Interwar Unemployment Flashcards
Trends in IW unemployment (7)
- Hatton & Thomas (2013) – picture seems to be that NAIRU increase in 1920s (due to TUs, hours, staples) & stayed there in 1930s (with higher cyclical & LTU – Depression + hyst.)
- Take with grain of salt because sources vary tremendously
o “Board of Trade” – only in unionised sectors, which tend to be industrial, hence cyclical
o Census enumerations – under-survey the u/e (reluctant to admit it) – most accurate
o U/e benefits – extensive (2/3 of earners), but misses agriculture/services/govt/high earners (understates u/e) - Boyer & Hatton (2002) construct new index for unemployment data (from Board of Trade)
o Board of Trade data underestimates u/e and overemphasises fluctuations
o Average u/e rate 1870-1913 is 5.8%; average in interwar is 10.9%
o u/e c. 8% 1923-29, then up to 17% by 1932, then back to 8.5% by 1937 – never actually lower than 7%, whereas in pre-war period, rarely exceeded 8% = persistence. - Solomou (1996) – very cyclical, severity increases over time & hysteresis are key features
- Highest in: Traditional staple industries (e.g. coal, cotton), N England, Wales, Scotland (In 1929, unemployment was 3.3% in South East, 18.1% in Wales), Older people.
- In 1920s, LTU was rare. Rose to 25% of unemployment by 1930s (Crafts)
Causes of High Unemployment
- 1919 Shock to Working Hours - (Broadberry 1986, 1990)
- Unemployment Benefits (Benjamin & Kochin, 1979)
- Trade Unions
- Structural Unemployment
- Demand-Side Explanations
- Exogenous shocks
- Hysteresis = Persistence (LT unemployment is a subset of this)
1919 Shock to Working Hours (Broadberry 1986, 1990) (4)
- Two main channels:
o During WWI, consumers saved (real C down 14% due to diversion of resources), increasing their purchasing power, so demanded more leisure (normal good), so less hours working.
o (Dowie, 1975) In 1919, the average working week fell from 54 hours to 47 hours – 8 hour day. - Nominal wage rigidity (Lennard, 2022) meant real hourly wage rate rose 13% on average.
- Not helped by contractionary monetary policy to raise the exchange rate in order to facilitate return to Gold at pre-war parity, putting downward pressure on prices = further increase in real wage.
- Created a “wage gap” (gap between labour productivity & real wages), sustained until 1940
o Unit labour costs, ULC = WL/PY = W/L ÷ P/Y; wage gap increased ULCs via real wages.
o Higher ULCs makes UK less competitive, reducing exports and hence AD contraction = demand-deficient unemployment.
o Represents a one-off shift in SRAS, explaining persistently higher natural rate of unemployment in interwar period from 2% to 8% (Dowie, Broadberry).
o Demand fluctuations explain variations about this level.
1919 Shock to Working Hours - HOWEVER (3)
However
- Solomou (1996):
o Broadberry deflates the nominal wage series with the RPI.
Key to hiring is product real wage, not consumption real wage.
Therefore, should deflate with the price of final output, i.e. GDP deflator.
o Doing this, wage gap is eliminated by 1922 – only a short-term transitory effect.
o If this explains all interwar unemployment, it has to occur via hysteresis.
- Hatton & Thomas (2010) – increases unemployment by 3% by 1922, but largely disappeared by 1929.
- Scott & Spadavetcchia (2011):
o Majority of lost weekly output from less hours was made up for (in LR) through increased hourly productivity (as working environments adapted).
o Reflects improvements in speed/quality of work, abolition of in-work rest breaks/periods of idleness due to bottlenecks & reductions in sickness/accidents.
o Reduction in working hours occurred in almost all other industrialised countries:
UK experienced lowest proportional reduction in hours over 1913-1920, except France.
UK real wage growth 1913-1920 was 6.9%, slightly above Australia/US, but much lower than European countries – relative competitiveness should not have suffered
o Major ‘social multiplier’ effects associated with reduction in hours – e.g. growth of industries associated with working-class leisure.
Unemployment Benefits (Benjamin & Kochin 1979) (4+3+Conc)
- 1911 National Insurance Act: contributory u/e insurance benefits available to 15% of the workforce (2.2m workers).
- 1920 Unemployment Insurance Act – increased coverage to 11m workers (workforce at 20.7m so 50%) & increased weekly benefits by 40%.
Benjamin and Kochin (1979): - Unemployment benefits are the primary cause of interwar unemployment – ‘volunteer army’.
