the gap between the rich and the poor across these years of Thatcher and New Labour economic policy. Those questions are answered by the data in Figure 2 which shows the second way of presenting the data.  


Figure 2 Household Incomes across the Distribution

The top and bottom left charts in Figure 2 show increases in average annual household incomes (all at constant 2019 prices) for the top and bottom deciles respectively. The years shown cover the start and end of the Thatcher premiership and the start and end of the New Labour ones. The message is the same as from Figure1 but now we can put numbers on those income differences between Thatcher and New Labour.


Considering equivalised disposable incomes under Thatcher incomes in the top decile increased from £ 36,150 to £64,976, a rise of 80 per cent. In contrast incomes for the bottom decile increased from £6,748 to £7,113, a rise of just over 5 per cent.


The comparable numbers for New Labour are an increase from £68,452 to £89,596 for the top decile, a rise of 31 per cent. For the bottom decile the increase is from £7,681 to £10,256, a rise 33.5 per cent. This is less than that shown in Figure 1 as there was some slowing of the growth rate for the bottom decile during the later New Labour years. However, both approaches point to the reversal of the pattern shown during the Thatcher years for the New Labour ones.


Clearly the bottom and top deciles are the extremes of the distribution. In the top right I show the pattern for the median households. For these the two period saw virtually identical increases if we look at changes between the beginning and end points from Figure 2. If we look at trends, as used in Figure 1, New Labour does slightly better. 


In the bottom right of Figure 2 we show the gap between the top and bottom deciles. It is indeed this gap which people often mean when they talk of increasing inequality. As the chart shows it is true that the gap between the top and bottom deciles has been growing both under both Thatcher and New Labour. But that reflects, not the failure of New Labour, but the assault by the Conservative government on the poor such that by1997 the level relative to the rich was so low that even the much  better performance for the poor under New Labour was insufficient to close the gap.


New labour represented a revolution in the success of economic policy toward the poorest parts of society. It is unsurprising that the Tories remain silent on this success. That they are joined in this silence by members of the labour party is a tragedy for the people that New Labour was so successful at helping.


Notes on the data used in the charts:

The data is provided by the ONS. The most recent data can be found at:


The data in the Figures is drawn from past publications of this data which seeks to provide estimates of the redistributive role of taxes and benefits on household income and inequality. The data are from our Living Costs and Food Survey (LCF), a voluntary sample survey of around 5,000 private households in the UK.


These statistics are produced at the individual level meaning, for example, that income quintiles are derived by ordering people, rather than households, on an equivalised household disposable income basis. This method is consistent with the statistics reported in Average household income, UK: financial year ending 2019 and Household income inequality, UK: financial year ending 2019, and it ensures the variance in household size across the income distribution is better accounted for.


My book – The Poor and the Plutocrats (to be published by Oxford University Press on March 24 2021) - uses this data source, presenting quintiles rather than deciles, to show how the US compares with the UK in terms of how levels of income across the distribution have changed. The book also provides more details on the data used.

The causes of the collapse of productivity in the UK Economy after the Financial Crisis

    In an earlier blog I showed that after the financial crisis of 2007-08 real wages fell, while the amount of output produced by each ho...