- (Maki & Spindler, 1979) 3 mechanisms through which benefits create unemployment:
o SA –> SA’ people join L force without genuinely seeking employment (claimants).
o SE <– SE’ – effective labour supply falls as searches lengthen/people voluntarily leave jobs.
o D <– D’ – more willing to lay off workers during downturns and bear cost of employer contributions. - 3 key features of u/e benefits in this period that make them especially bad for u/e:
o Contributions unrelated to u/e experience.
o Not proportional to wages.
o Payable for short periods, especially if previously unemployed. - Present 3 pieces of evidence:
1) Time-Series Regression
o U = f(B/W) where B/W is the replacement ratio.
o Find correlation on the replacement ratio is statistically significant.
o Conclude that generous u/e insurance increased interwar u/e by 5-8pp.
2) Juvenile Unemployment (16-17 year olds)
o Harder for juveniles to get benefits (20-30 weekly contributions required) & paid at a lower rate (doubled at 18).
o Juvenile unemployment is much lower (5%) than average (15%).
o Running time series for juveniles, replacement ratio coefficient not significantly different from 0.
3) Female Unemployment
o 1931 Anomalies Regulations – increased tests for married women to get benefits.
o After this, lower u/e among married women relative to single & women vs. men too. - Conclude that without benefits, u/e would have been ‘at normal levels through much of the period’.
Criticisms of Benjamin & Kochin (5)
- Compositional evidence is flawed
o Eichengreen (1987) – little incentive to register as u/e if not eligible for benefits – this affected both juvenile u/e (throughout) and female u/e (after 1931)
o Hatton (1983) – juvenile u/e because they are more desirable workers, from even before WW1 (Beveridge)
o Broadberry (1983) – female participation is too erratic to conclude
o Therefore, BK79 rely on the regression - Hatton (2004) – replacement ratio used is inappropriate
o Measures benefits for adult male with a wife and two children
o Wage series is average wages for whole population
o Therefore B/W ratio biased upwards – using average, B/W = 0.27 vs 0.37 – so in fact, below subsistence for much of period, so no incentive to get benefits - Irish (1980) – trending is a problem
o Significance of α1 comes from fact that Ut and B/Wt are trended upwards, whereas log(Q/Q*)t is not
o Adding any trended upwards variable, such as population, makes α1 insignificant - Solomou (1996) – identification problems
o Hard to see it as a demand or supply function – replacement ratio influences labour supply, but output gap influences labour demand
o Therefore, isn’t a proper model – can’t distinguish between D and S shifts
o U/e and wages determined simultaneously, so causation runs both ways
o Hatton (1983) finds effect is via employers being more willing to lay off workers, not labour supply - Cross (1982) – endogeneity problem
o High u/e may mean that govt responds with policy
Unemployment benefits - alternative studies (3)
- Harris (1988) – living standards show u/e was not voluntary
o Uses data on average height of school children (proxy for malnutrition)
o Average height decreases in 1930s (shorter in 1933 than 1929)
o Also a negative relationship between local unemployment and children’s stature
o Therefore, u/e not voluntary – has an adverse impact on their children
o Rowntree - in 1936, 73% of u/e workers lived below poverty line - Eichengreen (1987) – secondary vs. household heads
o Considered individual level data from “New Survey of London Life and Labour 1928”
o Effect depends on elasticity of supply of labour – no relationship between replacement ratio for household heads, but there is one for secondary workers (23% of sample)
o Counterfactual u/e rate with 1913 benefit levels is 6.8-7.4% (vs actual of 8.6%) (for adult males), so not more than a fifth of actual u/e (ct 1/3 to ½ of BK79)
o Data limitations: truncated data (biasing influence upwards), focus on London (biasing influence upwards – little involuntary u/e) & absence of key variables - Hatton & Bailey (2002)
o Use individual level data for London between 1929 and 1931 – 30k workers
o Effect of replacement ratio is biased upwards by occupation and skill not being controlled for, leading to ‘spuriously positive effects’
o Carry out regression for adult males, juveniles & females
o Find that in all population cohorts, controlling for occupation and skill, the replacement ratio is insignificant – ‘no evidence that benefits induced u/e’
Voluntary unemployment assumption is unrealistic (5)
- Collins (1982) – relationship drops away
o Disaggregates calculations by industry for 12 – not statistically significant for 9
o u/e in industries not positively related to industry specific replacement ratio - Metcalf et al. (1982) – replacement ratio just as high post-WWII, but not same levels of u/e
- Eichengreen & Hatton (1988) – Belgian u/e increase much more rapidly in 1930s than GB, despite cuts in the rates of benefits and increasingly stringent qualification requirements
- Crafts (1987) – no explanatory power over LTU, which is increasingly large % of L in 1930s
- Hatton (2004) – 95% of vacancies taken within a week, suggesting u/e not choosy/picky
Defences of Benjamin & Kochin (1)
- Crafts (1989) – specifies regression model with many variables to assess determinants of wage growth in GB, 1925-1939, and replacement ratio is one of only a few statistically significant determinants (along with trade union militancy)
Trade unions (9)
- Pre-WWI, TUs were concentrated in only a few industries (e.g. crafts and building)
- TUs spread in WWI in order to maintain production & minimise disputes – fairness & cost of living
- Also encouraged by government – Trade Boards created & industry collective bargaining was standard
- Union density up to 45% by 1920 (ct 22% in 1913) & 60% of employees covered by industry-wide agreements in the mid-1920s (but density down to 25% by 1929 (Hatton & Thomas, ’10)
- Hines (1964) – changes in TU unionisation account for over 90% of variation in wages (by regression) – significant for unemployment (especially in 1930s)
- Competing Claims Framework (Layard et al, 1991) (see diagram)
o “Feasible real wage” – level of real wage consistent with profit-maximising behaviour of firms (prices set via wage markup)
o “Target real wage” – employee supply of labour/collective bargaining
o Equilibrium is intersection, giving us the NAIRU (no unexpected inflation/deflation)
o TU bargaining power increased greatly into interwar, shifting target real wage right, increasing real wage and unemployment
o Crafts (1989) – shows with regression model on wage growth that trade union militancy was statistically significant determinant of wage growth, 1925-1939
o Feasible real wage rose too due to productivity improvements – compensated a bit - Broadberry (1986) & Pigou argue that the rapid spread of unionisation up to the 1920s led to greater nominal wage rigidity, explaining persistent u/e of 1920s
- Calmfors & Driffill (1988) – inverted U shape relationship between real wages and degree of centralisation of collective bargaining (supported theoretically & empirically)
o At firm level, wage bargainers account for high PED of firm in industry – moderated
o At national level, bargainers take into account economy wide effects – moderated
o At industry level, negotiators don’t take into account the externalities of their actions on the wider economy, but also have some market power
o UK NAIRU dropped to 3% after WWII when “post-war settlement” meant that collective bargaining was much more centralised at national level (72% ct 42%) – suggests that not trade unions themselves, but degree of centralisation
o Probably need more empirical data/research to determine true significance - Gazely & Newell (2007)
o Calmfors/Driffill model weak because doesn’t account for open economy relationships and doesn’t include parameter for relative bargaining power of trade unions & employers
o TUs have less bargaining power as unemployment increases
Structural unemployment (‘Old Optimists) (3)
- Unemployment appears to be high in areas where old staples industries concentrated
o June 1932 – u/e is 28% in Scotland, compared to 4% in London (new industries)
o 1923-38 – u/e 23% in mining & 24% in metal trades vs. 7% in paper & 9% in food, drink and tobacco (Hatton, 2004) - Old industries particularly during the Interwar period
o World trade grew at significantly reduced rate & collapsed in the 1930s
o Loss of overseas markets – competition, own-country production & tariffs
o Out-dated equipment and processes in the UK - Mechanism
o GB moving to new industries in this period- cars, chemicals etc
o Labour not fully mobile as jobs switch regions – creates frictional u/e
o Moreover, new industries more capital intensive (labour-saving technology)
Structural Unemployment - Matthews et al. (1982) (3)
- Create index of structural change
- Index was 1.2 for 1924-37, compared to 1.6 for 1913-24 & 1.5 for 1937-51
- Suggests pace faster post-WWII (when NAIRU around 3%)
Structural Unemployment - von Tunzelmann (1982) (2)
- New industries not less labour intensive - in fact, reverse is true
- Ratio of share of employment to share of capital was 2 in new vs. 1.5 in old
Structural Unemployment - Hatton (1986) (2)
- Differences in u/e by industry/region are due to different incidence of cyclical fluctuations on different industries – old staples much more ‘cyclically sensitive’
- However, regional u/e differences remain after controlling for mix of industries
Structural unemployment - not independent of economic activity (3)
- Recession increases degree of structural u/e
- Hatton (1986) – internal migration influenced by AD – higher AD supports more migration, which would have reduced structural u/e
- Broadberry (1986) – structural u/e occurred in all countries with overvalued ERs in 1920s, so ER policy was partly to blame